{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 5000, "global_step": 13066, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 7.653451706719731e-05, "grad_norm": 0.5543133022888164, "learning_rate": 0.0, "loss": 1.1603, "step": 1 }, { "epoch": 0.00015306903413439462, "grad_norm": 0.9006133420840196, "learning_rate": 2.5510204081632652e-08, "loss": 1.7088, "step": 2 }, { "epoch": 0.00022960355120159193, "grad_norm": 0.7792124611182072, "learning_rate": 5.1020408163265303e-08, "loss": 1.654, "step": 3 }, { "epoch": 0.00030613806826878924, "grad_norm": 0.754177329677479, "learning_rate": 7.653061224489796e-08, "loss": 1.601, "step": 4 }, { "epoch": 0.0003826725853359865, "grad_norm": 0.7585485917747323, "learning_rate": 1.0204081632653061e-07, "loss": 1.3269, "step": 5 }, { "epoch": 0.00045920710240318386, "grad_norm": 0.90773958620214, "learning_rate": 1.2755102040816328e-07, "loss": 1.7799, "step": 6 }, { "epoch": 0.0005357416194703812, "grad_norm": 0.5626296102043628, "learning_rate": 1.5306122448979592e-07, "loss": 1.1232, "step": 7 }, { "epoch": 0.0006122761365375785, "grad_norm": 0.6825058708482228, "learning_rate": 1.7857142857142858e-07, "loss": 1.4467, "step": 8 }, { "epoch": 0.0006888106536047758, "grad_norm": 0.7252676726288547, "learning_rate": 2.0408163265306121e-07, "loss": 1.4227, "step": 9 }, { "epoch": 0.000765345170671973, "grad_norm": 0.969665394166545, "learning_rate": 2.295918367346939e-07, "loss": 1.2758, "step": 10 }, { "epoch": 0.0008418796877391703, "grad_norm": 0.9573879146114386, "learning_rate": 2.5510204081632656e-07, "loss": 1.7517, "step": 11 }, { "epoch": 0.0009184142048063677, "grad_norm": 0.6683179202886369, "learning_rate": 2.806122448979592e-07, "loss": 0.9698, "step": 12 }, { "epoch": 0.0009949487218735649, "grad_norm": 0.6492349344303812, "learning_rate": 3.0612244897959183e-07, "loss": 1.4301, "step": 13 }, { "epoch": 0.0010714832389407624, "grad_norm": 0.6392256359335632, "learning_rate": 3.3163265306122455e-07, "loss": 1.1753, "step": 14 }, { "epoch": 0.0011480177560079597, "grad_norm": 0.5994404433550519, "learning_rate": 3.5714285714285716e-07, "loss": 1.2235, "step": 15 }, { "epoch": 0.001224552273075157, "grad_norm": 1.050213843158958, "learning_rate": 3.826530612244898e-07, "loss": 1.8951, "step": 16 }, { "epoch": 0.0013010867901423542, "grad_norm": 0.8507121594297411, "learning_rate": 4.0816326530612243e-07, "loss": 1.5465, "step": 17 }, { "epoch": 0.0013776213072095515, "grad_norm": 0.8789546607307192, "learning_rate": 4.3367346938775514e-07, "loss": 1.7006, "step": 18 }, { "epoch": 0.0014541558242767488, "grad_norm": 2.0085990100043367, "learning_rate": 4.591836734693878e-07, "loss": 1.7983, "step": 19 }, { "epoch": 0.001530690341343946, "grad_norm": 0.697373442499593, "learning_rate": 4.846938775510205e-07, "loss": 1.5122, "step": 20 }, { "epoch": 0.0016072248584111434, "grad_norm": 0.6869546816893197, "learning_rate": 5.102040816326531e-07, "loss": 1.482, "step": 21 }, { "epoch": 0.0016837593754783406, "grad_norm": 0.7283794274194103, "learning_rate": 5.357142857142857e-07, "loss": 1.6182, "step": 22 }, { "epoch": 0.0017602938925455381, "grad_norm": 0.7654154788998775, "learning_rate": 5.612244897959184e-07, "loss": 1.5774, "step": 23 }, { "epoch": 0.0018368284096127354, "grad_norm": 0.8474146830546749, "learning_rate": 5.867346938775511e-07, "loss": 1.5569, "step": 24 }, { "epoch": 0.0019133629266799327, "grad_norm": 1.058520098809233, "learning_rate": 6.122448979591837e-07, "loss": 1.8506, "step": 25 }, { "epoch": 0.0019898974437471298, "grad_norm": 0.6576902894041956, "learning_rate": 6.377551020408164e-07, "loss": 1.3666, "step": 26 }, { "epoch": 0.002066431960814327, "grad_norm": 0.9175193725419177, "learning_rate": 6.632653061224491e-07, "loss": 1.577, "step": 27 }, { "epoch": 0.0021429664778815248, "grad_norm": 0.632994554449033, "learning_rate": 6.887755102040817e-07, "loss": 1.3329, "step": 28 }, { "epoch": 0.002219500994948722, "grad_norm": 1.3208584970931616, "learning_rate": 7.142857142857143e-07, "loss": 1.8562, "step": 29 }, { "epoch": 0.0022960355120159193, "grad_norm": 0.7546528669346105, "learning_rate": 7.39795918367347e-07, "loss": 1.513, "step": 30 }, { "epoch": 0.0023725700290831166, "grad_norm": 0.7457884772807881, "learning_rate": 7.653061224489796e-07, "loss": 1.589, "step": 31 }, { "epoch": 0.002449104546150314, "grad_norm": 1.7478320732023302, "learning_rate": 7.908163265306124e-07, "loss": 1.5735, "step": 32 }, { "epoch": 0.002525639063217511, "grad_norm": 0.6863495869311668, "learning_rate": 8.163265306122449e-07, "loss": 1.6458, "step": 33 }, { "epoch": 0.0026021735802847085, "grad_norm": 0.7252086591464023, "learning_rate": 8.418367346938776e-07, "loss": 1.6137, "step": 34 }, { "epoch": 0.0026787080973519057, "grad_norm": 0.784664180164915, "learning_rate": 8.673469387755103e-07, "loss": 1.8443, "step": 35 }, { "epoch": 0.002755242614419103, "grad_norm": 0.634864859161251, "learning_rate": 8.928571428571429e-07, "loss": 1.357, "step": 36 }, { "epoch": 0.0028317771314863003, "grad_norm": 0.709802135597134, "learning_rate": 9.183673469387756e-07, "loss": 1.6163, "step": 37 }, { "epoch": 0.0029083116485534976, "grad_norm": 0.7999940071560423, "learning_rate": 9.438775510204082e-07, "loss": 1.6945, "step": 38 }, { "epoch": 0.002984846165620695, "grad_norm": 0.770149177246219, "learning_rate": 9.69387755102041e-07, "loss": 1.6161, "step": 39 }, { "epoch": 0.003061380682687892, "grad_norm": 0.8238851776326207, "learning_rate": 9.948979591836735e-07, "loss": 1.5933, "step": 40 }, { "epoch": 0.0031379151997550894, "grad_norm": 0.6721699377720447, "learning_rate": 1.0204081632653063e-06, "loss": 1.1067, "step": 41 }, { "epoch": 0.0032144497168222867, "grad_norm": 0.59412620957463, "learning_rate": 1.0459183673469388e-06, "loss": 1.2108, "step": 42 }, { "epoch": 0.003290984233889484, "grad_norm": 0.8918442475768654, "learning_rate": 1.0714285714285714e-06, "loss": 1.5236, "step": 43 }, { "epoch": 0.0033675187509566813, "grad_norm": 0.7490315204325602, "learning_rate": 1.0969387755102041e-06, "loss": 1.4144, "step": 44 }, { "epoch": 0.0034440532680238786, "grad_norm": 0.822438280349189, "learning_rate": 1.122448979591837e-06, "loss": 1.4294, "step": 45 }, { "epoch": 0.0035205877850910763, "grad_norm": 0.7100019114771426, "learning_rate": 1.1479591836734695e-06, "loss": 1.5025, "step": 46 }, { "epoch": 0.0035971223021582736, "grad_norm": 0.8687018245853326, "learning_rate": 1.1734693877551022e-06, "loss": 1.946, "step": 47 }, { "epoch": 0.003673656819225471, "grad_norm": 0.6057337487285013, "learning_rate": 1.1989795918367348e-06, "loss": 1.1751, "step": 48 }, { "epoch": 0.003750191336292668, "grad_norm": 0.7666235314036431, "learning_rate": 1.2244897959183673e-06, "loss": 1.9054, "step": 49 }, { "epoch": 0.0038267258533598654, "grad_norm": 0.7309194279797597, "learning_rate": 1.25e-06, "loss": 1.4753, "step": 50 }, { "epoch": 0.0039032603704270627, "grad_norm": 0.6593894378571503, "learning_rate": 1.2755102040816329e-06, "loss": 1.5711, "step": 51 }, { "epoch": 0.0039797948874942595, "grad_norm": 0.828483701880011, "learning_rate": 1.3010204081632654e-06, "loss": 1.7903, "step": 52 }, { "epoch": 0.004056329404561457, "grad_norm": 0.7652782324064107, "learning_rate": 1.3265306122448982e-06, "loss": 1.6296, "step": 53 }, { "epoch": 0.004132863921628654, "grad_norm": 0.7748857806347332, "learning_rate": 1.3520408163265307e-06, "loss": 1.7072, "step": 54 }, { "epoch": 0.004209398438695852, "grad_norm": 0.8176103503409349, "learning_rate": 1.3775510204081633e-06, "loss": 1.5772, "step": 55 }, { "epoch": 0.0042859329557630495, "grad_norm": 0.7189712284151318, "learning_rate": 1.4030612244897959e-06, "loss": 1.5373, "step": 56 }, { "epoch": 0.004362467472830246, "grad_norm": 0.7204262236487476, "learning_rate": 1.4285714285714286e-06, "loss": 1.4572, "step": 57 }, { "epoch": 0.004439001989897444, "grad_norm": 0.6816565119763551, "learning_rate": 1.4540816326530614e-06, "loss": 1.5635, "step": 58 }, { "epoch": 0.004515536506964641, "grad_norm": 0.774621548843093, "learning_rate": 1.479591836734694e-06, "loss": 1.5998, "step": 59 }, { "epoch": 0.004592071024031839, "grad_norm": 0.7081985686963639, "learning_rate": 1.5051020408163267e-06, "loss": 1.2431, "step": 60 }, { "epoch": 0.0046686055410990355, "grad_norm": 0.6964701665269689, "learning_rate": 1.5306122448979593e-06, "loss": 1.5439, "step": 61 }, { "epoch": 0.004745140058166233, "grad_norm": 0.6439244432695842, "learning_rate": 1.556122448979592e-06, "loss": 1.3595, "step": 62 }, { "epoch": 0.00482167457523343, "grad_norm": 0.8023262290437888, "learning_rate": 1.5816326530612248e-06, "loss": 1.5674, "step": 63 }, { "epoch": 0.004898209092300628, "grad_norm": 0.8151120145100474, "learning_rate": 1.6071428571428574e-06, "loss": 1.4523, "step": 64 }, { "epoch": 0.004974743609367825, "grad_norm": 0.8113153248451547, "learning_rate": 1.6326530612244897e-06, "loss": 1.7658, "step": 65 }, { "epoch": 0.005051278126435022, "grad_norm": 0.7795457898462568, "learning_rate": 1.6581632653061225e-06, "loss": 1.2737, "step": 66 }, { "epoch": 0.005127812643502219, "grad_norm": 0.7682472839235571, "learning_rate": 1.6836734693877552e-06, "loss": 1.7055, "step": 67 }, { "epoch": 0.005204347160569417, "grad_norm": 0.7080798215994974, "learning_rate": 1.7091836734693878e-06, "loss": 1.3124, "step": 68 }, { "epoch": 0.005280881677636614, "grad_norm": 0.925741104845603, "learning_rate": 1.7346938775510206e-06, "loss": 1.7575, "step": 69 }, { "epoch": 0.0053574161947038115, "grad_norm": 0.5961027170192289, "learning_rate": 1.7602040816326531e-06, "loss": 1.1694, "step": 70 }, { "epoch": 0.005433950711771008, "grad_norm": 1.2307792147789915, "learning_rate": 1.7857142857142859e-06, "loss": 1.9707, "step": 71 }, { "epoch": 0.005510485228838206, "grad_norm": 0.6217523349772048, "learning_rate": 1.8112244897959187e-06, "loss": 1.4015, "step": 72 }, { "epoch": 0.005587019745905403, "grad_norm": 1.901937211945271, "learning_rate": 1.8367346938775512e-06, "loss": 1.8445, "step": 73 }, { "epoch": 0.005663554262972601, "grad_norm": 0.8120907428854304, "learning_rate": 1.862244897959184e-06, "loss": 1.7468, "step": 74 }, { "epoch": 0.005740088780039798, "grad_norm": 0.8136863109609926, "learning_rate": 1.8877551020408163e-06, "loss": 1.5678, "step": 75 }, { "epoch": 0.005816623297106995, "grad_norm": 1.0489580911183387, "learning_rate": 1.913265306122449e-06, "loss": 1.5078, "step": 76 }, { "epoch": 0.005893157814174193, "grad_norm": 0.8495224944859957, "learning_rate": 1.938775510204082e-06, "loss": 1.7306, "step": 77 }, { "epoch": 0.00596969233124139, "grad_norm": 0.8599391835186481, "learning_rate": 1.9642857142857144e-06, "loss": 1.7983, "step": 78 }, { "epoch": 0.0060462268483085875, "grad_norm": 0.6849876791074689, "learning_rate": 1.989795918367347e-06, "loss": 1.6884, "step": 79 }, { "epoch": 0.006122761365375784, "grad_norm": 0.5551447473058452, "learning_rate": 2.01530612244898e-06, "loss": 1.0439, "step": 80 }, { "epoch": 0.006199295882442982, "grad_norm": 0.7489728888785726, "learning_rate": 2.0408163265306125e-06, "loss": 1.4229, "step": 81 }, { "epoch": 0.006275830399510179, "grad_norm": 0.6798465389873332, "learning_rate": 2.066326530612245e-06, "loss": 1.232, "step": 82 }, { "epoch": 0.006352364916577377, "grad_norm": 0.9003384732168881, "learning_rate": 2.0918367346938776e-06, "loss": 1.6315, "step": 83 }, { "epoch": 0.0064288994336445734, "grad_norm": 0.8616680340778915, "learning_rate": 2.1173469387755106e-06, "loss": 1.6154, "step": 84 }, { "epoch": 0.006505433950711771, "grad_norm": 1.4489231466520376, "learning_rate": 2.1428571428571427e-06, "loss": 1.4384, "step": 85 }, { "epoch": 0.006581968467778968, "grad_norm": 1.0441692287322952, "learning_rate": 2.1683673469387757e-06, "loss": 1.68, "step": 86 }, { "epoch": 0.006658502984846166, "grad_norm": 1.0046888714530315, "learning_rate": 2.1938775510204083e-06, "loss": 1.9058, "step": 87 }, { "epoch": 0.006735037501913363, "grad_norm": 0.8308657476597857, "learning_rate": 2.219387755102041e-06, "loss": 1.6426, "step": 88 }, { "epoch": 0.00681157201898056, "grad_norm": 0.8035110285302401, "learning_rate": 2.244897959183674e-06, "loss": 1.4602, "step": 89 }, { "epoch": 0.006888106536047757, "grad_norm": 0.7086985190586965, "learning_rate": 2.2704081632653064e-06, "loss": 1.5129, "step": 90 }, { "epoch": 0.006964641053114955, "grad_norm": 1.291765311094669, "learning_rate": 2.295918367346939e-06, "loss": 2.0349, "step": 91 }, { "epoch": 0.007041175570182153, "grad_norm": 0.853496847825476, "learning_rate": 2.321428571428572e-06, "loss": 1.6662, "step": 92 }, { "epoch": 0.007117710087249349, "grad_norm": 1.48322178224944, "learning_rate": 2.3469387755102044e-06, "loss": 1.3706, "step": 93 }, { "epoch": 0.007194244604316547, "grad_norm": 0.7508436857498241, "learning_rate": 2.372448979591837e-06, "loss": 1.4618, "step": 94 }, { "epoch": 0.007270779121383744, "grad_norm": 0.6809714935514284, "learning_rate": 2.3979591836734696e-06, "loss": 1.4257, "step": 95 }, { "epoch": 0.007347313638450942, "grad_norm": 0.7675646644239258, "learning_rate": 2.423469387755102e-06, "loss": 1.5536, "step": 96 }, { "epoch": 0.0074238481555181386, "grad_norm": 0.6350673575900304, "learning_rate": 2.4489795918367347e-06, "loss": 1.3779, "step": 97 }, { "epoch": 0.007500382672585336, "grad_norm": 0.7665438513709646, "learning_rate": 2.4744897959183676e-06, "loss": 1.781, "step": 98 }, { "epoch": 0.007576917189652533, "grad_norm": 0.7159037840830718, "learning_rate": 2.5e-06, "loss": 1.6121, "step": 99 }, { "epoch": 0.007653451706719731, "grad_norm": 0.6268142920220352, "learning_rate": 2.5255102040816328e-06, "loss": 1.2404, "step": 100 }, { "epoch": 0.007729986223786928, "grad_norm": 0.9658899949588496, "learning_rate": 2.5510204081632657e-06, "loss": 2.0373, "step": 101 }, { "epoch": 0.007806520740854125, "grad_norm": 0.6208872368591688, "learning_rate": 2.576530612244898e-06, "loss": 1.1427, "step": 102 }, { "epoch": 0.007883055257921323, "grad_norm": 0.7460702823142458, "learning_rate": 2.602040816326531e-06, "loss": 1.5496, "step": 103 }, { "epoch": 0.007959589774988519, "grad_norm": 0.856042533861203, "learning_rate": 2.6275510204081634e-06, "loss": 1.48, "step": 104 }, { "epoch": 0.008036124292055717, "grad_norm": 0.7104918072204623, "learning_rate": 2.6530612244897964e-06, "loss": 1.4287, "step": 105 }, { "epoch": 0.008112658809122915, "grad_norm": 0.8812844028910253, "learning_rate": 2.6785714285714285e-06, "loss": 1.7963, "step": 106 }, { "epoch": 0.008189193326190112, "grad_norm": 0.7545690707665534, "learning_rate": 2.7040816326530615e-06, "loss": 1.6429, "step": 107 }, { "epoch": 0.008265727843257308, "grad_norm": 0.6073614256676381, "learning_rate": 2.729591836734694e-06, "loss": 1.1786, "step": 108 }, { "epoch": 0.008342262360324506, "grad_norm": 1.2073517142912267, "learning_rate": 2.7551020408163266e-06, "loss": 1.5181, "step": 109 }, { "epoch": 0.008418796877391704, "grad_norm": 0.8854600882579503, "learning_rate": 2.7806122448979596e-06, "loss": 1.7806, "step": 110 }, { "epoch": 0.008495331394458901, "grad_norm": 0.6413548208305045, "learning_rate": 2.8061224489795917e-06, "loss": 1.3915, "step": 111 }, { "epoch": 0.008571865911526099, "grad_norm": 0.8183809682031535, "learning_rate": 2.8316326530612247e-06, "loss": 1.6441, "step": 112 }, { "epoch": 0.008648400428593295, "grad_norm": 0.5948648351110447, "learning_rate": 2.8571428571428573e-06, "loss": 1.0855, "step": 113 }, { "epoch": 0.008724934945660493, "grad_norm": 0.8539927277091405, "learning_rate": 2.8826530612244902e-06, "loss": 1.9269, "step": 114 }, { "epoch": 0.00880146946272769, "grad_norm": 0.7116711051126505, "learning_rate": 2.908163265306123e-06, "loss": 1.5236, "step": 115 }, { "epoch": 0.008878003979794888, "grad_norm": 0.9319788232019757, "learning_rate": 2.9336734693877553e-06, "loss": 1.684, "step": 116 }, { "epoch": 0.008954538496862084, "grad_norm": 0.631703928763773, "learning_rate": 2.959183673469388e-06, "loss": 1.3099, "step": 117 }, { "epoch": 0.009031073013929282, "grad_norm": 0.8505966305073485, "learning_rate": 2.984693877551021e-06, "loss": 1.4169, "step": 118 }, { "epoch": 0.00910760753099648, "grad_norm": 0.530667580492609, "learning_rate": 3.0102040816326534e-06, "loss": 0.9156, "step": 119 }, { "epoch": 0.009184142048063677, "grad_norm": 0.7762153495066573, "learning_rate": 3.0357142857142856e-06, "loss": 1.6133, "step": 120 }, { "epoch": 0.009260676565130873, "grad_norm": 0.7386971797144123, "learning_rate": 3.0612244897959185e-06, "loss": 1.4193, "step": 121 }, { "epoch": 0.009337211082198071, "grad_norm": 0.9731325946181612, "learning_rate": 3.086734693877551e-06, "loss": 1.7543, "step": 122 }, { "epoch": 0.009413745599265269, "grad_norm": 0.5578761736148721, "learning_rate": 3.112244897959184e-06, "loss": 1.0145, "step": 123 }, { "epoch": 0.009490280116332466, "grad_norm": 0.7241780691467339, "learning_rate": 3.1377551020408166e-06, "loss": 1.3835, "step": 124 }, { "epoch": 0.009566814633399662, "grad_norm": 0.6629927153802713, "learning_rate": 3.1632653061224496e-06, "loss": 1.4684, "step": 125 }, { "epoch": 0.00964334915046686, "grad_norm": 0.6667923748729764, "learning_rate": 3.1887755102040818e-06, "loss": 1.3967, "step": 126 }, { "epoch": 0.009719883667534058, "grad_norm": 0.6156385734499219, "learning_rate": 3.2142857142857147e-06, "loss": 1.2421, "step": 127 }, { "epoch": 0.009796418184601256, "grad_norm": 0.8139758336593877, "learning_rate": 3.2397959183673473e-06, "loss": 1.7393, "step": 128 }, { "epoch": 0.009872952701668453, "grad_norm": 0.6549760843148348, "learning_rate": 3.2653061224489794e-06, "loss": 1.4336, "step": 129 }, { "epoch": 0.00994948721873565, "grad_norm": 0.7928672684143342, "learning_rate": 3.2908163265306124e-06, "loss": 1.4502, "step": 130 }, { "epoch": 0.010026021735802847, "grad_norm": 1.0344867237165796, "learning_rate": 3.316326530612245e-06, "loss": 1.6478, "step": 131 }, { "epoch": 0.010102556252870045, "grad_norm": 0.5939956322312424, "learning_rate": 3.341836734693878e-06, "loss": 1.2907, "step": 132 }, { "epoch": 0.010179090769937242, "grad_norm": 0.7236321253027866, "learning_rate": 3.3673469387755105e-06, "loss": 1.3796, "step": 133 }, { "epoch": 0.010255625287004438, "grad_norm": 0.9305519138917686, "learning_rate": 3.3928571428571435e-06, "loss": 1.8589, "step": 134 }, { "epoch": 0.010332159804071636, "grad_norm": 0.6065623181452972, "learning_rate": 3.4183673469387756e-06, "loss": 1.4092, "step": 135 }, { "epoch": 0.010408694321138834, "grad_norm": 0.8963407597541586, "learning_rate": 3.4438775510204086e-06, "loss": 1.4055, "step": 136 }, { "epoch": 0.010485228838206032, "grad_norm": 0.7543011062127578, "learning_rate": 3.469387755102041e-06, "loss": 1.3896, "step": 137 }, { "epoch": 0.010561763355273228, "grad_norm": 0.6562545214204587, "learning_rate": 3.494897959183674e-06, "loss": 1.4153, "step": 138 }, { "epoch": 0.010638297872340425, "grad_norm": 0.7158735443532045, "learning_rate": 3.5204081632653062e-06, "loss": 1.5838, "step": 139 }, { "epoch": 0.010714832389407623, "grad_norm": 0.634593456765445, "learning_rate": 3.545918367346939e-06, "loss": 1.5092, "step": 140 }, { "epoch": 0.01079136690647482, "grad_norm": 0.6399977592699894, "learning_rate": 3.5714285714285718e-06, "loss": 1.3567, "step": 141 }, { "epoch": 0.010867901423542017, "grad_norm": 0.8841932924457651, "learning_rate": 3.5969387755102043e-06, "loss": 1.6661, "step": 142 }, { "epoch": 0.010944435940609214, "grad_norm": 0.892040761575269, "learning_rate": 3.6224489795918373e-06, "loss": 1.5834, "step": 143 }, { "epoch": 0.011020970457676412, "grad_norm": 0.716994294352364, "learning_rate": 3.6479591836734694e-06, "loss": 1.5915, "step": 144 }, { "epoch": 0.01109750497474361, "grad_norm": 0.7978669746478189, "learning_rate": 3.6734693877551024e-06, "loss": 1.4997, "step": 145 }, { "epoch": 0.011174039491810806, "grad_norm": 0.7436326124758637, "learning_rate": 3.698979591836735e-06, "loss": 1.4026, "step": 146 }, { "epoch": 0.011250574008878004, "grad_norm": 0.5981469595209354, "learning_rate": 3.724489795918368e-06, "loss": 1.2186, "step": 147 }, { "epoch": 0.011327108525945201, "grad_norm": 0.7010244673481033, "learning_rate": 3.7500000000000005e-06, "loss": 1.5132, "step": 148 }, { "epoch": 0.011403643043012399, "grad_norm": 0.5972263621409107, "learning_rate": 3.7755102040816327e-06, "loss": 1.2463, "step": 149 }, { "epoch": 0.011480177560079597, "grad_norm": 0.7343124050992713, "learning_rate": 3.8010204081632656e-06, "loss": 1.3161, "step": 150 }, { "epoch": 0.011556712077146793, "grad_norm": 0.8271095614449644, "learning_rate": 3.826530612244898e-06, "loss": 1.5803, "step": 151 }, { "epoch": 0.01163324659421399, "grad_norm": 0.774749082449683, "learning_rate": 3.852040816326531e-06, "loss": 1.8567, "step": 152 }, { "epoch": 0.011709781111281188, "grad_norm": 0.5945040364660721, "learning_rate": 3.877551020408164e-06, "loss": 1.2794, "step": 153 }, { "epoch": 0.011786315628348386, "grad_norm": 0.7263731752950312, "learning_rate": 3.903061224489797e-06, "loss": 1.7082, "step": 154 }, { "epoch": 0.011862850145415582, "grad_norm": 0.7210410756160391, "learning_rate": 3.928571428571429e-06, "loss": 1.7094, "step": 155 }, { "epoch": 0.01193938466248278, "grad_norm": 2.4792647702176147, "learning_rate": 3.954081632653062e-06, "loss": 1.7753, "step": 156 }, { "epoch": 0.012015919179549977, "grad_norm": 0.7465816599645385, "learning_rate": 3.979591836734694e-06, "loss": 1.5678, "step": 157 }, { "epoch": 0.012092453696617175, "grad_norm": 0.693374619944547, "learning_rate": 4.005102040816327e-06, "loss": 1.4726, "step": 158 }, { "epoch": 0.012168988213684371, "grad_norm": 0.6058613167960101, "learning_rate": 4.03061224489796e-06, "loss": 1.1603, "step": 159 }, { "epoch": 0.012245522730751569, "grad_norm": 0.7530013238271518, "learning_rate": 4.056122448979592e-06, "loss": 1.5631, "step": 160 }, { "epoch": 0.012322057247818766, "grad_norm": 0.855687098470826, "learning_rate": 4.081632653061225e-06, "loss": 1.7446, "step": 161 }, { "epoch": 0.012398591764885964, "grad_norm": 1.0310735707203063, "learning_rate": 4.107142857142857e-06, "loss": 1.6486, "step": 162 }, { "epoch": 0.01247512628195316, "grad_norm": 0.6334551735020523, "learning_rate": 4.13265306122449e-06, "loss": 1.3502, "step": 163 }, { "epoch": 0.012551660799020358, "grad_norm": 0.8634376741123527, "learning_rate": 4.158163265306123e-06, "loss": 1.3277, "step": 164 }, { "epoch": 0.012628195316087555, "grad_norm": 0.7034893542850986, "learning_rate": 4.183673469387755e-06, "loss": 1.3395, "step": 165 }, { "epoch": 0.012704729833154753, "grad_norm": 0.7540285762400373, "learning_rate": 4.209183673469388e-06, "loss": 1.5974, "step": 166 }, { "epoch": 0.012781264350221951, "grad_norm": 1.2306922568379532, "learning_rate": 4.234693877551021e-06, "loss": 1.6233, "step": 167 }, { "epoch": 0.012857798867289147, "grad_norm": 0.607616649865397, "learning_rate": 4.260204081632653e-06, "loss": 1.2299, "step": 168 }, { "epoch": 0.012934333384356345, "grad_norm": 0.6241979654000134, "learning_rate": 4.2857142857142855e-06, "loss": 1.3931, "step": 169 }, { "epoch": 0.013010867901423542, "grad_norm": 0.8521939514767665, "learning_rate": 4.3112244897959184e-06, "loss": 1.7166, "step": 170 }, { "epoch": 0.01308740241849074, "grad_norm": 0.6561132415780986, "learning_rate": 4.336734693877551e-06, "loss": 1.3275, "step": 171 }, { "epoch": 0.013163936935557936, "grad_norm": 0.9495705271973147, "learning_rate": 4.362244897959184e-06, "loss": 1.8855, "step": 172 }, { "epoch": 0.013240471452625134, "grad_norm": 0.7867516626357992, "learning_rate": 4.3877551020408165e-06, "loss": 1.7422, "step": 173 }, { "epoch": 0.013317005969692331, "grad_norm": 0.9069150942459083, "learning_rate": 4.4132653061224495e-06, "loss": 1.5921, "step": 174 }, { "epoch": 0.01339354048675953, "grad_norm": 0.7036297002580745, "learning_rate": 4.438775510204082e-06, "loss": 1.4655, "step": 175 }, { "epoch": 0.013470075003826725, "grad_norm": 0.6644926924180714, "learning_rate": 4.464285714285715e-06, "loss": 1.3462, "step": 176 }, { "epoch": 0.013546609520893923, "grad_norm": 0.6467099264424655, "learning_rate": 4.489795918367348e-06, "loss": 1.201, "step": 177 }, { "epoch": 0.01362314403796112, "grad_norm": 0.7566424142566419, "learning_rate": 4.51530612244898e-06, "loss": 1.4682, "step": 178 }, { "epoch": 0.013699678555028318, "grad_norm": 0.7702229431479125, "learning_rate": 4.540816326530613e-06, "loss": 1.4945, "step": 179 }, { "epoch": 0.013776213072095514, "grad_norm": 0.7706478307837394, "learning_rate": 4.566326530612245e-06, "loss": 1.2997, "step": 180 }, { "epoch": 0.013852747589162712, "grad_norm": 0.5663572263230623, "learning_rate": 4.591836734693878e-06, "loss": 1.2201, "step": 181 }, { "epoch": 0.01392928210622991, "grad_norm": 0.7945396188366861, "learning_rate": 4.617346938775511e-06, "loss": 1.6658, "step": 182 }, { "epoch": 0.014005816623297107, "grad_norm": 0.6901441108315396, "learning_rate": 4.642857142857144e-06, "loss": 1.3499, "step": 183 }, { "epoch": 0.014082351140364305, "grad_norm": 1.229468498932526, "learning_rate": 4.668367346938776e-06, "loss": 1.5371, "step": 184 }, { "epoch": 0.014158885657431501, "grad_norm": 0.7884594273768784, "learning_rate": 4.693877551020409e-06, "loss": 1.4948, "step": 185 }, { "epoch": 0.014235420174498699, "grad_norm": 1.1360544892609152, "learning_rate": 4.719387755102041e-06, "loss": 1.8002, "step": 186 }, { "epoch": 0.014311954691565897, "grad_norm": 0.7300075119690221, "learning_rate": 4.744897959183674e-06, "loss": 1.5238, "step": 187 }, { "epoch": 0.014388489208633094, "grad_norm": 0.6653276021590062, "learning_rate": 4.770408163265306e-06, "loss": 1.6973, "step": 188 }, { "epoch": 0.01446502372570029, "grad_norm": 0.6855680408856885, "learning_rate": 4.795918367346939e-06, "loss": 1.522, "step": 189 }, { "epoch": 0.014541558242767488, "grad_norm": 0.6534151337426618, "learning_rate": 4.821428571428572e-06, "loss": 1.5034, "step": 190 }, { "epoch": 0.014618092759834686, "grad_norm": 0.8018330751807697, "learning_rate": 4.846938775510204e-06, "loss": 1.4782, "step": 191 }, { "epoch": 0.014694627276901883, "grad_norm": 0.7413625434903166, "learning_rate": 4.872448979591837e-06, "loss": 1.334, "step": 192 }, { "epoch": 0.01477116179396908, "grad_norm": 0.9866053128806102, "learning_rate": 4.897959183673469e-06, "loss": 1.8488, "step": 193 }, { "epoch": 0.014847696311036277, "grad_norm": 0.8073983937174588, "learning_rate": 4.923469387755102e-06, "loss": 1.8562, "step": 194 }, { "epoch": 0.014924230828103475, "grad_norm": 0.6578792532885083, "learning_rate": 4.948979591836735e-06, "loss": 1.3481, "step": 195 }, { "epoch": 0.015000765345170673, "grad_norm": 0.6844105693976417, "learning_rate": 4.974489795918368e-06, "loss": 1.4438, "step": 196 }, { "epoch": 0.015077299862237869, "grad_norm": 1.1191018039707172, "learning_rate": 5e-06, "loss": 1.4358, "step": 197 }, { "epoch": 0.015153834379305066, "grad_norm": 0.6935539247928558, "learning_rate": 5.0255102040816325e-06, "loss": 1.6452, "step": 198 }, { "epoch": 0.015230368896372264, "grad_norm": 0.7415691414563097, "learning_rate": 5.0510204081632655e-06, "loss": 1.6005, "step": 199 }, { "epoch": 0.015306903413439462, "grad_norm": 0.6921571538065485, "learning_rate": 5.0765306122448985e-06, "loss": 1.7049, "step": 200 }, { "epoch": 0.015383437930506658, "grad_norm": 0.6169448035047199, "learning_rate": 5.1020408163265315e-06, "loss": 1.3457, "step": 201 }, { "epoch": 0.015459972447573855, "grad_norm": 0.8583613830817494, "learning_rate": 5.127551020408164e-06, "loss": 1.8552, "step": 202 }, { "epoch": 0.015536506964641053, "grad_norm": 0.6466235598298162, "learning_rate": 5.153061224489796e-06, "loss": 1.5724, "step": 203 }, { "epoch": 0.01561304148170825, "grad_norm": 0.7158918852686649, "learning_rate": 5.1785714285714296e-06, "loss": 1.5695, "step": 204 }, { "epoch": 0.01568957599877545, "grad_norm": 0.7018118982638661, "learning_rate": 5.204081632653062e-06, "loss": 1.5026, "step": 205 }, { "epoch": 0.015766110515842646, "grad_norm": 0.8331729097140679, "learning_rate": 5.229591836734695e-06, "loss": 1.7757, "step": 206 }, { "epoch": 0.015842645032909844, "grad_norm": 0.6036778407106412, "learning_rate": 5.255102040816327e-06, "loss": 1.37, "step": 207 }, { "epoch": 0.015919179549977038, "grad_norm": 1.1062424670583089, "learning_rate": 5.280612244897959e-06, "loss": 1.6523, "step": 208 }, { "epoch": 0.015995714067044236, "grad_norm": 0.7230842676774977, "learning_rate": 5.306122448979593e-06, "loss": 1.3412, "step": 209 }, { "epoch": 0.016072248584111434, "grad_norm": 2.032625538925077, "learning_rate": 5.331632653061225e-06, "loss": 2.0888, "step": 210 }, { "epoch": 0.01614878310117863, "grad_norm": 0.7697205902821688, "learning_rate": 5.357142857142857e-06, "loss": 1.7527, "step": 211 }, { "epoch": 0.01622531761824583, "grad_norm": 0.6572478756582484, "learning_rate": 5.38265306122449e-06, "loss": 1.4246, "step": 212 }, { "epoch": 0.016301852135313027, "grad_norm": 0.8604031674899681, "learning_rate": 5.408163265306123e-06, "loss": 1.8135, "step": 213 }, { "epoch": 0.016378386652380224, "grad_norm": 0.7990586242880797, "learning_rate": 5.433673469387756e-06, "loss": 1.464, "step": 214 }, { "epoch": 0.016454921169447422, "grad_norm": 0.7785748601478146, "learning_rate": 5.459183673469388e-06, "loss": 1.3905, "step": 215 }, { "epoch": 0.016531455686514616, "grad_norm": 0.6213699784611225, "learning_rate": 5.48469387755102e-06, "loss": 1.357, "step": 216 }, { "epoch": 0.016607990203581814, "grad_norm": 0.6423101213505271, "learning_rate": 5.510204081632653e-06, "loss": 1.2869, "step": 217 }, { "epoch": 0.016684524720649012, "grad_norm": 0.5659524810107474, "learning_rate": 5.535714285714286e-06, "loss": 1.2906, "step": 218 }, { "epoch": 0.01676105923771621, "grad_norm": 0.6532668723811067, "learning_rate": 5.561224489795919e-06, "loss": 1.3317, "step": 219 }, { "epoch": 0.016837593754783407, "grad_norm": 0.679120013333432, "learning_rate": 5.586734693877551e-06, "loss": 1.4965, "step": 220 }, { "epoch": 0.016914128271850605, "grad_norm": 1.032320138497014, "learning_rate": 5.6122448979591834e-06, "loss": 1.6449, "step": 221 }, { "epoch": 0.016990662788917803, "grad_norm": 0.6754860569342603, "learning_rate": 5.637755102040817e-06, "loss": 1.4925, "step": 222 }, { "epoch": 0.017067197305985, "grad_norm": 0.9270647836284723, "learning_rate": 5.663265306122449e-06, "loss": 1.7722, "step": 223 }, { "epoch": 0.017143731823052198, "grad_norm": 0.7332168978995982, "learning_rate": 5.688775510204082e-06, "loss": 1.48, "step": 224 }, { "epoch": 0.017220266340119392, "grad_norm": 0.6175465248634819, "learning_rate": 5.7142857142857145e-06, "loss": 1.4586, "step": 225 }, { "epoch": 0.01729680085718659, "grad_norm": 0.7409280242924055, "learning_rate": 5.739795918367348e-06, "loss": 1.5945, "step": 226 }, { "epoch": 0.017373335374253788, "grad_norm": 0.5603914011369461, "learning_rate": 5.7653061224489805e-06, "loss": 1.3271, "step": 227 }, { "epoch": 0.017449869891320986, "grad_norm": 0.6888866134251944, "learning_rate": 5.790816326530613e-06, "loss": 1.6521, "step": 228 }, { "epoch": 0.017526404408388183, "grad_norm": 0.7097333539232203, "learning_rate": 5.816326530612246e-06, "loss": 1.4923, "step": 229 }, { "epoch": 0.01760293892545538, "grad_norm": 0.9180655852647148, "learning_rate": 5.841836734693878e-06, "loss": 2.0186, "step": 230 }, { "epoch": 0.01767947344252258, "grad_norm": 0.7393964379572977, "learning_rate": 5.867346938775511e-06, "loss": 1.731, "step": 231 }, { "epoch": 0.017756007959589776, "grad_norm": 0.7334365047597092, "learning_rate": 5.892857142857144e-06, "loss": 1.6755, "step": 232 }, { "epoch": 0.01783254247665697, "grad_norm": 0.5783394779016475, "learning_rate": 5.918367346938776e-06, "loss": 1.2511, "step": 233 }, { "epoch": 0.01790907699372417, "grad_norm": 0.8778264606613141, "learning_rate": 5.943877551020408e-06, "loss": 1.5294, "step": 234 }, { "epoch": 0.017985611510791366, "grad_norm": 0.5638863878191499, "learning_rate": 5.969387755102042e-06, "loss": 1.1741, "step": 235 }, { "epoch": 0.018062146027858564, "grad_norm": 0.7647486979883888, "learning_rate": 5.994897959183674e-06, "loss": 1.7372, "step": 236 }, { "epoch": 0.01813868054492576, "grad_norm": 0.7545011524346827, "learning_rate": 6.020408163265307e-06, "loss": 1.5121, "step": 237 }, { "epoch": 0.01821521506199296, "grad_norm": 0.7352907747106404, "learning_rate": 6.045918367346939e-06, "loss": 1.5366, "step": 238 }, { "epoch": 0.018291749579060157, "grad_norm": 0.6474122391710009, "learning_rate": 6.071428571428571e-06, "loss": 1.4883, "step": 239 }, { "epoch": 0.018368284096127355, "grad_norm": 0.6318908895702792, "learning_rate": 6.096938775510205e-06, "loss": 1.429, "step": 240 }, { "epoch": 0.018444818613194552, "grad_norm": 0.7545838276689402, "learning_rate": 6.122448979591837e-06, "loss": 1.6407, "step": 241 }, { "epoch": 0.018521353130261747, "grad_norm": 0.7090239332459496, "learning_rate": 6.14795918367347e-06, "loss": 1.4056, "step": 242 }, { "epoch": 0.018597887647328944, "grad_norm": 0.7088458695818002, "learning_rate": 6.173469387755102e-06, "loss": 1.7596, "step": 243 }, { "epoch": 0.018674422164396142, "grad_norm": 0.9073014053331018, "learning_rate": 6.198979591836736e-06, "loss": 1.5414, "step": 244 }, { "epoch": 0.01875095668146334, "grad_norm": 0.7539704975857463, "learning_rate": 6.224489795918368e-06, "loss": 1.4112, "step": 245 }, { "epoch": 0.018827491198530538, "grad_norm": 0.8496381360885228, "learning_rate": 6.25e-06, "loss": 1.8999, "step": 246 }, { "epoch": 0.018904025715597735, "grad_norm": 0.6094018318090667, "learning_rate": 6.275510204081633e-06, "loss": 1.3477, "step": 247 }, { "epoch": 0.018980560232664933, "grad_norm": 0.719221242362398, "learning_rate": 6.301020408163265e-06, "loss": 1.3999, "step": 248 }, { "epoch": 0.01905709474973213, "grad_norm": 0.6951857735499408, "learning_rate": 6.326530612244899e-06, "loss": 1.3316, "step": 249 }, { "epoch": 0.019133629266799325, "grad_norm": 0.6988012467497217, "learning_rate": 6.352040816326531e-06, "loss": 1.3376, "step": 250 }, { "epoch": 0.019210163783866523, "grad_norm": 0.7458642087469981, "learning_rate": 6.3775510204081635e-06, "loss": 1.5154, "step": 251 }, { "epoch": 0.01928669830093372, "grad_norm": 0.6525088842422955, "learning_rate": 6.403061224489796e-06, "loss": 1.2367, "step": 252 }, { "epoch": 0.019363232818000918, "grad_norm": 0.7961322444015424, "learning_rate": 6.4285714285714295e-06, "loss": 1.3827, "step": 253 }, { "epoch": 0.019439767335068116, "grad_norm": 0.7504999962415796, "learning_rate": 6.454081632653062e-06, "loss": 1.0714, "step": 254 }, { "epoch": 0.019516301852135313, "grad_norm": 0.636780456418222, "learning_rate": 6.4795918367346946e-06, "loss": 1.5421, "step": 255 }, { "epoch": 0.01959283636920251, "grad_norm": 1.4191421773566295, "learning_rate": 6.505102040816327e-06, "loss": 1.8002, "step": 256 }, { "epoch": 0.01966937088626971, "grad_norm": 2.851602228782996, "learning_rate": 6.530612244897959e-06, "loss": 2.0519, "step": 257 }, { "epoch": 0.019745905403336907, "grad_norm": 0.6484245281739885, "learning_rate": 6.556122448979593e-06, "loss": 1.4218, "step": 258 }, { "epoch": 0.0198224399204041, "grad_norm": 0.6801840589806463, "learning_rate": 6.581632653061225e-06, "loss": 1.5507, "step": 259 }, { "epoch": 0.0198989744374713, "grad_norm": 1.021803253329601, "learning_rate": 6.607142857142858e-06, "loss": 1.2527, "step": 260 }, { "epoch": 0.019975508954538496, "grad_norm": 0.6712570297326487, "learning_rate": 6.63265306122449e-06, "loss": 1.2541, "step": 261 }, { "epoch": 0.020052043471605694, "grad_norm": 0.5942970060569573, "learning_rate": 6.658163265306124e-06, "loss": 1.2196, "step": 262 }, { "epoch": 0.020128577988672892, "grad_norm": 0.573976886312932, "learning_rate": 6.683673469387756e-06, "loss": 1.213, "step": 263 }, { "epoch": 0.02020511250574009, "grad_norm": 1.1596778912889156, "learning_rate": 6.709183673469388e-06, "loss": 1.8277, "step": 264 }, { "epoch": 0.020281647022807287, "grad_norm": 0.5493022497501052, "learning_rate": 6.734693877551021e-06, "loss": 1.2612, "step": 265 }, { "epoch": 0.020358181539874485, "grad_norm": 0.6921272718386753, "learning_rate": 6.760204081632653e-06, "loss": 1.3563, "step": 266 }, { "epoch": 0.02043471605694168, "grad_norm": 0.685006291144245, "learning_rate": 6.785714285714287e-06, "loss": 1.6003, "step": 267 }, { "epoch": 0.020511250574008877, "grad_norm": 0.6524174649604806, "learning_rate": 6.811224489795919e-06, "loss": 1.524, "step": 268 }, { "epoch": 0.020587785091076075, "grad_norm": 0.7781259935647241, "learning_rate": 6.836734693877551e-06, "loss": 1.5875, "step": 269 }, { "epoch": 0.020664319608143272, "grad_norm": 0.7379574093016172, "learning_rate": 6.862244897959184e-06, "loss": 1.7364, "step": 270 }, { "epoch": 0.02074085412521047, "grad_norm": 1.280619516824216, "learning_rate": 6.887755102040817e-06, "loss": 1.5662, "step": 271 }, { "epoch": 0.020817388642277668, "grad_norm": 0.6092478994147836, "learning_rate": 6.91326530612245e-06, "loss": 1.4251, "step": 272 }, { "epoch": 0.020893923159344865, "grad_norm": 1.5948600159573383, "learning_rate": 6.938775510204082e-06, "loss": 1.8629, "step": 273 }, { "epoch": 0.020970457676412063, "grad_norm": 0.7015594457720886, "learning_rate": 6.964285714285714e-06, "loss": 1.8898, "step": 274 }, { "epoch": 0.02104699219347926, "grad_norm": 0.8099837884091626, "learning_rate": 6.989795918367348e-06, "loss": 1.6101, "step": 275 }, { "epoch": 0.021123526710546455, "grad_norm": 0.5699642674283432, "learning_rate": 7.01530612244898e-06, "loss": 1.2124, "step": 276 }, { "epoch": 0.021200061227613653, "grad_norm": 0.7057195416130652, "learning_rate": 7.0408163265306125e-06, "loss": 1.3341, "step": 277 }, { "epoch": 0.02127659574468085, "grad_norm": 0.7375673612396559, "learning_rate": 7.0663265306122455e-06, "loss": 1.647, "step": 278 }, { "epoch": 0.021353130261748048, "grad_norm": 0.7733691975386522, "learning_rate": 7.091836734693878e-06, "loss": 1.8007, "step": 279 }, { "epoch": 0.021429664778815246, "grad_norm": 0.6182318466966219, "learning_rate": 7.1173469387755114e-06, "loss": 1.3898, "step": 280 }, { "epoch": 0.021506199295882444, "grad_norm": 0.6098976999829387, "learning_rate": 7.1428571428571436e-06, "loss": 1.3449, "step": 281 }, { "epoch": 0.02158273381294964, "grad_norm": 0.8005922722984998, "learning_rate": 7.168367346938776e-06, "loss": 1.6623, "step": 282 }, { "epoch": 0.02165926833001684, "grad_norm": 1.5149788612376172, "learning_rate": 7.193877551020409e-06, "loss": 1.4528, "step": 283 }, { "epoch": 0.021735802847084033, "grad_norm": 0.7343144733867973, "learning_rate": 7.219387755102042e-06, "loss": 1.7698, "step": 284 }, { "epoch": 0.02181233736415123, "grad_norm": 0.7369243265561233, "learning_rate": 7.244897959183675e-06, "loss": 1.5564, "step": 285 }, { "epoch": 0.02188887188121843, "grad_norm": 0.7049708024469841, "learning_rate": 7.270408163265307e-06, "loss": 1.8036, "step": 286 }, { "epoch": 0.021965406398285627, "grad_norm": 0.7729336279678173, "learning_rate": 7.295918367346939e-06, "loss": 1.6542, "step": 287 }, { "epoch": 0.022041940915352824, "grad_norm": 0.6798521506857779, "learning_rate": 7.321428571428572e-06, "loss": 1.6183, "step": 288 }, { "epoch": 0.022118475432420022, "grad_norm": 0.8702908344092192, "learning_rate": 7.346938775510205e-06, "loss": 1.7593, "step": 289 }, { "epoch": 0.02219500994948722, "grad_norm": 0.7390582383719602, "learning_rate": 7.372448979591838e-06, "loss": 1.5353, "step": 290 }, { "epoch": 0.022271544466554417, "grad_norm": 0.6023973518164804, "learning_rate": 7.39795918367347e-06, "loss": 1.3331, "step": 291 }, { "epoch": 0.02234807898362161, "grad_norm": 0.7140548313293337, "learning_rate": 7.423469387755102e-06, "loss": 1.8135, "step": 292 }, { "epoch": 0.02242461350068881, "grad_norm": 0.6439489573502003, "learning_rate": 7.448979591836736e-06, "loss": 1.4787, "step": 293 }, { "epoch": 0.022501148017756007, "grad_norm": 0.6492092310373763, "learning_rate": 7.474489795918368e-06, "loss": 1.3324, "step": 294 }, { "epoch": 0.022577682534823205, "grad_norm": 0.6578802313704962, "learning_rate": 7.500000000000001e-06, "loss": 1.3961, "step": 295 }, { "epoch": 0.022654217051890402, "grad_norm": 0.6354314658240812, "learning_rate": 7.525510204081633e-06, "loss": 1.6094, "step": 296 }, { "epoch": 0.0227307515689576, "grad_norm": 0.6130234942152479, "learning_rate": 7.551020408163265e-06, "loss": 1.2973, "step": 297 }, { "epoch": 0.022807286086024798, "grad_norm": 0.7244804972654644, "learning_rate": 7.576530612244899e-06, "loss": 1.5757, "step": 298 }, { "epoch": 0.022883820603091996, "grad_norm": 0.6246668979471546, "learning_rate": 7.602040816326531e-06, "loss": 1.3853, "step": 299 }, { "epoch": 0.022960355120159193, "grad_norm": 0.9093501010606855, "learning_rate": 7.627551020408163e-06, "loss": 1.9647, "step": 300 }, { "epoch": 0.023036889637226388, "grad_norm": 0.6766846676497035, "learning_rate": 7.653061224489796e-06, "loss": 1.4237, "step": 301 }, { "epoch": 0.023113424154293585, "grad_norm": 0.595828995454396, "learning_rate": 7.67857142857143e-06, "loss": 1.3746, "step": 302 }, { "epoch": 0.023189958671360783, "grad_norm": 0.6898370871029432, "learning_rate": 7.704081632653061e-06, "loss": 1.563, "step": 303 }, { "epoch": 0.02326649318842798, "grad_norm": 0.9516948923420088, "learning_rate": 7.729591836734695e-06, "loss": 1.8355, "step": 304 }, { "epoch": 0.02334302770549518, "grad_norm": 0.7365907952892128, "learning_rate": 7.755102040816327e-06, "loss": 1.6097, "step": 305 }, { "epoch": 0.023419562222562376, "grad_norm": 0.5436563866888214, "learning_rate": 7.78061224489796e-06, "loss": 1.2837, "step": 306 }, { "epoch": 0.023496096739629574, "grad_norm": 0.6849516172052329, "learning_rate": 7.806122448979593e-06, "loss": 1.5807, "step": 307 }, { "epoch": 0.02357263125669677, "grad_norm": 0.7285244276279486, "learning_rate": 7.831632653061226e-06, "loss": 1.5559, "step": 308 }, { "epoch": 0.023649165773763966, "grad_norm": 0.799160890223462, "learning_rate": 7.857142857142858e-06, "loss": 1.8442, "step": 309 }, { "epoch": 0.023725700290831164, "grad_norm": 0.7945360987359484, "learning_rate": 7.88265306122449e-06, "loss": 1.7251, "step": 310 }, { "epoch": 0.02380223480789836, "grad_norm": 0.9434950925989123, "learning_rate": 7.908163265306124e-06, "loss": 2.0396, "step": 311 }, { "epoch": 0.02387876932496556, "grad_norm": 0.6020256069104253, "learning_rate": 7.933673469387756e-06, "loss": 1.2859, "step": 312 }, { "epoch": 0.023955303842032757, "grad_norm": 0.659211824292014, "learning_rate": 7.959183673469388e-06, "loss": 1.5137, "step": 313 }, { "epoch": 0.024031838359099954, "grad_norm": 0.7152824606485948, "learning_rate": 7.98469387755102e-06, "loss": 1.5427, "step": 314 }, { "epoch": 0.024108372876167152, "grad_norm": 0.5870909844483021, "learning_rate": 8.010204081632654e-06, "loss": 1.2141, "step": 315 }, { "epoch": 0.02418490739323435, "grad_norm": 0.9038786660298859, "learning_rate": 8.035714285714286e-06, "loss": 2.0714, "step": 316 }, { "epoch": 0.024261441910301548, "grad_norm": 0.7580837279566359, "learning_rate": 8.06122448979592e-06, "loss": 1.9174, "step": 317 }, { "epoch": 0.024337976427368742, "grad_norm": 0.6357506023151837, "learning_rate": 8.086734693877552e-06, "loss": 1.5946, "step": 318 }, { "epoch": 0.02441451094443594, "grad_norm": 0.8132580507294848, "learning_rate": 8.112244897959184e-06, "loss": 1.7895, "step": 319 }, { "epoch": 0.024491045461503137, "grad_norm": 0.692114389458356, "learning_rate": 8.137755102040818e-06, "loss": 1.6021, "step": 320 }, { "epoch": 0.024567579978570335, "grad_norm": 0.5409788789747733, "learning_rate": 8.16326530612245e-06, "loss": 1.123, "step": 321 }, { "epoch": 0.024644114495637533, "grad_norm": 0.7180646987216044, "learning_rate": 8.188775510204082e-06, "loss": 1.4075, "step": 322 }, { "epoch": 0.02472064901270473, "grad_norm": 0.7722483086051932, "learning_rate": 8.214285714285714e-06, "loss": 1.2169, "step": 323 }, { "epoch": 0.024797183529771928, "grad_norm": 0.6707862944741039, "learning_rate": 8.239795918367348e-06, "loss": 1.5619, "step": 324 }, { "epoch": 0.024873718046839126, "grad_norm": 0.6330236100130755, "learning_rate": 8.26530612244898e-06, "loss": 1.3603, "step": 325 }, { "epoch": 0.02495025256390632, "grad_norm": 0.6155515795803709, "learning_rate": 8.290816326530612e-06, "loss": 1.5061, "step": 326 }, { "epoch": 0.025026787080973518, "grad_norm": 0.5814744820828933, "learning_rate": 8.316326530612246e-06, "loss": 1.4084, "step": 327 }, { "epoch": 0.025103321598040716, "grad_norm": 1.5261830158055725, "learning_rate": 8.341836734693878e-06, "loss": 1.6218, "step": 328 }, { "epoch": 0.025179856115107913, "grad_norm": 0.7153869853222908, "learning_rate": 8.36734693877551e-06, "loss": 1.7553, "step": 329 }, { "epoch": 0.02525639063217511, "grad_norm": 0.7433871197417138, "learning_rate": 8.392857142857144e-06, "loss": 1.6314, "step": 330 }, { "epoch": 0.02533292514924231, "grad_norm": 0.6088467777118977, "learning_rate": 8.418367346938776e-06, "loss": 1.5112, "step": 331 }, { "epoch": 0.025409459666309506, "grad_norm": 0.8945473449440358, "learning_rate": 8.443877551020409e-06, "loss": 2.0626, "step": 332 }, { "epoch": 0.025485994183376704, "grad_norm": 0.786866105899759, "learning_rate": 8.469387755102042e-06, "loss": 1.4192, "step": 333 }, { "epoch": 0.025562528700443902, "grad_norm": 0.5661422309955172, "learning_rate": 8.494897959183675e-06, "loss": 1.1995, "step": 334 }, { "epoch": 0.025639063217511096, "grad_norm": 1.0663951510516496, "learning_rate": 8.520408163265307e-06, "loss": 1.2061, "step": 335 }, { "epoch": 0.025715597734578294, "grad_norm": 0.6600175214128197, "learning_rate": 8.545918367346939e-06, "loss": 1.4981, "step": 336 }, { "epoch": 0.02579213225164549, "grad_norm": 0.7615992942159607, "learning_rate": 8.571428571428571e-06, "loss": 1.665, "step": 337 }, { "epoch": 0.02586866676871269, "grad_norm": 1.0574419238095056, "learning_rate": 8.596938775510205e-06, "loss": 1.8391, "step": 338 }, { "epoch": 0.025945201285779887, "grad_norm": 0.604794081524408, "learning_rate": 8.622448979591837e-06, "loss": 1.2307, "step": 339 }, { "epoch": 0.026021735802847085, "grad_norm": 2.3226939360870547, "learning_rate": 8.64795918367347e-06, "loss": 1.6168, "step": 340 }, { "epoch": 0.026098270319914282, "grad_norm": 0.634793102410073, "learning_rate": 8.673469387755103e-06, "loss": 1.3511, "step": 341 }, { "epoch": 0.02617480483698148, "grad_norm": 0.734807341701062, "learning_rate": 8.698979591836737e-06, "loss": 1.5243, "step": 342 }, { "epoch": 0.026251339354048674, "grad_norm": 0.5721178754540122, "learning_rate": 8.724489795918369e-06, "loss": 1.1777, "step": 343 }, { "epoch": 0.026327873871115872, "grad_norm": 1.297713144355545, "learning_rate": 8.750000000000001e-06, "loss": 1.7424, "step": 344 }, { "epoch": 0.02640440838818307, "grad_norm": 0.6696055991745414, "learning_rate": 8.775510204081633e-06, "loss": 1.4929, "step": 345 }, { "epoch": 0.026480942905250267, "grad_norm": 0.660927045070603, "learning_rate": 8.801020408163265e-06, "loss": 1.3742, "step": 346 }, { "epoch": 0.026557477422317465, "grad_norm": 0.6633728725113553, "learning_rate": 8.826530612244899e-06, "loss": 1.4308, "step": 347 }, { "epoch": 0.026634011939384663, "grad_norm": 0.5985872043671933, "learning_rate": 8.852040816326531e-06, "loss": 1.325, "step": 348 }, { "epoch": 0.02671054645645186, "grad_norm": 0.6571521346330775, "learning_rate": 8.877551020408163e-06, "loss": 1.3307, "step": 349 }, { "epoch": 0.02678708097351906, "grad_norm": 0.6816312627161274, "learning_rate": 8.903061224489795e-06, "loss": 1.606, "step": 350 }, { "epoch": 0.026863615490586256, "grad_norm": 0.7119906749814333, "learning_rate": 8.92857142857143e-06, "loss": 1.4483, "step": 351 }, { "epoch": 0.02694015000765345, "grad_norm": 0.8280274526942504, "learning_rate": 8.954081632653061e-06, "loss": 1.6625, "step": 352 }, { "epoch": 0.027016684524720648, "grad_norm": 0.8099422348731073, "learning_rate": 8.979591836734695e-06, "loss": 1.6384, "step": 353 }, { "epoch": 0.027093219041787846, "grad_norm": 0.684810478589788, "learning_rate": 9.005102040816327e-06, "loss": 1.6723, "step": 354 }, { "epoch": 0.027169753558855043, "grad_norm": 0.6114756121288059, "learning_rate": 9.03061224489796e-06, "loss": 1.4264, "step": 355 }, { "epoch": 0.02724628807592224, "grad_norm": 0.7662105195220993, "learning_rate": 9.056122448979593e-06, "loss": 1.6071, "step": 356 }, { "epoch": 0.02732282259298944, "grad_norm": 0.8449772505612343, "learning_rate": 9.081632653061225e-06, "loss": 1.7994, "step": 357 }, { "epoch": 0.027399357110056637, "grad_norm": 0.629820431400909, "learning_rate": 9.107142857142858e-06, "loss": 1.4605, "step": 358 }, { "epoch": 0.027475891627123834, "grad_norm": 0.6202520180969769, "learning_rate": 9.13265306122449e-06, "loss": 1.2835, "step": 359 }, { "epoch": 0.02755242614419103, "grad_norm": 0.8201697414968221, "learning_rate": 9.158163265306124e-06, "loss": 1.5492, "step": 360 }, { "epoch": 0.027628960661258226, "grad_norm": 0.7554924661681546, "learning_rate": 9.183673469387756e-06, "loss": 1.391, "step": 361 }, { "epoch": 0.027705495178325424, "grad_norm": 0.6319956259029368, "learning_rate": 9.209183673469388e-06, "loss": 1.5438, "step": 362 }, { "epoch": 0.02778202969539262, "grad_norm": 0.6636338153218464, "learning_rate": 9.234693877551022e-06, "loss": 1.6516, "step": 363 }, { "epoch": 0.02785856421245982, "grad_norm": 0.7159503159787959, "learning_rate": 9.260204081632654e-06, "loss": 1.3446, "step": 364 }, { "epoch": 0.027935098729527017, "grad_norm": 0.624506708488036, "learning_rate": 9.285714285714288e-06, "loss": 1.5919, "step": 365 }, { "epoch": 0.028011633246594215, "grad_norm": 0.6360443482856901, "learning_rate": 9.31122448979592e-06, "loss": 1.4463, "step": 366 }, { "epoch": 0.028088167763661413, "grad_norm": 0.6785138232469766, "learning_rate": 9.336734693877552e-06, "loss": 1.3119, "step": 367 }, { "epoch": 0.02816470228072861, "grad_norm": 0.5366079099965269, "learning_rate": 9.362244897959184e-06, "loss": 1.2635, "step": 368 }, { "epoch": 0.028241236797795805, "grad_norm": 0.5650566889826765, "learning_rate": 9.387755102040818e-06, "loss": 1.3616, "step": 369 }, { "epoch": 0.028317771314863002, "grad_norm": 0.7073500338388389, "learning_rate": 9.41326530612245e-06, "loss": 1.657, "step": 370 }, { "epoch": 0.0283943058319302, "grad_norm": 0.7998778451249302, "learning_rate": 9.438775510204082e-06, "loss": 1.4405, "step": 371 }, { "epoch": 0.028470840348997398, "grad_norm": 0.8118189308505762, "learning_rate": 9.464285714285714e-06, "loss": 1.5542, "step": 372 }, { "epoch": 0.028547374866064595, "grad_norm": 0.6120231415007104, "learning_rate": 9.489795918367348e-06, "loss": 1.3243, "step": 373 }, { "epoch": 0.028623909383131793, "grad_norm": 0.885362321432426, "learning_rate": 9.51530612244898e-06, "loss": 1.9519, "step": 374 }, { "epoch": 0.02870044390019899, "grad_norm": 0.6778245274188315, "learning_rate": 9.540816326530612e-06, "loss": 1.6265, "step": 375 }, { "epoch": 0.02877697841726619, "grad_norm": 0.8946040668484295, "learning_rate": 9.566326530612246e-06, "loss": 1.325, "step": 376 }, { "epoch": 0.028853512934333383, "grad_norm": 0.6444746241711294, "learning_rate": 9.591836734693878e-06, "loss": 1.5015, "step": 377 }, { "epoch": 0.02893004745140058, "grad_norm": 0.6528237793599856, "learning_rate": 9.617346938775512e-06, "loss": 1.4463, "step": 378 }, { "epoch": 0.029006581968467778, "grad_norm": 0.6838268772006857, "learning_rate": 9.642857142857144e-06, "loss": 1.4184, "step": 379 }, { "epoch": 0.029083116485534976, "grad_norm": 0.7811906249838196, "learning_rate": 9.668367346938776e-06, "loss": 1.5281, "step": 380 }, { "epoch": 0.029159651002602174, "grad_norm": 0.7973762498426971, "learning_rate": 9.693877551020408e-06, "loss": 1.5317, "step": 381 }, { "epoch": 0.02923618551966937, "grad_norm": 0.7077437920626721, "learning_rate": 9.719387755102042e-06, "loss": 1.7085, "step": 382 }, { "epoch": 0.02931272003673657, "grad_norm": 0.5714547762179556, "learning_rate": 9.744897959183674e-06, "loss": 1.1338, "step": 383 }, { "epoch": 0.029389254553803767, "grad_norm": 0.7108934953237472, "learning_rate": 9.770408163265307e-06, "loss": 1.7002, "step": 384 }, { "epoch": 0.029465789070870965, "grad_norm": 0.6030734734893004, "learning_rate": 9.795918367346939e-06, "loss": 1.2419, "step": 385 }, { "epoch": 0.02954232358793816, "grad_norm": 0.6740339954932842, "learning_rate": 9.821428571428573e-06, "loss": 1.7557, "step": 386 }, { "epoch": 0.029618858105005356, "grad_norm": 0.6030407838125619, "learning_rate": 9.846938775510205e-06, "loss": 1.3856, "step": 387 }, { "epoch": 0.029695392622072554, "grad_norm": 0.9195844204938564, "learning_rate": 9.872448979591838e-06, "loss": 1.8105, "step": 388 }, { "epoch": 0.029771927139139752, "grad_norm": 0.6605825314370996, "learning_rate": 9.89795918367347e-06, "loss": 1.5161, "step": 389 }, { "epoch": 0.02984846165620695, "grad_norm": 0.5886555193399452, "learning_rate": 9.923469387755103e-06, "loss": 1.4482, "step": 390 }, { "epoch": 0.029924996173274147, "grad_norm": 0.6071661908204804, "learning_rate": 9.948979591836737e-06, "loss": 1.4484, "step": 391 }, { "epoch": 0.030001530690341345, "grad_norm": 0.5571196861711257, "learning_rate": 9.974489795918369e-06, "loss": 1.0996, "step": 392 }, { "epoch": 0.030078065207408543, "grad_norm": 0.6454540922176127, "learning_rate": 1e-05, "loss": 1.6842, "step": 393 }, { "epoch": 0.030154599724475737, "grad_norm": 0.6769065492966317, "learning_rate": 9.999999877114021e-06, "loss": 1.7059, "step": 394 }, { "epoch": 0.030231134241542935, "grad_norm": 0.7192682560893499, "learning_rate": 9.999999508456094e-06, "loss": 1.7073, "step": 395 }, { "epoch": 0.030307668758610132, "grad_norm": 0.9890352579423197, "learning_rate": 9.999998894026237e-06, "loss": 1.3904, "step": 396 }, { "epoch": 0.03038420327567733, "grad_norm": 0.7467777383789381, "learning_rate": 9.99999803382449e-06, "loss": 1.688, "step": 397 }, { "epoch": 0.030460737792744528, "grad_norm": 0.7287015392406395, "learning_rate": 9.999996927850907e-06, "loss": 1.763, "step": 398 }, { "epoch": 0.030537272309811726, "grad_norm": 0.6871191126936096, "learning_rate": 9.999995576105555e-06, "loss": 1.6028, "step": 399 }, { "epoch": 0.030613806826878923, "grad_norm": 0.615148569616621, "learning_rate": 9.999993978588516e-06, "loss": 1.4732, "step": 400 }, { "epoch": 0.03069034134394612, "grad_norm": 0.6188637561851388, "learning_rate": 9.99999213529989e-06, "loss": 1.442, "step": 401 }, { "epoch": 0.030766875861013315, "grad_norm": 0.6420605455220184, "learning_rate": 9.99999004623979e-06, "loss": 1.4323, "step": 402 }, { "epoch": 0.030843410378080513, "grad_norm": 0.6321371606717198, "learning_rate": 9.999987711408343e-06, "loss": 1.5096, "step": 403 }, { "epoch": 0.03091994489514771, "grad_norm": 0.5807026435351609, "learning_rate": 9.999985130805695e-06, "loss": 1.4314, "step": 404 }, { "epoch": 0.03099647941221491, "grad_norm": 0.8741793660975439, "learning_rate": 9.999982304432001e-06, "loss": 1.8337, "step": 405 }, { "epoch": 0.031073013929282106, "grad_norm": 1.831576934995252, "learning_rate": 9.999979232287437e-06, "loss": 1.7742, "step": 406 }, { "epoch": 0.031149548446349304, "grad_norm": 0.6208876715043614, "learning_rate": 9.999975914372193e-06, "loss": 1.2702, "step": 407 }, { "epoch": 0.0312260829634165, "grad_norm": 0.8286457197690517, "learning_rate": 9.99997235068647e-06, "loss": 1.571, "step": 408 }, { "epoch": 0.0313026174804837, "grad_norm": 0.5581612588111717, "learning_rate": 9.999968541230487e-06, "loss": 1.1855, "step": 409 }, { "epoch": 0.0313791519975509, "grad_norm": 0.6976120018277467, "learning_rate": 9.99996448600448e-06, "loss": 1.2591, "step": 410 }, { "epoch": 0.031455686514618095, "grad_norm": 0.8531168226781983, "learning_rate": 9.999960185008697e-06, "loss": 1.5779, "step": 411 }, { "epoch": 0.03153222103168529, "grad_norm": 0.6718210496658464, "learning_rate": 9.999955638243404e-06, "loss": 1.4925, "step": 412 }, { "epoch": 0.03160875554875249, "grad_norm": 0.7930370027365535, "learning_rate": 9.999950845708876e-06, "loss": 1.7151, "step": 413 }, { "epoch": 0.03168529006581969, "grad_norm": 0.759512451255485, "learning_rate": 9.999945807405414e-06, "loss": 1.4385, "step": 414 }, { "epoch": 0.03176182458288688, "grad_norm": 0.7191485214661393, "learning_rate": 9.999940523333323e-06, "loss": 1.5124, "step": 415 }, { "epoch": 0.031838359099954076, "grad_norm": 0.6295491861182331, "learning_rate": 9.999934993492928e-06, "loss": 1.5466, "step": 416 }, { "epoch": 0.031914893617021274, "grad_norm": 0.576114948881194, "learning_rate": 9.99992921788457e-06, "loss": 1.1285, "step": 417 }, { "epoch": 0.03199142813408847, "grad_norm": 0.5976456130100406, "learning_rate": 9.999923196508603e-06, "loss": 1.4072, "step": 418 }, { "epoch": 0.03206796265115567, "grad_norm": 0.6216491147335452, "learning_rate": 9.999916929365396e-06, "loss": 1.5602, "step": 419 }, { "epoch": 0.03214449716822287, "grad_norm": 0.8254519231087039, "learning_rate": 9.999910416455337e-06, "loss": 1.6471, "step": 420 }, { "epoch": 0.032221031685290065, "grad_norm": 0.5650341834840662, "learning_rate": 9.999903657778825e-06, "loss": 1.1243, "step": 421 }, { "epoch": 0.03229756620235726, "grad_norm": 0.6442809367003439, "learning_rate": 9.999896653336275e-06, "loss": 1.5787, "step": 422 }, { "epoch": 0.03237410071942446, "grad_norm": 0.6923899498753237, "learning_rate": 9.999889403128115e-06, "loss": 1.4107, "step": 423 }, { "epoch": 0.03245063523649166, "grad_norm": 0.855932081540141, "learning_rate": 9.999881907154796e-06, "loss": 1.7252, "step": 424 }, { "epoch": 0.032527169753558856, "grad_norm": 0.5950515895590557, "learning_rate": 9.999874165416772e-06, "loss": 1.4805, "step": 425 }, { "epoch": 0.032603704270626054, "grad_norm": 0.6094878732724563, "learning_rate": 9.999866177914523e-06, "loss": 1.3748, "step": 426 }, { "epoch": 0.03268023878769325, "grad_norm": 0.5812056704657917, "learning_rate": 9.999857944648539e-06, "loss": 1.126, "step": 427 }, { "epoch": 0.03275677330476045, "grad_norm": 0.6711671566151249, "learning_rate": 9.999849465619326e-06, "loss": 1.5165, "step": 428 }, { "epoch": 0.03283330782182765, "grad_norm": 0.6512898324384714, "learning_rate": 9.999840740827404e-06, "loss": 1.384, "step": 429 }, { "epoch": 0.032909842338894844, "grad_norm": 0.6470826860576577, "learning_rate": 9.999831770273309e-06, "loss": 1.5903, "step": 430 }, { "epoch": 0.03298637685596204, "grad_norm": 0.5754290994319857, "learning_rate": 9.999822553957593e-06, "loss": 1.2156, "step": 431 }, { "epoch": 0.03306291137302923, "grad_norm": 0.5399035229267263, "learning_rate": 9.999813091880824e-06, "loss": 0.9859, "step": 432 }, { "epoch": 0.03313944589009643, "grad_norm": 0.9651313407584389, "learning_rate": 9.99980338404358e-06, "loss": 1.9829, "step": 433 }, { "epoch": 0.03321598040716363, "grad_norm": 0.8499739325098147, "learning_rate": 9.999793430446458e-06, "loss": 1.6234, "step": 434 }, { "epoch": 0.033292514924230826, "grad_norm": 0.6246400712377089, "learning_rate": 9.99978323109007e-06, "loss": 1.3155, "step": 435 }, { "epoch": 0.033369049441298024, "grad_norm": 0.7911101347335681, "learning_rate": 9.999772785975047e-06, "loss": 1.6416, "step": 436 }, { "epoch": 0.03344558395836522, "grad_norm": 0.7070426745770564, "learning_rate": 9.999762095102025e-06, "loss": 1.5228, "step": 437 }, { "epoch": 0.03352211847543242, "grad_norm": 0.7164930968519839, "learning_rate": 9.999751158471661e-06, "loss": 1.4374, "step": 438 }, { "epoch": 0.03359865299249962, "grad_norm": 0.6461892877691247, "learning_rate": 9.99973997608463e-06, "loss": 1.3329, "step": 439 }, { "epoch": 0.033675187509566815, "grad_norm": 0.7734889293170166, "learning_rate": 9.999728547941619e-06, "loss": 1.6268, "step": 440 }, { "epoch": 0.03375172202663401, "grad_norm": 0.6332780881600383, "learning_rate": 9.99971687404333e-06, "loss": 1.4125, "step": 441 }, { "epoch": 0.03382825654370121, "grad_norm": 0.7810239587859762, "learning_rate": 9.999704954390476e-06, "loss": 1.8539, "step": 442 }, { "epoch": 0.03390479106076841, "grad_norm": 0.6364742046240492, "learning_rate": 9.999692788983796e-06, "loss": 1.6013, "step": 443 }, { "epoch": 0.033981325577835605, "grad_norm": 0.5975682182306367, "learning_rate": 9.999680377824032e-06, "loss": 1.459, "step": 444 }, { "epoch": 0.0340578600949028, "grad_norm": 0.5517020597725469, "learning_rate": 9.99966772091195e-06, "loss": 1.274, "step": 445 }, { "epoch": 0.03413439461197, "grad_norm": 0.5886680324599908, "learning_rate": 9.999654818248326e-06, "loss": 1.4921, "step": 446 }, { "epoch": 0.0342109291290372, "grad_norm": 0.5536920840078005, "learning_rate": 9.999641669833953e-06, "loss": 1.0781, "step": 447 }, { "epoch": 0.034287463646104396, "grad_norm": 0.7032617035067749, "learning_rate": 9.99962827566964e-06, "loss": 1.5559, "step": 448 }, { "epoch": 0.03436399816317159, "grad_norm": 0.9068913997209191, "learning_rate": 9.999614635756208e-06, "loss": 1.3355, "step": 449 }, { "epoch": 0.034440532680238785, "grad_norm": 0.8247508952111727, "learning_rate": 9.999600750094499e-06, "loss": 1.7135, "step": 450 }, { "epoch": 0.03451706719730598, "grad_norm": 0.7731093496392444, "learning_rate": 9.99958661868536e-06, "loss": 1.5837, "step": 451 }, { "epoch": 0.03459360171437318, "grad_norm": 0.5781721087890882, "learning_rate": 9.999572241529664e-06, "loss": 1.3468, "step": 452 }, { "epoch": 0.03467013623144038, "grad_norm": 0.7139717035930251, "learning_rate": 9.999557618628294e-06, "loss": 1.6767, "step": 453 }, { "epoch": 0.034746670748507576, "grad_norm": 0.6355644071495807, "learning_rate": 9.999542749982146e-06, "loss": 1.6209, "step": 454 }, { "epoch": 0.03482320526557477, "grad_norm": 0.7894783612032478, "learning_rate": 9.999527635592135e-06, "loss": 1.5506, "step": 455 }, { "epoch": 0.03489973978264197, "grad_norm": 0.7235683763132742, "learning_rate": 9.999512275459191e-06, "loss": 1.7079, "step": 456 }, { "epoch": 0.03497627429970917, "grad_norm": 0.706467824088446, "learning_rate": 9.999496669584256e-06, "loss": 1.741, "step": 457 }, { "epoch": 0.03505280881677637, "grad_norm": 0.5156769521465482, "learning_rate": 9.99948081796829e-06, "loss": 0.9105, "step": 458 }, { "epoch": 0.035129343333843564, "grad_norm": 0.7396469358719138, "learning_rate": 9.999464720612265e-06, "loss": 1.6812, "step": 459 }, { "epoch": 0.03520587785091076, "grad_norm": 0.6804845292743167, "learning_rate": 9.999448377517171e-06, "loss": 1.472, "step": 460 }, { "epoch": 0.03528241236797796, "grad_norm": 0.5865032121490821, "learning_rate": 9.999431788684015e-06, "loss": 1.3026, "step": 461 }, { "epoch": 0.03535894688504516, "grad_norm": 0.6805829591694867, "learning_rate": 9.999414954113813e-06, "loss": 1.6972, "step": 462 }, { "epoch": 0.035435481402112355, "grad_norm": 0.5256944284322652, "learning_rate": 9.999397873807599e-06, "loss": 1.0988, "step": 463 }, { "epoch": 0.03551201591917955, "grad_norm": 0.49816165247754784, "learning_rate": 9.999380547766424e-06, "loss": 1.075, "step": 464 }, { "epoch": 0.03558855043624675, "grad_norm": 0.703659042862334, "learning_rate": 9.999362975991354e-06, "loss": 1.5669, "step": 465 }, { "epoch": 0.03566508495331394, "grad_norm": 0.6807927474495371, "learning_rate": 9.999345158483465e-06, "loss": 1.5188, "step": 466 }, { "epoch": 0.03574161947038114, "grad_norm": 0.603631238351739, "learning_rate": 9.999327095243856e-06, "loss": 1.4805, "step": 467 }, { "epoch": 0.03581815398744834, "grad_norm": 0.6169583716562697, "learning_rate": 9.999308786273632e-06, "loss": 1.579, "step": 468 }, { "epoch": 0.035894688504515534, "grad_norm": 0.6686987673392215, "learning_rate": 9.999290231573922e-06, "loss": 1.5623, "step": 469 }, { "epoch": 0.03597122302158273, "grad_norm": 0.7278218332126127, "learning_rate": 9.999271431145862e-06, "loss": 1.8601, "step": 470 }, { "epoch": 0.03604775753864993, "grad_norm": 0.5553673318012063, "learning_rate": 9.999252384990613e-06, "loss": 1.3074, "step": 471 }, { "epoch": 0.03612429205571713, "grad_norm": 0.5712510331694836, "learning_rate": 9.99923309310934e-06, "loss": 1.1387, "step": 472 }, { "epoch": 0.036200826572784325, "grad_norm": 0.5794547268732084, "learning_rate": 9.999213555503232e-06, "loss": 1.1799, "step": 473 }, { "epoch": 0.03627736108985152, "grad_norm": 0.7029436490223196, "learning_rate": 9.999193772173487e-06, "loss": 1.5847, "step": 474 }, { "epoch": 0.03635389560691872, "grad_norm": 0.6877339825232794, "learning_rate": 9.99917374312132e-06, "loss": 1.5231, "step": 475 }, { "epoch": 0.03643043012398592, "grad_norm": 0.5722108850168397, "learning_rate": 9.999153468347965e-06, "loss": 1.1667, "step": 476 }, { "epoch": 0.036506964641053116, "grad_norm": 0.7805161610600883, "learning_rate": 9.999132947854665e-06, "loss": 1.3715, "step": 477 }, { "epoch": 0.036583499158120314, "grad_norm": 0.6805824249413469, "learning_rate": 9.999112181642683e-06, "loss": 1.5456, "step": 478 }, { "epoch": 0.03666003367518751, "grad_norm": 1.2359228371002406, "learning_rate": 9.99909116971329e-06, "loss": 1.3083, "step": 479 }, { "epoch": 0.03673656819225471, "grad_norm": 0.5796662646579451, "learning_rate": 9.999069912067782e-06, "loss": 1.2714, "step": 480 }, { "epoch": 0.03681310270932191, "grad_norm": 0.5756179486257061, "learning_rate": 9.999048408707466e-06, "loss": 1.2928, "step": 481 }, { "epoch": 0.036889637226389105, "grad_norm": 0.8734910591804516, "learning_rate": 9.999026659633658e-06, "loss": 1.6598, "step": 482 }, { "epoch": 0.036966171743456296, "grad_norm": 0.6165252087030384, "learning_rate": 9.9990046648477e-06, "loss": 1.4877, "step": 483 }, { "epoch": 0.03704270626052349, "grad_norm": 0.5771023008759348, "learning_rate": 9.998982424350938e-06, "loss": 1.3866, "step": 484 }, { "epoch": 0.03711924077759069, "grad_norm": 0.6227424833368532, "learning_rate": 9.99895993814474e-06, "loss": 1.496, "step": 485 }, { "epoch": 0.03719577529465789, "grad_norm": 1.268581950380618, "learning_rate": 9.998937206230493e-06, "loss": 1.8228, "step": 486 }, { "epoch": 0.037272309811725086, "grad_norm": 0.6388252481270139, "learning_rate": 9.998914228609586e-06, "loss": 1.3721, "step": 487 }, { "epoch": 0.037348844328792284, "grad_norm": 0.6967930117158199, "learning_rate": 9.998891005283436e-06, "loss": 1.3782, "step": 488 }, { "epoch": 0.03742537884585948, "grad_norm": 0.601523726848999, "learning_rate": 9.998867536253469e-06, "loss": 1.2956, "step": 489 }, { "epoch": 0.03750191336292668, "grad_norm": 0.6323169216663258, "learning_rate": 9.998843821521126e-06, "loss": 1.4636, "step": 490 }, { "epoch": 0.03757844787999388, "grad_norm": 0.7363160541108016, "learning_rate": 9.998819861087863e-06, "loss": 1.5131, "step": 491 }, { "epoch": 0.037654982397061075, "grad_norm": 0.8672207419211637, "learning_rate": 9.998795654955153e-06, "loss": 1.7876, "step": 492 }, { "epoch": 0.03773151691412827, "grad_norm": 0.7589930128517743, "learning_rate": 9.998771203124485e-06, "loss": 1.55, "step": 493 }, { "epoch": 0.03780805143119547, "grad_norm": 0.6028912428058674, "learning_rate": 9.998746505597358e-06, "loss": 1.3065, "step": 494 }, { "epoch": 0.03788458594826267, "grad_norm": 0.6299764390516099, "learning_rate": 9.998721562375295e-06, "loss": 1.5302, "step": 495 }, { "epoch": 0.037961120465329866, "grad_norm": 0.7329253118542907, "learning_rate": 9.998696373459824e-06, "loss": 1.7986, "step": 496 }, { "epoch": 0.038037654982397064, "grad_norm": 0.6744288214650577, "learning_rate": 9.998670938852495e-06, "loss": 1.331, "step": 497 }, { "epoch": 0.03811418949946426, "grad_norm": 0.5484884514882429, "learning_rate": 9.998645258554868e-06, "loss": 1.1731, "step": 498 }, { "epoch": 0.03819072401653146, "grad_norm": 0.5014865359804401, "learning_rate": 9.998619332568525e-06, "loss": 1.0975, "step": 499 }, { "epoch": 0.03826725853359865, "grad_norm": 0.6661293776173325, "learning_rate": 9.998593160895054e-06, "loss": 1.4936, "step": 500 }, { "epoch": 0.03834379305066585, "grad_norm": 1.1109820225999751, "learning_rate": 9.998566743536068e-06, "loss": 1.6986, "step": 501 }, { "epoch": 0.038420327567733045, "grad_norm": 0.7046347794344198, "learning_rate": 9.998540080493186e-06, "loss": 1.5147, "step": 502 }, { "epoch": 0.03849686208480024, "grad_norm": 0.8854435607794352, "learning_rate": 9.998513171768048e-06, "loss": 1.9439, "step": 503 }, { "epoch": 0.03857339660186744, "grad_norm": 0.6504048322063676, "learning_rate": 9.99848601736231e-06, "loss": 1.4477, "step": 504 }, { "epoch": 0.03864993111893464, "grad_norm": 0.6409124986306645, "learning_rate": 9.998458617277635e-06, "loss": 1.6356, "step": 505 }, { "epoch": 0.038726465636001836, "grad_norm": 0.5710179709795011, "learning_rate": 9.99843097151571e-06, "loss": 1.414, "step": 506 }, { "epoch": 0.038803000153069034, "grad_norm": 0.5753591733729182, "learning_rate": 9.998403080078235e-06, "loss": 1.2394, "step": 507 }, { "epoch": 0.03887953467013623, "grad_norm": 0.6509198235261964, "learning_rate": 9.99837494296692e-06, "loss": 1.4383, "step": 508 }, { "epoch": 0.03895606918720343, "grad_norm": 0.6274182328823258, "learning_rate": 9.998346560183494e-06, "loss": 1.5938, "step": 509 }, { "epoch": 0.03903260370427063, "grad_norm": 0.7904884027268887, "learning_rate": 9.998317931729707e-06, "loss": 1.6514, "step": 510 }, { "epoch": 0.039109138221337825, "grad_norm": 0.5782547328484015, "learning_rate": 9.998289057607309e-06, "loss": 1.3789, "step": 511 }, { "epoch": 0.03918567273840502, "grad_norm": 0.7264639906276591, "learning_rate": 9.99825993781808e-06, "loss": 1.5689, "step": 512 }, { "epoch": 0.03926220725547222, "grad_norm": 0.7259049718507538, "learning_rate": 9.998230572363809e-06, "loss": 1.6924, "step": 513 }, { "epoch": 0.03933874177253942, "grad_norm": 0.9701047881041294, "learning_rate": 9.998200961246301e-06, "loss": 1.6508, "step": 514 }, { "epoch": 0.039415276289606616, "grad_norm": 0.615024152712932, "learning_rate": 9.998171104467371e-06, "loss": 1.4278, "step": 515 }, { "epoch": 0.03949181080667381, "grad_norm": 0.721812131389861, "learning_rate": 9.998141002028856e-06, "loss": 1.222, "step": 516 }, { "epoch": 0.039568345323741004, "grad_norm": 0.8305392068974389, "learning_rate": 9.998110653932607e-06, "loss": 1.4415, "step": 517 }, { "epoch": 0.0396448798408082, "grad_norm": 0.6177402558094828, "learning_rate": 9.998080060180487e-06, "loss": 1.4409, "step": 518 }, { "epoch": 0.0397214143578754, "grad_norm": 1.0248498142334026, "learning_rate": 9.998049220774375e-06, "loss": 1.5534, "step": 519 }, { "epoch": 0.0397979488749426, "grad_norm": 0.5834758245022349, "learning_rate": 9.998018135716168e-06, "loss": 1.2941, "step": 520 }, { "epoch": 0.039874483392009795, "grad_norm": 0.6258860512061952, "learning_rate": 9.997986805007775e-06, "loss": 1.3204, "step": 521 }, { "epoch": 0.03995101790907699, "grad_norm": 0.6847155077549484, "learning_rate": 9.99795522865112e-06, "loss": 1.3922, "step": 522 }, { "epoch": 0.04002755242614419, "grad_norm": 0.7358486125022399, "learning_rate": 9.997923406648145e-06, "loss": 1.6131, "step": 523 }, { "epoch": 0.04010408694321139, "grad_norm": 0.6457930790192739, "learning_rate": 9.997891339000805e-06, "loss": 1.5658, "step": 524 }, { "epoch": 0.040180621460278586, "grad_norm": 0.5565614167053554, "learning_rate": 9.997859025711068e-06, "loss": 1.218, "step": 525 }, { "epoch": 0.040257155977345783, "grad_norm": 0.682312606507225, "learning_rate": 9.997826466780921e-06, "loss": 1.2752, "step": 526 }, { "epoch": 0.04033369049441298, "grad_norm": 0.5798426805032405, "learning_rate": 9.997793662212368e-06, "loss": 1.1604, "step": 527 }, { "epoch": 0.04041022501148018, "grad_norm": 0.6372220283949261, "learning_rate": 9.997760612007418e-06, "loss": 1.6113, "step": 528 }, { "epoch": 0.04048675952854738, "grad_norm": 0.6431970462414955, "learning_rate": 9.997727316168105e-06, "loss": 1.533, "step": 529 }, { "epoch": 0.040563294045614574, "grad_norm": 0.9708311680739204, "learning_rate": 9.997693774696476e-06, "loss": 1.6276, "step": 530 }, { "epoch": 0.04063982856268177, "grad_norm": 0.912522739953705, "learning_rate": 9.99765998759459e-06, "loss": 1.6837, "step": 531 }, { "epoch": 0.04071636307974897, "grad_norm": 0.5294647713552644, "learning_rate": 9.997625954864525e-06, "loss": 1.1086, "step": 532 }, { "epoch": 0.04079289759681617, "grad_norm": 1.4500483470203294, "learning_rate": 9.99759167650837e-06, "loss": 1.4546, "step": 533 }, { "epoch": 0.04086943211388336, "grad_norm": 0.590082971090789, "learning_rate": 9.997557152528231e-06, "loss": 1.3597, "step": 534 }, { "epoch": 0.040945966630950556, "grad_norm": 0.7136769200555003, "learning_rate": 9.99752238292623e-06, "loss": 1.4781, "step": 535 }, { "epoch": 0.041022501148017754, "grad_norm": 0.5376368778412209, "learning_rate": 9.997487367704505e-06, "loss": 1.0779, "step": 536 }, { "epoch": 0.04109903566508495, "grad_norm": 0.634437410604844, "learning_rate": 9.997452106865204e-06, "loss": 1.4624, "step": 537 }, { "epoch": 0.04117557018215215, "grad_norm": 0.5748995089403264, "learning_rate": 9.997416600410498e-06, "loss": 1.2426, "step": 538 }, { "epoch": 0.04125210469921935, "grad_norm": 0.8988598770272104, "learning_rate": 9.997380848342566e-06, "loss": 1.5136, "step": 539 }, { "epoch": 0.041328639216286545, "grad_norm": 0.5477062925831563, "learning_rate": 9.997344850663602e-06, "loss": 1.0694, "step": 540 }, { "epoch": 0.04140517373335374, "grad_norm": 0.7617574712980616, "learning_rate": 9.997308607375822e-06, "loss": 1.7011, "step": 541 }, { "epoch": 0.04148170825042094, "grad_norm": 0.5494239355058425, "learning_rate": 9.99727211848145e-06, "loss": 1.0967, "step": 542 }, { "epoch": 0.04155824276748814, "grad_norm": 0.5815189119342715, "learning_rate": 9.997235383982733e-06, "loss": 1.0745, "step": 543 }, { "epoch": 0.041634777284555335, "grad_norm": 0.6619076333411321, "learning_rate": 9.997198403881922e-06, "loss": 1.4141, "step": 544 }, { "epoch": 0.04171131180162253, "grad_norm": 0.5597362795016078, "learning_rate": 9.997161178181291e-06, "loss": 1.2204, "step": 545 }, { "epoch": 0.04178784631868973, "grad_norm": 0.6113757158833196, "learning_rate": 9.997123706883129e-06, "loss": 1.4172, "step": 546 }, { "epoch": 0.04186438083575693, "grad_norm": 0.7312615053125672, "learning_rate": 9.997085989989739e-06, "loss": 1.582, "step": 547 }, { "epoch": 0.041940915352824126, "grad_norm": 0.6786073360504016, "learning_rate": 9.997048027503435e-06, "loss": 1.779, "step": 548 }, { "epoch": 0.042017449869891324, "grad_norm": 0.7810602460472668, "learning_rate": 9.99700981942655e-06, "loss": 1.5014, "step": 549 }, { "epoch": 0.04209398438695852, "grad_norm": 0.6531550641052708, "learning_rate": 9.996971365761436e-06, "loss": 1.153, "step": 550 }, { "epoch": 0.04217051890402571, "grad_norm": 0.6634525437135284, "learning_rate": 9.99693266651045e-06, "loss": 1.4015, "step": 551 }, { "epoch": 0.04224705342109291, "grad_norm": 0.6786151339580214, "learning_rate": 9.996893721675974e-06, "loss": 1.4996, "step": 552 }, { "epoch": 0.04232358793816011, "grad_norm": 0.5835540681767395, "learning_rate": 9.9968545312604e-06, "loss": 1.2197, "step": 553 }, { "epoch": 0.042400122455227306, "grad_norm": 0.7409848317897059, "learning_rate": 9.996815095266134e-06, "loss": 1.6848, "step": 554 }, { "epoch": 0.0424766569722945, "grad_norm": 0.7162856399925703, "learning_rate": 9.996775413695598e-06, "loss": 1.5444, "step": 555 }, { "epoch": 0.0425531914893617, "grad_norm": 0.5927107837955705, "learning_rate": 9.996735486551237e-06, "loss": 1.2774, "step": 556 }, { "epoch": 0.0426297260064289, "grad_norm": 0.6665413302218725, "learning_rate": 9.996695313835497e-06, "loss": 1.5962, "step": 557 }, { "epoch": 0.042706260523496097, "grad_norm": 0.8135660649197484, "learning_rate": 9.99665489555085e-06, "loss": 1.7722, "step": 558 }, { "epoch": 0.042782795040563294, "grad_norm": 0.6430114035818093, "learning_rate": 9.996614231699778e-06, "loss": 1.7135, "step": 559 }, { "epoch": 0.04285932955763049, "grad_norm": 0.660716230302945, "learning_rate": 9.99657332228478e-06, "loss": 1.5551, "step": 560 }, { "epoch": 0.04293586407469769, "grad_norm": 0.6065332740098596, "learning_rate": 9.996532167308371e-06, "loss": 1.4312, "step": 561 }, { "epoch": 0.04301239859176489, "grad_norm": 0.7465905112106715, "learning_rate": 9.996490766773076e-06, "loss": 1.4714, "step": 562 }, { "epoch": 0.043088933108832085, "grad_norm": 0.7163905435063859, "learning_rate": 9.996449120681443e-06, "loss": 1.457, "step": 563 }, { "epoch": 0.04316546762589928, "grad_norm": 0.7227087888875402, "learning_rate": 9.996407229036029e-06, "loss": 1.8126, "step": 564 }, { "epoch": 0.04324200214296648, "grad_norm": 0.5358024141575536, "learning_rate": 9.996365091839406e-06, "loss": 1.2798, "step": 565 }, { "epoch": 0.04331853666003368, "grad_norm": 0.6431068497559485, "learning_rate": 9.996322709094165e-06, "loss": 1.6115, "step": 566 }, { "epoch": 0.04339507117710087, "grad_norm": 0.5914452353387483, "learning_rate": 9.996280080802911e-06, "loss": 1.3864, "step": 567 }, { "epoch": 0.04347160569416807, "grad_norm": 0.815441067154719, "learning_rate": 9.996237206968263e-06, "loss": 1.429, "step": 568 }, { "epoch": 0.043548140211235264, "grad_norm": 1.2692494499982299, "learning_rate": 9.996194087592854e-06, "loss": 1.7064, "step": 569 }, { "epoch": 0.04362467472830246, "grad_norm": 0.6632083845041026, "learning_rate": 9.996150722679331e-06, "loss": 1.5963, "step": 570 }, { "epoch": 0.04370120924536966, "grad_norm": 0.5785268108182091, "learning_rate": 9.996107112230364e-06, "loss": 1.4436, "step": 571 }, { "epoch": 0.04377774376243686, "grad_norm": 0.7319977438469727, "learning_rate": 9.996063256248632e-06, "loss": 1.6701, "step": 572 }, { "epoch": 0.043854278279504055, "grad_norm": 0.7411393031693925, "learning_rate": 9.996019154736825e-06, "loss": 1.6618, "step": 573 }, { "epoch": 0.04393081279657125, "grad_norm": 0.6048413267710796, "learning_rate": 9.995974807697656e-06, "loss": 1.4051, "step": 574 }, { "epoch": 0.04400734731363845, "grad_norm": 0.5605848934398427, "learning_rate": 9.995930215133848e-06, "loss": 1.273, "step": 575 }, { "epoch": 0.04408388183070565, "grad_norm": 0.6504731519068633, "learning_rate": 9.995885377048144e-06, "loss": 1.6846, "step": 576 }, { "epoch": 0.044160416347772846, "grad_norm": 0.6079176011646894, "learning_rate": 9.995840293443295e-06, "loss": 1.372, "step": 577 }, { "epoch": 0.044236950864840044, "grad_norm": 0.7051103470929236, "learning_rate": 9.995794964322074e-06, "loss": 1.86, "step": 578 }, { "epoch": 0.04431348538190724, "grad_norm": 1.1971923809047544, "learning_rate": 9.995749389687266e-06, "loss": 1.6468, "step": 579 }, { "epoch": 0.04439001989897444, "grad_norm": 0.7131746445478965, "learning_rate": 9.995703569541669e-06, "loss": 1.8089, "step": 580 }, { "epoch": 0.04446655441604164, "grad_norm": 0.5644048191472967, "learning_rate": 9.995657503888102e-06, "loss": 1.2945, "step": 581 }, { "epoch": 0.044543088933108835, "grad_norm": 0.8022769748022014, "learning_rate": 9.99561119272939e-06, "loss": 1.6151, "step": 582 }, { "epoch": 0.04461962345017603, "grad_norm": 0.7258853657631779, "learning_rate": 9.995564636068385e-06, "loss": 1.6756, "step": 583 }, { "epoch": 0.04469615796724322, "grad_norm": 0.5610260757595193, "learning_rate": 9.995517833907942e-06, "loss": 1.2056, "step": 584 }, { "epoch": 0.04477269248431042, "grad_norm": 0.7108717306006663, "learning_rate": 9.995470786250941e-06, "loss": 1.6234, "step": 585 }, { "epoch": 0.04484922700137762, "grad_norm": 0.7394059439269186, "learning_rate": 9.995423493100268e-06, "loss": 1.75, "step": 586 }, { "epoch": 0.044925761518444816, "grad_norm": 1.3064338984668609, "learning_rate": 9.995375954458832e-06, "loss": 1.7611, "step": 587 }, { "epoch": 0.045002296035512014, "grad_norm": 1.2789400337863026, "learning_rate": 9.995328170329554e-06, "loss": 1.497, "step": 588 }, { "epoch": 0.04507883055257921, "grad_norm": 0.6212324069295975, "learning_rate": 9.99528014071537e-06, "loss": 1.4591, "step": 589 }, { "epoch": 0.04515536506964641, "grad_norm": 0.936916874891491, "learning_rate": 9.995231865619232e-06, "loss": 1.258, "step": 590 }, { "epoch": 0.04523189958671361, "grad_norm": 0.6460980277623618, "learning_rate": 9.995183345044103e-06, "loss": 1.4575, "step": 591 }, { "epoch": 0.045308434103780805, "grad_norm": 0.6423482209362804, "learning_rate": 9.995134578992966e-06, "loss": 1.432, "step": 592 }, { "epoch": 0.045384968620848, "grad_norm": 0.6231648933112746, "learning_rate": 9.995085567468818e-06, "loss": 1.4859, "step": 593 }, { "epoch": 0.0454615031379152, "grad_norm": 0.6957945452597163, "learning_rate": 9.995036310474669e-06, "loss": 1.3448, "step": 594 }, { "epoch": 0.0455380376549824, "grad_norm": 1.016056549244617, "learning_rate": 9.994986808013546e-06, "loss": 1.5762, "step": 595 }, { "epoch": 0.045614572172049596, "grad_norm": 0.6247378941354753, "learning_rate": 9.994937060088491e-06, "loss": 1.5365, "step": 596 }, { "epoch": 0.045691106689116794, "grad_norm": 0.6179467362275838, "learning_rate": 9.99488706670256e-06, "loss": 1.5172, "step": 597 }, { "epoch": 0.04576764120618399, "grad_norm": 1.3292206675207658, "learning_rate": 9.994836827858825e-06, "loss": 1.6804, "step": 598 }, { "epoch": 0.04584417572325119, "grad_norm": 0.6788464932461565, "learning_rate": 9.994786343560375e-06, "loss": 1.5674, "step": 599 }, { "epoch": 0.04592071024031839, "grad_norm": 0.6513150832379057, "learning_rate": 9.99473561381031e-06, "loss": 1.5702, "step": 600 }, { "epoch": 0.04599724475738558, "grad_norm": 0.6320741437565888, "learning_rate": 9.994684638611743e-06, "loss": 1.4583, "step": 601 }, { "epoch": 0.046073779274452775, "grad_norm": 0.7417671622693168, "learning_rate": 9.994633417967813e-06, "loss": 1.6651, "step": 602 }, { "epoch": 0.04615031379151997, "grad_norm": 0.7279452872828515, "learning_rate": 9.994581951881664e-06, "loss": 1.6262, "step": 603 }, { "epoch": 0.04622684830858717, "grad_norm": 0.7822542463452328, "learning_rate": 9.994530240356457e-06, "loss": 1.9151, "step": 604 }, { "epoch": 0.04630338282565437, "grad_norm": 0.8852430653423689, "learning_rate": 9.994478283395373e-06, "loss": 1.3811, "step": 605 }, { "epoch": 0.046379917342721566, "grad_norm": 0.7093546219956073, "learning_rate": 9.9944260810016e-06, "loss": 1.4812, "step": 606 }, { "epoch": 0.046456451859788764, "grad_norm": 0.7264987277137186, "learning_rate": 9.994373633178348e-06, "loss": 1.5348, "step": 607 }, { "epoch": 0.04653298637685596, "grad_norm": 0.5979496642933173, "learning_rate": 9.99432093992884e-06, "loss": 1.3033, "step": 608 }, { "epoch": 0.04660952089392316, "grad_norm": 0.5862573886527216, "learning_rate": 9.99426800125631e-06, "loss": 1.4804, "step": 609 }, { "epoch": 0.04668605541099036, "grad_norm": 1.155540258152031, "learning_rate": 9.994214817164016e-06, "loss": 2.0104, "step": 610 }, { "epoch": 0.046762589928057555, "grad_norm": 0.6582904334038634, "learning_rate": 9.994161387655223e-06, "loss": 1.5506, "step": 611 }, { "epoch": 0.04683912444512475, "grad_norm": 0.5992302090023207, "learning_rate": 9.994107712733213e-06, "loss": 1.5679, "step": 612 }, { "epoch": 0.04691565896219195, "grad_norm": 0.6722964710413789, "learning_rate": 9.994053792401285e-06, "loss": 1.5547, "step": 613 }, { "epoch": 0.04699219347925915, "grad_norm": 0.5125317722490081, "learning_rate": 9.993999626662754e-06, "loss": 1.0878, "step": 614 }, { "epoch": 0.047068727996326346, "grad_norm": 0.8060804324292828, "learning_rate": 9.993945215520944e-06, "loss": 1.4924, "step": 615 }, { "epoch": 0.04714526251339354, "grad_norm": 0.7683076062966825, "learning_rate": 9.993890558979202e-06, "loss": 1.7938, "step": 616 }, { "epoch": 0.04722179703046074, "grad_norm": 0.6303571172638638, "learning_rate": 9.993835657040884e-06, "loss": 1.4107, "step": 617 }, { "epoch": 0.04729833154752793, "grad_norm": 0.6463027712105108, "learning_rate": 9.993780509709363e-06, "loss": 1.5741, "step": 618 }, { "epoch": 0.04737486606459513, "grad_norm": 0.658528426506199, "learning_rate": 9.99372511698803e-06, "loss": 1.6022, "step": 619 }, { "epoch": 0.04745140058166233, "grad_norm": 0.5717791921743663, "learning_rate": 9.993669478880286e-06, "loss": 1.142, "step": 620 }, { "epoch": 0.047527935098729525, "grad_norm": 0.5690594039288301, "learning_rate": 9.99361359538955e-06, "loss": 1.3369, "step": 621 }, { "epoch": 0.04760446961579672, "grad_norm": 0.9818029388050408, "learning_rate": 9.993557466519257e-06, "loss": 1.716, "step": 622 }, { "epoch": 0.04768100413286392, "grad_norm": 0.6608930085803161, "learning_rate": 9.993501092272853e-06, "loss": 1.7088, "step": 623 }, { "epoch": 0.04775753864993112, "grad_norm": 0.7181636399673431, "learning_rate": 9.993444472653806e-06, "loss": 1.5845, "step": 624 }, { "epoch": 0.047834073166998316, "grad_norm": 0.5956152392292143, "learning_rate": 9.993387607665591e-06, "loss": 1.3561, "step": 625 }, { "epoch": 0.04791060768406551, "grad_norm": 0.7311540044365868, "learning_rate": 9.993330497311704e-06, "loss": 1.5284, "step": 626 }, { "epoch": 0.04798714220113271, "grad_norm": 0.8938144026815045, "learning_rate": 9.993273141595653e-06, "loss": 1.6648, "step": 627 }, { "epoch": 0.04806367671819991, "grad_norm": 0.8072178265162382, "learning_rate": 9.993215540520962e-06, "loss": 1.7306, "step": 628 }, { "epoch": 0.04814021123526711, "grad_norm": 0.6177453261133534, "learning_rate": 9.99315769409117e-06, "loss": 1.4696, "step": 629 }, { "epoch": 0.048216745752334304, "grad_norm": 0.541016633749938, "learning_rate": 9.993099602309836e-06, "loss": 1.0845, "step": 630 }, { "epoch": 0.0482932802694015, "grad_norm": 0.6160116452197681, "learning_rate": 9.993041265180521e-06, "loss": 1.5673, "step": 631 }, { "epoch": 0.0483698147864687, "grad_norm": 0.5257369193575284, "learning_rate": 9.992982682706817e-06, "loss": 1.1971, "step": 632 }, { "epoch": 0.0484463493035359, "grad_norm": 0.7134988108446907, "learning_rate": 9.992923854892319e-06, "loss": 1.5962, "step": 633 }, { "epoch": 0.048522883820603095, "grad_norm": 0.8285965998598201, "learning_rate": 9.992864781740643e-06, "loss": 1.8567, "step": 634 }, { "epoch": 0.048599418337670286, "grad_norm": 0.6680027316841289, "learning_rate": 9.992805463255418e-06, "loss": 1.4976, "step": 635 }, { "epoch": 0.048675952854737484, "grad_norm": 0.7250493400446115, "learning_rate": 9.992745899440288e-06, "loss": 1.4578, "step": 636 }, { "epoch": 0.04875248737180468, "grad_norm": 1.3406512514738318, "learning_rate": 9.992686090298915e-06, "loss": 1.259, "step": 637 }, { "epoch": 0.04882902188887188, "grad_norm": 0.7758300068438102, "learning_rate": 9.992626035834973e-06, "loss": 1.7191, "step": 638 }, { "epoch": 0.04890555640593908, "grad_norm": 0.5929428300552925, "learning_rate": 9.992565736052152e-06, "loss": 1.4568, "step": 639 }, { "epoch": 0.048982090923006275, "grad_norm": 0.8150766102528282, "learning_rate": 9.992505190954156e-06, "loss": 1.7671, "step": 640 }, { "epoch": 0.04905862544007347, "grad_norm": 0.7824915538864081, "learning_rate": 9.992444400544704e-06, "loss": 1.7927, "step": 641 }, { "epoch": 0.04913515995714067, "grad_norm": 0.5487183548119657, "learning_rate": 9.992383364827536e-06, "loss": 1.2607, "step": 642 }, { "epoch": 0.04921169447420787, "grad_norm": 0.80745979630256, "learning_rate": 9.992322083806398e-06, "loss": 1.5996, "step": 643 }, { "epoch": 0.049288228991275065, "grad_norm": 1.1539697083925027, "learning_rate": 9.992260557485056e-06, "loss": 1.5189, "step": 644 }, { "epoch": 0.04936476350834226, "grad_norm": 0.7394871391118539, "learning_rate": 9.99219878586729e-06, "loss": 1.4572, "step": 645 }, { "epoch": 0.04944129802540946, "grad_norm": 0.7489425291185243, "learning_rate": 9.9921367689569e-06, "loss": 1.8203, "step": 646 }, { "epoch": 0.04951783254247666, "grad_norm": 0.5911738686160434, "learning_rate": 9.992074506757689e-06, "loss": 1.3092, "step": 647 }, { "epoch": 0.049594367059543856, "grad_norm": 0.6326363039034044, "learning_rate": 9.992011999273486e-06, "loss": 1.4868, "step": 648 }, { "epoch": 0.049670901576611054, "grad_norm": 0.6728793151110449, "learning_rate": 9.991949246508133e-06, "loss": 1.8019, "step": 649 }, { "epoch": 0.04974743609367825, "grad_norm": 0.6671598779727315, "learning_rate": 9.991886248465487e-06, "loss": 1.6441, "step": 650 }, { "epoch": 0.04982397061074545, "grad_norm": 0.6452439041695024, "learning_rate": 9.991823005149412e-06, "loss": 1.3843, "step": 651 }, { "epoch": 0.04990050512781264, "grad_norm": 0.5188168145060931, "learning_rate": 9.991759516563801e-06, "loss": 1.0788, "step": 652 }, { "epoch": 0.04997703964487984, "grad_norm": 0.60451365879961, "learning_rate": 9.991695782712551e-06, "loss": 1.5327, "step": 653 }, { "epoch": 0.050053574161947036, "grad_norm": 0.6220626118276713, "learning_rate": 9.99163180359958e-06, "loss": 1.4205, "step": 654 }, { "epoch": 0.05013010867901423, "grad_norm": 0.6318750245862768, "learning_rate": 9.99156757922882e-06, "loss": 1.5769, "step": 655 }, { "epoch": 0.05020664319608143, "grad_norm": 0.6843569468764439, "learning_rate": 9.991503109604212e-06, "loss": 1.5457, "step": 656 }, { "epoch": 0.05028317771314863, "grad_norm": 0.6713656273817138, "learning_rate": 9.991438394729723e-06, "loss": 1.5542, "step": 657 }, { "epoch": 0.050359712230215826, "grad_norm": 1.7808843905042713, "learning_rate": 9.991373434609326e-06, "loss": 1.6383, "step": 658 }, { "epoch": 0.050436246747283024, "grad_norm": 0.6105268036372317, "learning_rate": 9.991308229247013e-06, "loss": 1.3508, "step": 659 }, { "epoch": 0.05051278126435022, "grad_norm": 0.6901825065873464, "learning_rate": 9.991242778646792e-06, "loss": 1.6003, "step": 660 }, { "epoch": 0.05058931578141742, "grad_norm": 0.7414179628130674, "learning_rate": 9.991177082812682e-06, "loss": 1.8539, "step": 661 }, { "epoch": 0.05066585029848462, "grad_norm": 0.5390769635540159, "learning_rate": 9.99111114174872e-06, "loss": 1.2026, "step": 662 }, { "epoch": 0.050742384815551815, "grad_norm": 0.5932816866790651, "learning_rate": 9.99104495545896e-06, "loss": 1.5666, "step": 663 }, { "epoch": 0.05081891933261901, "grad_norm": 0.9106053182931979, "learning_rate": 9.990978523947465e-06, "loss": 1.7971, "step": 664 }, { "epoch": 0.05089545384968621, "grad_norm": 0.630303910810329, "learning_rate": 9.99091184721832e-06, "loss": 1.3564, "step": 665 }, { "epoch": 0.05097198836675341, "grad_norm": 0.6578578700522288, "learning_rate": 9.990844925275621e-06, "loss": 1.509, "step": 666 }, { "epoch": 0.051048522883820606, "grad_norm": 1.2573144195486337, "learning_rate": 9.990777758123478e-06, "loss": 1.6485, "step": 667 }, { "epoch": 0.051125057400887804, "grad_norm": 2.0782385064132716, "learning_rate": 9.990710345766022e-06, "loss": 1.7332, "step": 668 }, { "epoch": 0.051201591917954994, "grad_norm": 0.7235703271717161, "learning_rate": 9.99064268820739e-06, "loss": 1.4692, "step": 669 }, { "epoch": 0.05127812643502219, "grad_norm": 0.6025920774091921, "learning_rate": 9.990574785451742e-06, "loss": 1.4887, "step": 670 }, { "epoch": 0.05135466095208939, "grad_norm": 0.5922796745157011, "learning_rate": 9.990506637503249e-06, "loss": 1.1102, "step": 671 }, { "epoch": 0.05143119546915659, "grad_norm": 0.6663435856731305, "learning_rate": 9.9904382443661e-06, "loss": 1.6096, "step": 672 }, { "epoch": 0.051507729986223785, "grad_norm": 0.93844639788989, "learning_rate": 9.990369606044497e-06, "loss": 1.6942, "step": 673 }, { "epoch": 0.05158426450329098, "grad_norm": 2.284982004217665, "learning_rate": 9.990300722542655e-06, "loss": 1.6393, "step": 674 }, { "epoch": 0.05166079902035818, "grad_norm": 0.7287792304268685, "learning_rate": 9.990231593864806e-06, "loss": 1.5998, "step": 675 }, { "epoch": 0.05173733353742538, "grad_norm": 0.5890637606089953, "learning_rate": 9.990162220015203e-06, "loss": 1.3772, "step": 676 }, { "epoch": 0.051813868054492576, "grad_norm": 0.6692977465575033, "learning_rate": 9.990092600998102e-06, "loss": 1.5833, "step": 677 }, { "epoch": 0.051890402571559774, "grad_norm": 0.6386591608742658, "learning_rate": 9.990022736817785e-06, "loss": 1.5146, "step": 678 }, { "epoch": 0.05196693708862697, "grad_norm": 0.6413483284435051, "learning_rate": 9.989952627478543e-06, "loss": 1.5152, "step": 679 }, { "epoch": 0.05204347160569417, "grad_norm": 0.6947658407748465, "learning_rate": 9.989882272984685e-06, "loss": 1.4766, "step": 680 }, { "epoch": 0.05212000612276137, "grad_norm": 0.5679213693035644, "learning_rate": 9.989811673340531e-06, "loss": 1.386, "step": 681 }, { "epoch": 0.052196540639828565, "grad_norm": 1.117929365071203, "learning_rate": 9.989740828550422e-06, "loss": 1.4894, "step": 682 }, { "epoch": 0.05227307515689576, "grad_norm": 0.7539062470379206, "learning_rate": 9.989669738618709e-06, "loss": 1.5773, "step": 683 }, { "epoch": 0.05234960967396296, "grad_norm": 0.6656717617027177, "learning_rate": 9.989598403549759e-06, "loss": 1.5291, "step": 684 }, { "epoch": 0.05242614419103016, "grad_norm": 0.5725079469090081, "learning_rate": 9.989526823347957e-06, "loss": 1.4439, "step": 685 }, { "epoch": 0.05250267870809735, "grad_norm": 0.5675837175563044, "learning_rate": 9.9894549980177e-06, "loss": 1.1956, "step": 686 }, { "epoch": 0.052579213225164546, "grad_norm": 0.5838177042641622, "learning_rate": 9.989382927563404e-06, "loss": 1.4931, "step": 687 }, { "epoch": 0.052655747742231744, "grad_norm": 0.5706573516978142, "learning_rate": 9.989310611989492e-06, "loss": 1.3705, "step": 688 }, { "epoch": 0.05273228225929894, "grad_norm": 0.7331727151806354, "learning_rate": 9.989238051300411e-06, "loss": 1.4529, "step": 689 }, { "epoch": 0.05280881677636614, "grad_norm": 1.1217969685212377, "learning_rate": 9.989165245500621e-06, "loss": 1.5481, "step": 690 }, { "epoch": 0.05288535129343334, "grad_norm": 1.3747347680788735, "learning_rate": 9.989092194594592e-06, "loss": 1.7014, "step": 691 }, { "epoch": 0.052961885810500535, "grad_norm": 0.6176627966206873, "learning_rate": 9.989018898586813e-06, "loss": 1.566, "step": 692 }, { "epoch": 0.05303842032756773, "grad_norm": 0.7882561607393551, "learning_rate": 9.988945357481787e-06, "loss": 1.8086, "step": 693 }, { "epoch": 0.05311495484463493, "grad_norm": 0.61165065842479, "learning_rate": 9.988871571284034e-06, "loss": 1.4807, "step": 694 }, { "epoch": 0.05319148936170213, "grad_norm": 0.939259309804922, "learning_rate": 9.988797539998088e-06, "loss": 1.4314, "step": 695 }, { "epoch": 0.053268023878769326, "grad_norm": 0.7175510878596943, "learning_rate": 9.988723263628496e-06, "loss": 1.6118, "step": 696 }, { "epoch": 0.053344558395836524, "grad_norm": 0.6942793866421285, "learning_rate": 9.988648742179824e-06, "loss": 1.4498, "step": 697 }, { "epoch": 0.05342109291290372, "grad_norm": 0.573397205816099, "learning_rate": 9.988573975656648e-06, "loss": 1.3877, "step": 698 }, { "epoch": 0.05349762742997092, "grad_norm": 0.6480060815984128, "learning_rate": 9.988498964063565e-06, "loss": 1.4514, "step": 699 }, { "epoch": 0.05357416194703812, "grad_norm": 0.583729213855497, "learning_rate": 9.988423707405179e-06, "loss": 1.2769, "step": 700 }, { "epoch": 0.053650696464105314, "grad_norm": 0.6442826085256947, "learning_rate": 9.98834820568612e-06, "loss": 1.5112, "step": 701 }, { "epoch": 0.05372723098117251, "grad_norm": 0.602192472720062, "learning_rate": 9.988272458911025e-06, "loss": 1.3619, "step": 702 }, { "epoch": 0.0538037654982397, "grad_norm": 0.6286193579083845, "learning_rate": 9.988196467084546e-06, "loss": 1.3629, "step": 703 }, { "epoch": 0.0538803000153069, "grad_norm": 0.6557511668003185, "learning_rate": 9.988120230211354e-06, "loss": 1.5184, "step": 704 }, { "epoch": 0.0539568345323741, "grad_norm": 0.6982814038911924, "learning_rate": 9.988043748296132e-06, "loss": 1.4728, "step": 705 }, { "epoch": 0.054033369049441296, "grad_norm": 0.7257507637496482, "learning_rate": 9.987967021343582e-06, "loss": 1.6175, "step": 706 }, { "epoch": 0.054109903566508494, "grad_norm": 0.5761798380988099, "learning_rate": 9.987890049358415e-06, "loss": 1.454, "step": 707 }, { "epoch": 0.05418643808357569, "grad_norm": 0.5718959258488314, "learning_rate": 9.987812832345363e-06, "loss": 1.2142, "step": 708 }, { "epoch": 0.05426297260064289, "grad_norm": 0.7030779468157732, "learning_rate": 9.987735370309169e-06, "loss": 1.8078, "step": 709 }, { "epoch": 0.05433950711771009, "grad_norm": 0.6524006476842624, "learning_rate": 9.987657663254592e-06, "loss": 1.7931, "step": 710 }, { "epoch": 0.054416041634777285, "grad_norm": 0.6471690693783413, "learning_rate": 9.987579711186408e-06, "loss": 1.5816, "step": 711 }, { "epoch": 0.05449257615184448, "grad_norm": 0.6729425446197053, "learning_rate": 9.987501514109405e-06, "loss": 1.4069, "step": 712 }, { "epoch": 0.05456911066891168, "grad_norm": 0.5959405902756209, "learning_rate": 9.98742307202839e-06, "loss": 1.407, "step": 713 }, { "epoch": 0.05464564518597888, "grad_norm": 0.6076385255506996, "learning_rate": 9.98734438494818e-06, "loss": 1.5134, "step": 714 }, { "epoch": 0.054722179703046075, "grad_norm": 0.7052704477427119, "learning_rate": 9.987265452873612e-06, "loss": 1.4071, "step": 715 }, { "epoch": 0.05479871422011327, "grad_norm": 0.665862386525014, "learning_rate": 9.987186275809535e-06, "loss": 1.4725, "step": 716 }, { "epoch": 0.05487524873718047, "grad_norm": 0.6729711199181311, "learning_rate": 9.987106853760812e-06, "loss": 1.6997, "step": 717 }, { "epoch": 0.05495178325424767, "grad_norm": 0.6123945194457796, "learning_rate": 9.987027186732328e-06, "loss": 1.1687, "step": 718 }, { "epoch": 0.055028317771314866, "grad_norm": 0.6847629587344455, "learning_rate": 9.986947274728971e-06, "loss": 1.6347, "step": 719 }, { "epoch": 0.05510485228838206, "grad_norm": 0.5629754220171046, "learning_rate": 9.986867117755656e-06, "loss": 1.1964, "step": 720 }, { "epoch": 0.055181386805449255, "grad_norm": 0.7634659137829468, "learning_rate": 9.986786715817309e-06, "loss": 1.3923, "step": 721 }, { "epoch": 0.05525792132251645, "grad_norm": 0.7154093997791272, "learning_rate": 9.986706068918865e-06, "loss": 1.6078, "step": 722 }, { "epoch": 0.05533445583958365, "grad_norm": 0.6146849938851849, "learning_rate": 9.986625177065284e-06, "loss": 1.3852, "step": 723 }, { "epoch": 0.05541099035665085, "grad_norm": 0.8724765588979096, "learning_rate": 9.986544040261535e-06, "loss": 1.6378, "step": 724 }, { "epoch": 0.055487524873718046, "grad_norm": 0.6421540064463557, "learning_rate": 9.986462658512601e-06, "loss": 1.4792, "step": 725 }, { "epoch": 0.05556405939078524, "grad_norm": 0.8520109501674896, "learning_rate": 9.986381031823485e-06, "loss": 1.9305, "step": 726 }, { "epoch": 0.05564059390785244, "grad_norm": 0.9276274337994743, "learning_rate": 9.986299160199201e-06, "loss": 1.5947, "step": 727 }, { "epoch": 0.05571712842491964, "grad_norm": 0.7498286093953891, "learning_rate": 9.98621704364478e-06, "loss": 1.8478, "step": 728 }, { "epoch": 0.05579366294198684, "grad_norm": 0.6846692357128826, "learning_rate": 9.986134682165267e-06, "loss": 1.4729, "step": 729 }, { "epoch": 0.055870197459054034, "grad_norm": 0.6493885296439994, "learning_rate": 9.986052075765723e-06, "loss": 1.4494, "step": 730 }, { "epoch": 0.05594673197612123, "grad_norm": 0.5983681136703454, "learning_rate": 9.985969224451225e-06, "loss": 1.421, "step": 731 }, { "epoch": 0.05602326649318843, "grad_norm": 0.6549029841029363, "learning_rate": 9.985886128226858e-06, "loss": 1.4246, "step": 732 }, { "epoch": 0.05609980101025563, "grad_norm": 0.6420470349305156, "learning_rate": 9.985802787097735e-06, "loss": 1.4902, "step": 733 }, { "epoch": 0.056176335527322825, "grad_norm": 0.6303340397183825, "learning_rate": 9.985719201068973e-06, "loss": 1.2599, "step": 734 }, { "epoch": 0.05625287004439002, "grad_norm": 0.5931509293657249, "learning_rate": 9.985635370145707e-06, "loss": 1.4244, "step": 735 }, { "epoch": 0.05632940456145722, "grad_norm": 0.651354943455417, "learning_rate": 9.985551294333091e-06, "loss": 1.2216, "step": 736 }, { "epoch": 0.05640593907852441, "grad_norm": 0.5443937503106236, "learning_rate": 9.985466973636287e-06, "loss": 1.1966, "step": 737 }, { "epoch": 0.05648247359559161, "grad_norm": 0.8728979982790196, "learning_rate": 9.985382408060479e-06, "loss": 1.3516, "step": 738 }, { "epoch": 0.05655900811265881, "grad_norm": 0.5603073764853448, "learning_rate": 9.985297597610863e-06, "loss": 1.2134, "step": 739 }, { "epoch": 0.056635542629726005, "grad_norm": 0.7312579360532852, "learning_rate": 9.985212542292646e-06, "loss": 1.4765, "step": 740 }, { "epoch": 0.0567120771467932, "grad_norm": 0.6406708723668353, "learning_rate": 9.985127242111059e-06, "loss": 1.6865, "step": 741 }, { "epoch": 0.0567886116638604, "grad_norm": 0.5535266077368708, "learning_rate": 9.98504169707134e-06, "loss": 1.0927, "step": 742 }, { "epoch": 0.0568651461809276, "grad_norm": 0.5795940334905942, "learning_rate": 9.984955907178746e-06, "loss": 1.491, "step": 743 }, { "epoch": 0.056941680697994795, "grad_norm": 0.6325103446816489, "learning_rate": 9.98486987243855e-06, "loss": 1.3666, "step": 744 }, { "epoch": 0.05701821521506199, "grad_norm": 0.740354989583893, "learning_rate": 9.984783592856034e-06, "loss": 1.5159, "step": 745 }, { "epoch": 0.05709474973212919, "grad_norm": 0.6880084774199505, "learning_rate": 9.984697068436504e-06, "loss": 1.6741, "step": 746 }, { "epoch": 0.05717128424919639, "grad_norm": 0.7955589356334773, "learning_rate": 9.984610299185273e-06, "loss": 1.7902, "step": 747 }, { "epoch": 0.057247818766263586, "grad_norm": 0.6491013551641477, "learning_rate": 9.984523285107675e-06, "loss": 1.3331, "step": 748 }, { "epoch": 0.057324353283330784, "grad_norm": 0.6187277644967982, "learning_rate": 9.984436026209052e-06, "loss": 1.2446, "step": 749 }, { "epoch": 0.05740088780039798, "grad_norm": 0.5787989075550694, "learning_rate": 9.98434852249477e-06, "loss": 1.2746, "step": 750 }, { "epoch": 0.05747742231746518, "grad_norm": 0.5473465408682792, "learning_rate": 9.984260773970205e-06, "loss": 1.226, "step": 751 }, { "epoch": 0.05755395683453238, "grad_norm": 0.5474872779703601, "learning_rate": 9.984172780640747e-06, "loss": 1.2641, "step": 752 }, { "epoch": 0.057630491351599575, "grad_norm": 0.9504837680051527, "learning_rate": 9.9840845425118e-06, "loss": 1.8975, "step": 753 }, { "epoch": 0.057707025868666766, "grad_norm": 0.6242965133625774, "learning_rate": 9.983996059588792e-06, "loss": 1.4156, "step": 754 }, { "epoch": 0.05778356038573396, "grad_norm": 0.7686869991474725, "learning_rate": 9.983907331877153e-06, "loss": 1.5563, "step": 755 }, { "epoch": 0.05786009490280116, "grad_norm": 0.5950211482859488, "learning_rate": 9.983818359382342e-06, "loss": 1.3714, "step": 756 }, { "epoch": 0.05793662941986836, "grad_norm": 0.6262338865147938, "learning_rate": 9.983729142109821e-06, "loss": 1.5312, "step": 757 }, { "epoch": 0.058013163936935556, "grad_norm": 0.9846856889642174, "learning_rate": 9.983639680065071e-06, "loss": 1.6277, "step": 758 }, { "epoch": 0.058089698454002754, "grad_norm": 0.6116012909529875, "learning_rate": 9.983549973253588e-06, "loss": 1.5046, "step": 759 }, { "epoch": 0.05816623297106995, "grad_norm": 0.6801971362801995, "learning_rate": 9.983460021680888e-06, "loss": 1.5699, "step": 760 }, { "epoch": 0.05824276748813715, "grad_norm": 0.6939458687643455, "learning_rate": 9.983369825352499e-06, "loss": 1.6351, "step": 761 }, { "epoch": 0.05831930200520435, "grad_norm": 0.5476417958387136, "learning_rate": 9.983279384273955e-06, "loss": 1.2167, "step": 762 }, { "epoch": 0.058395836522271545, "grad_norm": 0.7004252145975964, "learning_rate": 9.98318869845082e-06, "loss": 1.3777, "step": 763 }, { "epoch": 0.05847237103933874, "grad_norm": 0.662719233638281, "learning_rate": 9.983097767888665e-06, "loss": 1.6483, "step": 764 }, { "epoch": 0.05854890555640594, "grad_norm": 0.5790706713088, "learning_rate": 9.983006592593075e-06, "loss": 1.3406, "step": 765 }, { "epoch": 0.05862544007347314, "grad_norm": 0.7445444847442024, "learning_rate": 9.982915172569653e-06, "loss": 1.6438, "step": 766 }, { "epoch": 0.058701974590540336, "grad_norm": 0.7399956643252621, "learning_rate": 9.982823507824016e-06, "loss": 1.5711, "step": 767 }, { "epoch": 0.058778509107607534, "grad_norm": 0.6590014931243676, "learning_rate": 9.982731598361796e-06, "loss": 1.5741, "step": 768 }, { "epoch": 0.05885504362467473, "grad_norm": 0.7489817335944764, "learning_rate": 9.982639444188641e-06, "loss": 1.7556, "step": 769 }, { "epoch": 0.05893157814174193, "grad_norm": 0.5714181769256972, "learning_rate": 9.982547045310213e-06, "loss": 1.314, "step": 770 }, { "epoch": 0.05900811265880912, "grad_norm": 0.7735946066717714, "learning_rate": 9.98245440173219e-06, "loss": 1.4317, "step": 771 }, { "epoch": 0.05908464717587632, "grad_norm": 0.7066429294222598, "learning_rate": 9.98236151346026e-06, "loss": 1.6513, "step": 772 }, { "epoch": 0.059161181692943515, "grad_norm": 0.6010528685036244, "learning_rate": 9.982268380500136e-06, "loss": 1.5336, "step": 773 }, { "epoch": 0.05923771621001071, "grad_norm": 0.594622520928362, "learning_rate": 9.982175002857537e-06, "loss": 1.4244, "step": 774 }, { "epoch": 0.05931425072707791, "grad_norm": 0.5538800731533413, "learning_rate": 9.982081380538202e-06, "loss": 1.2385, "step": 775 }, { "epoch": 0.05939078524414511, "grad_norm": 0.6151376922502969, "learning_rate": 9.981987513547883e-06, "loss": 1.419, "step": 776 }, { "epoch": 0.059467319761212306, "grad_norm": 0.6631961862204204, "learning_rate": 9.981893401892347e-06, "loss": 1.4732, "step": 777 }, { "epoch": 0.059543854278279504, "grad_norm": 0.647066391759199, "learning_rate": 9.981799045577378e-06, "loss": 1.6952, "step": 778 }, { "epoch": 0.0596203887953467, "grad_norm": 0.7866276240593837, "learning_rate": 9.98170444460877e-06, "loss": 1.6665, "step": 779 }, { "epoch": 0.0596969233124139, "grad_norm": 0.5558463627768985, "learning_rate": 9.98160959899234e-06, "loss": 1.1811, "step": 780 }, { "epoch": 0.0597734578294811, "grad_norm": 0.6492093178932401, "learning_rate": 9.981514508733911e-06, "loss": 1.5141, "step": 781 }, { "epoch": 0.059849992346548295, "grad_norm": 0.6858467993487476, "learning_rate": 9.98141917383933e-06, "loss": 1.5003, "step": 782 }, { "epoch": 0.05992652686361549, "grad_norm": 0.5464095180572798, "learning_rate": 9.981323594314454e-06, "loss": 1.3427, "step": 783 }, { "epoch": 0.06000306138068269, "grad_norm": 0.6512022121676633, "learning_rate": 9.981227770165152e-06, "loss": 1.5238, "step": 784 }, { "epoch": 0.06007959589774989, "grad_norm": 0.6724150704434988, "learning_rate": 9.981131701397316e-06, "loss": 1.6128, "step": 785 }, { "epoch": 0.060156130414817086, "grad_norm": 0.6969860243147172, "learning_rate": 9.981035388016846e-06, "loss": 1.6609, "step": 786 }, { "epoch": 0.06023266493188428, "grad_norm": 0.6265074717822433, "learning_rate": 9.98093883002966e-06, "loss": 1.4542, "step": 787 }, { "epoch": 0.060309199448951474, "grad_norm": 0.7631534273456245, "learning_rate": 9.980842027441693e-06, "loss": 1.6409, "step": 788 }, { "epoch": 0.06038573396601867, "grad_norm": 0.6281697494421584, "learning_rate": 9.980744980258891e-06, "loss": 1.5653, "step": 789 }, { "epoch": 0.06046226848308587, "grad_norm": 0.6526474415829988, "learning_rate": 9.980647688487217e-06, "loss": 1.4934, "step": 790 }, { "epoch": 0.06053880300015307, "grad_norm": 0.5311021809365543, "learning_rate": 9.980550152132649e-06, "loss": 1.2762, "step": 791 }, { "epoch": 0.060615337517220265, "grad_norm": 0.651468933547558, "learning_rate": 9.980452371201178e-06, "loss": 1.839, "step": 792 }, { "epoch": 0.06069187203428746, "grad_norm": 0.5970644282155502, "learning_rate": 9.980354345698817e-06, "loss": 1.46, "step": 793 }, { "epoch": 0.06076840655135466, "grad_norm": 0.6217450924525925, "learning_rate": 9.980256075631585e-06, "loss": 1.4713, "step": 794 }, { "epoch": 0.06084494106842186, "grad_norm": 0.5373888512517504, "learning_rate": 9.980157561005522e-06, "loss": 1.1779, "step": 795 }, { "epoch": 0.060921475585489056, "grad_norm": 0.6876962729935553, "learning_rate": 9.98005880182668e-06, "loss": 1.7142, "step": 796 }, { "epoch": 0.060998010102556253, "grad_norm": 0.7857649833072082, "learning_rate": 9.979959798101127e-06, "loss": 1.7275, "step": 797 }, { "epoch": 0.06107454461962345, "grad_norm": 0.6119775506653257, "learning_rate": 9.979860549834946e-06, "loss": 1.4319, "step": 798 }, { "epoch": 0.06115107913669065, "grad_norm": 0.7467088635618927, "learning_rate": 9.979761057034234e-06, "loss": 1.5345, "step": 799 }, { "epoch": 0.06122761365375785, "grad_norm": 0.6283445603311303, "learning_rate": 9.979661319705107e-06, "loss": 1.4591, "step": 800 }, { "epoch": 0.061304148170825044, "grad_norm": 0.603650945555776, "learning_rate": 9.979561337853689e-06, "loss": 1.5254, "step": 801 }, { "epoch": 0.06138068268789224, "grad_norm": 0.5826902538099237, "learning_rate": 9.979461111486129e-06, "loss": 1.3647, "step": 802 }, { "epoch": 0.06145721720495944, "grad_norm": 0.8156803282106639, "learning_rate": 9.979360640608579e-06, "loss": 1.796, "step": 803 }, { "epoch": 0.06153375172202663, "grad_norm": 0.6193620667704268, "learning_rate": 9.979259925227218e-06, "loss": 1.7288, "step": 804 }, { "epoch": 0.06161028623909383, "grad_norm": 0.7325371108937179, "learning_rate": 9.97915896534823e-06, "loss": 1.7724, "step": 805 }, { "epoch": 0.061686820756161026, "grad_norm": 0.8152366323028506, "learning_rate": 9.979057760977822e-06, "loss": 1.6553, "step": 806 }, { "epoch": 0.061763355273228224, "grad_norm": 0.7728134617405223, "learning_rate": 9.978956312122207e-06, "loss": 1.513, "step": 807 }, { "epoch": 0.06183988979029542, "grad_norm": 0.6506053301279874, "learning_rate": 9.978854618787622e-06, "loss": 1.5812, "step": 808 }, { "epoch": 0.06191642430736262, "grad_norm": 0.5286615294368435, "learning_rate": 9.978752680980316e-06, "loss": 1.2086, "step": 809 }, { "epoch": 0.06199295882442982, "grad_norm": 0.5741672385659453, "learning_rate": 9.97865049870655e-06, "loss": 1.3253, "step": 810 }, { "epoch": 0.062069493341497015, "grad_norm": 0.7078619117500148, "learning_rate": 9.978548071972604e-06, "loss": 1.431, "step": 811 }, { "epoch": 0.06214602785856421, "grad_norm": 0.6402271738190715, "learning_rate": 9.978445400784771e-06, "loss": 1.7401, "step": 812 }, { "epoch": 0.06222256237563141, "grad_norm": 0.5484185688238166, "learning_rate": 9.978342485149358e-06, "loss": 1.1452, "step": 813 }, { "epoch": 0.06229909689269861, "grad_norm": 0.6741591237400995, "learning_rate": 9.97823932507269e-06, "loss": 1.4675, "step": 814 }, { "epoch": 0.062375631409765805, "grad_norm": 0.7621380501168079, "learning_rate": 9.978135920561107e-06, "loss": 1.7849, "step": 815 }, { "epoch": 0.062452165926833, "grad_norm": 0.78463560044298, "learning_rate": 9.978032271620958e-06, "loss": 1.8981, "step": 816 }, { "epoch": 0.0625287004439002, "grad_norm": 0.6592484517295781, "learning_rate": 9.977928378258617e-06, "loss": 1.115, "step": 817 }, { "epoch": 0.0626052349609674, "grad_norm": 0.5716724355536559, "learning_rate": 9.977824240480462e-06, "loss": 1.3234, "step": 818 }, { "epoch": 0.0626817694780346, "grad_norm": 0.5723829227135163, "learning_rate": 9.977719858292898e-06, "loss": 1.243, "step": 819 }, { "epoch": 0.0627583039951018, "grad_norm": 0.6618739920865349, "learning_rate": 9.97761523170233e-06, "loss": 1.712, "step": 820 }, { "epoch": 0.06283483851216899, "grad_norm": 0.5842440355846785, "learning_rate": 9.977510360715194e-06, "loss": 1.1136, "step": 821 }, { "epoch": 0.06291137302923619, "grad_norm": 0.5707199626744361, "learning_rate": 9.97740524533793e-06, "loss": 1.4531, "step": 822 }, { "epoch": 0.06298790754630339, "grad_norm": 0.9564306110704849, "learning_rate": 9.977299885577e-06, "loss": 1.4616, "step": 823 }, { "epoch": 0.06306444206337058, "grad_norm": 0.6270170428105961, "learning_rate": 9.977194281438874e-06, "loss": 1.48, "step": 824 }, { "epoch": 0.06314097658043778, "grad_norm": 1.186860216252004, "learning_rate": 9.97708843293004e-06, "loss": 1.6581, "step": 825 }, { "epoch": 0.06321751109750498, "grad_norm": 0.6082217033180257, "learning_rate": 9.976982340057005e-06, "loss": 1.3247, "step": 826 }, { "epoch": 0.06329404561457218, "grad_norm": 0.6322956760231198, "learning_rate": 9.976876002826287e-06, "loss": 1.5376, "step": 827 }, { "epoch": 0.06337058013163938, "grad_norm": 0.6819156820719872, "learning_rate": 9.976769421244417e-06, "loss": 1.5185, "step": 828 }, { "epoch": 0.06344711464870657, "grad_norm": 0.6171901707383338, "learning_rate": 9.976662595317947e-06, "loss": 1.5057, "step": 829 }, { "epoch": 0.06352364916577376, "grad_norm": 0.6538582821553963, "learning_rate": 9.976555525053437e-06, "loss": 1.7518, "step": 830 }, { "epoch": 0.06360018368284096, "grad_norm": 0.5772900978317405, "learning_rate": 9.97644821045747e-06, "loss": 1.3416, "step": 831 }, { "epoch": 0.06367671819990815, "grad_norm": 0.5822913230566856, "learning_rate": 9.976340651536638e-06, "loss": 1.4772, "step": 832 }, { "epoch": 0.06375325271697535, "grad_norm": 0.7901514557535747, "learning_rate": 9.976232848297547e-06, "loss": 1.8289, "step": 833 }, { "epoch": 0.06382978723404255, "grad_norm": 0.6597029276316632, "learning_rate": 9.976124800746823e-06, "loss": 1.1781, "step": 834 }, { "epoch": 0.06390632175110975, "grad_norm": 0.7071397043267944, "learning_rate": 9.976016508891107e-06, "loss": 1.7762, "step": 835 }, { "epoch": 0.06398285626817694, "grad_norm": 0.5937382014054406, "learning_rate": 9.97590797273705e-06, "loss": 1.3489, "step": 836 }, { "epoch": 0.06405939078524414, "grad_norm": 0.6527812901215887, "learning_rate": 9.975799192291322e-06, "loss": 1.4629, "step": 837 }, { "epoch": 0.06413592530231134, "grad_norm": 0.6425567389803262, "learning_rate": 9.975690167560604e-06, "loss": 1.4607, "step": 838 }, { "epoch": 0.06421245981937854, "grad_norm": 0.5706286326528841, "learning_rate": 9.975580898551599e-06, "loss": 1.4761, "step": 839 }, { "epoch": 0.06428899433644573, "grad_norm": 0.5907062317121601, "learning_rate": 9.975471385271019e-06, "loss": 1.4075, "step": 840 }, { "epoch": 0.06436552885351293, "grad_norm": 0.5611982182820167, "learning_rate": 9.975361627725593e-06, "loss": 1.113, "step": 841 }, { "epoch": 0.06444206337058013, "grad_norm": 2.2100635970367852, "learning_rate": 9.975251625922062e-06, "loss": 1.3604, "step": 842 }, { "epoch": 0.06451859788764733, "grad_norm": 0.8271895396915916, "learning_rate": 9.97514137986719e-06, "loss": 1.7099, "step": 843 }, { "epoch": 0.06459513240471453, "grad_norm": 0.5756247968761332, "learning_rate": 9.975030889567747e-06, "loss": 1.3308, "step": 844 }, { "epoch": 0.06467166692178172, "grad_norm": 0.7441894006201129, "learning_rate": 9.974920155030522e-06, "loss": 1.7783, "step": 845 }, { "epoch": 0.06474820143884892, "grad_norm": 0.649021745360654, "learning_rate": 9.974809176262323e-06, "loss": 1.5486, "step": 846 }, { "epoch": 0.06482473595591612, "grad_norm": 0.5646952825610265, "learning_rate": 9.974697953269964e-06, "loss": 1.3964, "step": 847 }, { "epoch": 0.06490127047298332, "grad_norm": 0.57344286367632, "learning_rate": 9.974586486060281e-06, "loss": 1.1975, "step": 848 }, { "epoch": 0.06497780499005051, "grad_norm": 0.6343648163526557, "learning_rate": 9.974474774640122e-06, "loss": 1.5996, "step": 849 }, { "epoch": 0.06505433950711771, "grad_norm": 0.9986071519681556, "learning_rate": 9.974362819016352e-06, "loss": 1.7085, "step": 850 }, { "epoch": 0.06513087402418491, "grad_norm": 0.6059251577055456, "learning_rate": 9.974250619195848e-06, "loss": 1.3212, "step": 851 }, { "epoch": 0.06520740854125211, "grad_norm": 0.6850244176903891, "learning_rate": 9.974138175185508e-06, "loss": 1.6407, "step": 852 }, { "epoch": 0.0652839430583193, "grad_norm": 0.591119536698554, "learning_rate": 9.974025486992235e-06, "loss": 1.5132, "step": 853 }, { "epoch": 0.0653604775753865, "grad_norm": 0.674253875849217, "learning_rate": 9.973912554622958e-06, "loss": 1.3224, "step": 854 }, { "epoch": 0.0654370120924537, "grad_norm": 0.9074620784467838, "learning_rate": 9.973799378084613e-06, "loss": 1.6958, "step": 855 }, { "epoch": 0.0655135466095209, "grad_norm": 0.5768913086251113, "learning_rate": 9.973685957384155e-06, "loss": 1.3133, "step": 856 }, { "epoch": 0.0655900811265881, "grad_norm": 0.5790561935799222, "learning_rate": 9.973572292528553e-06, "loss": 1.2392, "step": 857 }, { "epoch": 0.0656666156436553, "grad_norm": 0.5734312151396328, "learning_rate": 9.97345838352479e-06, "loss": 1.463, "step": 858 }, { "epoch": 0.06574315016072249, "grad_norm": 0.7145423113653864, "learning_rate": 9.973344230379867e-06, "loss": 1.6075, "step": 859 }, { "epoch": 0.06581968467778969, "grad_norm": 0.8004807230318164, "learning_rate": 9.973229833100794e-06, "loss": 2.0879, "step": 860 }, { "epoch": 0.06589621919485689, "grad_norm": 0.5962919895775215, "learning_rate": 9.973115191694605e-06, "loss": 1.3316, "step": 861 }, { "epoch": 0.06597275371192408, "grad_norm": 0.740150732895893, "learning_rate": 9.973000306168339e-06, "loss": 1.6657, "step": 862 }, { "epoch": 0.06604928822899128, "grad_norm": 0.5266960376402233, "learning_rate": 9.972885176529059e-06, "loss": 1.2184, "step": 863 }, { "epoch": 0.06612582274605847, "grad_norm": 0.6093283510594384, "learning_rate": 9.972769802783835e-06, "loss": 1.3772, "step": 864 }, { "epoch": 0.06620235726312566, "grad_norm": 0.5293735773899904, "learning_rate": 9.97265418493976e-06, "loss": 1.2915, "step": 865 }, { "epoch": 0.06627889178019286, "grad_norm": 0.7122395501156866, "learning_rate": 9.972538323003934e-06, "loss": 1.5324, "step": 866 }, { "epoch": 0.06635542629726006, "grad_norm": 0.5212409441900523, "learning_rate": 9.97242221698348e-06, "loss": 1.1667, "step": 867 }, { "epoch": 0.06643196081432726, "grad_norm": 0.7896637984879709, "learning_rate": 9.972305866885527e-06, "loss": 1.4727, "step": 868 }, { "epoch": 0.06650849533139445, "grad_norm": 0.7213588162485269, "learning_rate": 9.97218927271723e-06, "loss": 1.6732, "step": 869 }, { "epoch": 0.06658502984846165, "grad_norm": 0.6804655789739121, "learning_rate": 9.972072434485748e-06, "loss": 1.6139, "step": 870 }, { "epoch": 0.06666156436552885, "grad_norm": 0.5377189508907576, "learning_rate": 9.971955352198262e-06, "loss": 1.3043, "step": 871 }, { "epoch": 0.06673809888259605, "grad_norm": 0.6927136763876138, "learning_rate": 9.971838025861963e-06, "loss": 1.625, "step": 872 }, { "epoch": 0.06681463339966325, "grad_norm": 0.629362491167203, "learning_rate": 9.971720455484064e-06, "loss": 1.3819, "step": 873 }, { "epoch": 0.06689116791673044, "grad_norm": 0.5331486992293432, "learning_rate": 9.971602641071787e-06, "loss": 1.352, "step": 874 }, { "epoch": 0.06696770243379764, "grad_norm": 0.7309474854542959, "learning_rate": 9.97148458263237e-06, "loss": 1.7574, "step": 875 }, { "epoch": 0.06704423695086484, "grad_norm": 0.6119055243507305, "learning_rate": 9.971366280173069e-06, "loss": 1.5473, "step": 876 }, { "epoch": 0.06712077146793204, "grad_norm": 0.6272585095037674, "learning_rate": 9.97124773370115e-06, "loss": 1.2884, "step": 877 }, { "epoch": 0.06719730598499923, "grad_norm": 0.6016110599601847, "learning_rate": 9.9711289432239e-06, "loss": 1.5541, "step": 878 }, { "epoch": 0.06727384050206643, "grad_norm": 0.6644389765985893, "learning_rate": 9.971009908748616e-06, "loss": 1.7037, "step": 879 }, { "epoch": 0.06735037501913363, "grad_norm": 0.6234378840242308, "learning_rate": 9.97089063028261e-06, "loss": 1.101, "step": 880 }, { "epoch": 0.06742690953620083, "grad_norm": 0.5885471268402535, "learning_rate": 9.970771107833215e-06, "loss": 1.3844, "step": 881 }, { "epoch": 0.06750344405326802, "grad_norm": 0.7019508187982708, "learning_rate": 9.970651341407775e-06, "loss": 1.6694, "step": 882 }, { "epoch": 0.06757997857033522, "grad_norm": 0.606372297638377, "learning_rate": 9.970531331013644e-06, "loss": 1.5535, "step": 883 }, { "epoch": 0.06765651308740242, "grad_norm": 1.4684470253898017, "learning_rate": 9.970411076658198e-06, "loss": 1.5525, "step": 884 }, { "epoch": 0.06773304760446962, "grad_norm": 0.6582838976266405, "learning_rate": 9.970290578348826e-06, "loss": 1.3995, "step": 885 }, { "epoch": 0.06780958212153682, "grad_norm": 0.6509662562892683, "learning_rate": 9.970169836092934e-06, "loss": 1.4291, "step": 886 }, { "epoch": 0.06788611663860401, "grad_norm": 0.5785711871905647, "learning_rate": 9.970048849897937e-06, "loss": 1.361, "step": 887 }, { "epoch": 0.06796265115567121, "grad_norm": 1.1101468630881761, "learning_rate": 9.96992761977127e-06, "loss": 1.562, "step": 888 }, { "epoch": 0.06803918567273841, "grad_norm": 0.5984055078505485, "learning_rate": 9.969806145720384e-06, "loss": 1.4717, "step": 889 }, { "epoch": 0.0681157201898056, "grad_norm": 0.685964686377667, "learning_rate": 9.96968442775274e-06, "loss": 1.6879, "step": 890 }, { "epoch": 0.0681922547068728, "grad_norm": 0.6526448687442691, "learning_rate": 9.969562465875817e-06, "loss": 1.4876, "step": 891 }, { "epoch": 0.06826878922394, "grad_norm": 0.8399882885624009, "learning_rate": 9.969440260097112e-06, "loss": 1.4359, "step": 892 }, { "epoch": 0.0683453237410072, "grad_norm": 0.6897510131628932, "learning_rate": 9.969317810424128e-06, "loss": 1.5343, "step": 893 }, { "epoch": 0.0684218582580744, "grad_norm": 0.9718197060552433, "learning_rate": 9.969195116864393e-06, "loss": 1.6157, "step": 894 }, { "epoch": 0.0684983927751416, "grad_norm": 0.7931282744116118, "learning_rate": 9.969072179425445e-06, "loss": 1.337, "step": 895 }, { "epoch": 0.06857492729220879, "grad_norm": 0.5768873063323314, "learning_rate": 9.968948998114836e-06, "loss": 1.2951, "step": 896 }, { "epoch": 0.06865146180927599, "grad_norm": 0.5190078586485717, "learning_rate": 9.968825572940136e-06, "loss": 1.112, "step": 897 }, { "epoch": 0.06872799632634317, "grad_norm": 0.8853032565936291, "learning_rate": 9.968701903908929e-06, "loss": 1.6256, "step": 898 }, { "epoch": 0.06880453084341037, "grad_norm": 0.8279615365357481, "learning_rate": 9.968577991028811e-06, "loss": 1.549, "step": 899 }, { "epoch": 0.06888106536047757, "grad_norm": 0.5300616543058274, "learning_rate": 9.9684538343074e-06, "loss": 1.0752, "step": 900 }, { "epoch": 0.06895759987754477, "grad_norm": 0.5806380597896166, "learning_rate": 9.96832943375232e-06, "loss": 1.3588, "step": 901 }, { "epoch": 0.06903413439461197, "grad_norm": 0.6223343541045858, "learning_rate": 9.968204789371216e-06, "loss": 1.2883, "step": 902 }, { "epoch": 0.06911066891167916, "grad_norm": 0.7191554860902811, "learning_rate": 9.968079901171747e-06, "loss": 1.7902, "step": 903 }, { "epoch": 0.06918720342874636, "grad_norm": 0.6745403315076588, "learning_rate": 9.967954769161589e-06, "loss": 1.1497, "step": 904 }, { "epoch": 0.06926373794581356, "grad_norm": 0.5823766335833943, "learning_rate": 9.967829393348427e-06, "loss": 1.3017, "step": 905 }, { "epoch": 0.06934027246288076, "grad_norm": 0.6449336652011257, "learning_rate": 9.967703773739965e-06, "loss": 1.619, "step": 906 }, { "epoch": 0.06941680697994795, "grad_norm": 0.6845426729808655, "learning_rate": 9.967577910343921e-06, "loss": 1.0818, "step": 907 }, { "epoch": 0.06949334149701515, "grad_norm": 0.6946055926909906, "learning_rate": 9.967451803168031e-06, "loss": 1.6398, "step": 908 }, { "epoch": 0.06956987601408235, "grad_norm": 0.5490633728585079, "learning_rate": 9.967325452220042e-06, "loss": 1.0184, "step": 909 }, { "epoch": 0.06964641053114955, "grad_norm": 0.675302990121441, "learning_rate": 9.967198857507717e-06, "loss": 1.5921, "step": 910 }, { "epoch": 0.06972294504821674, "grad_norm": 0.7107201990742179, "learning_rate": 9.967072019038834e-06, "loss": 1.5862, "step": 911 }, { "epoch": 0.06979947956528394, "grad_norm": 0.9576416160325513, "learning_rate": 9.966944936821187e-06, "loss": 1.6836, "step": 912 }, { "epoch": 0.06987601408235114, "grad_norm": 0.6553142090808683, "learning_rate": 9.966817610862585e-06, "loss": 1.6699, "step": 913 }, { "epoch": 0.06995254859941834, "grad_norm": 0.6527328940343784, "learning_rate": 9.96669004117085e-06, "loss": 1.5822, "step": 914 }, { "epoch": 0.07002908311648554, "grad_norm": 0.7395568982045879, "learning_rate": 9.96656222775382e-06, "loss": 1.7038, "step": 915 }, { "epoch": 0.07010561763355273, "grad_norm": 0.9457002628849891, "learning_rate": 9.966434170619347e-06, "loss": 1.5414, "step": 916 }, { "epoch": 0.07018215215061993, "grad_norm": 0.7195950763204083, "learning_rate": 9.966305869775306e-06, "loss": 1.8462, "step": 917 }, { "epoch": 0.07025868666768713, "grad_norm": 0.5905947551423899, "learning_rate": 9.966177325229572e-06, "loss": 1.3622, "step": 918 }, { "epoch": 0.07033522118475433, "grad_norm": 0.612708391189701, "learning_rate": 9.966048536990048e-06, "loss": 1.4584, "step": 919 }, { "epoch": 0.07041175570182152, "grad_norm": 0.5017963656030745, "learning_rate": 9.965919505064644e-06, "loss": 1.2704, "step": 920 }, { "epoch": 0.07048829021888872, "grad_norm": 0.6724769803794114, "learning_rate": 9.965790229461291e-06, "loss": 1.5476, "step": 921 }, { "epoch": 0.07056482473595592, "grad_norm": 0.5735465780592652, "learning_rate": 9.965660710187931e-06, "loss": 1.4403, "step": 922 }, { "epoch": 0.07064135925302312, "grad_norm": 0.7731296562156847, "learning_rate": 9.965530947252522e-06, "loss": 1.4182, "step": 923 }, { "epoch": 0.07071789377009031, "grad_norm": 0.7102547530531528, "learning_rate": 9.965400940663037e-06, "loss": 1.6978, "step": 924 }, { "epoch": 0.07079442828715751, "grad_norm": 0.7044517952547602, "learning_rate": 9.965270690427464e-06, "loss": 1.5766, "step": 925 }, { "epoch": 0.07087096280422471, "grad_norm": 0.9625773245735932, "learning_rate": 9.965140196553804e-06, "loss": 1.5176, "step": 926 }, { "epoch": 0.07094749732129191, "grad_norm": 0.5549886290371676, "learning_rate": 9.965009459050078e-06, "loss": 1.1741, "step": 927 }, { "epoch": 0.0710240318383591, "grad_norm": 0.5842372915581019, "learning_rate": 9.964878477924319e-06, "loss": 1.4929, "step": 928 }, { "epoch": 0.0711005663554263, "grad_norm": 0.7432487300213058, "learning_rate": 9.964747253184572e-06, "loss": 1.5069, "step": 929 }, { "epoch": 0.0711771008724935, "grad_norm": 0.7355363919173148, "learning_rate": 9.9646157848389e-06, "loss": 1.4957, "step": 930 }, { "epoch": 0.07125363538956068, "grad_norm": 0.5738055468031754, "learning_rate": 9.964484072895385e-06, "loss": 1.3249, "step": 931 }, { "epoch": 0.07133016990662788, "grad_norm": 0.7591863472133552, "learning_rate": 9.964352117362116e-06, "loss": 1.802, "step": 932 }, { "epoch": 0.07140670442369508, "grad_norm": 0.6943197038078134, "learning_rate": 9.964219918247202e-06, "loss": 1.8274, "step": 933 }, { "epoch": 0.07148323894076228, "grad_norm": 0.9414705593614743, "learning_rate": 9.964087475558765e-06, "loss": 1.9768, "step": 934 }, { "epoch": 0.07155977345782948, "grad_norm": 0.6477619556897418, "learning_rate": 9.963954789304944e-06, "loss": 1.6981, "step": 935 }, { "epoch": 0.07163630797489667, "grad_norm": 0.7868803634229027, "learning_rate": 9.96382185949389e-06, "loss": 1.5809, "step": 936 }, { "epoch": 0.07171284249196387, "grad_norm": 0.5592150365396008, "learning_rate": 9.963688686133772e-06, "loss": 1.2865, "step": 937 }, { "epoch": 0.07178937700903107, "grad_norm": 0.888029875059795, "learning_rate": 9.963555269232772e-06, "loss": 1.5448, "step": 938 }, { "epoch": 0.07186591152609827, "grad_norm": 0.6267606003141575, "learning_rate": 9.963421608799087e-06, "loss": 1.6014, "step": 939 }, { "epoch": 0.07194244604316546, "grad_norm": 0.6781460083719826, "learning_rate": 9.96328770484093e-06, "loss": 1.6504, "step": 940 }, { "epoch": 0.07201898056023266, "grad_norm": 0.5360629092930747, "learning_rate": 9.963153557366527e-06, "loss": 1.3083, "step": 941 }, { "epoch": 0.07209551507729986, "grad_norm": 0.8029883157642667, "learning_rate": 9.963019166384125e-06, "loss": 1.6199, "step": 942 }, { "epoch": 0.07217204959436706, "grad_norm": 0.7336839512439449, "learning_rate": 9.962884531901975e-06, "loss": 1.6002, "step": 943 }, { "epoch": 0.07224858411143426, "grad_norm": 0.895567853715097, "learning_rate": 9.962749653928354e-06, "loss": 1.6783, "step": 944 }, { "epoch": 0.07232511862850145, "grad_norm": 0.7228740367928783, "learning_rate": 9.962614532471548e-06, "loss": 1.5127, "step": 945 }, { "epoch": 0.07240165314556865, "grad_norm": 0.63344538599879, "learning_rate": 9.96247916753986e-06, "loss": 1.4352, "step": 946 }, { "epoch": 0.07247818766263585, "grad_norm": 0.632605539384707, "learning_rate": 9.962343559141606e-06, "loss": 1.4371, "step": 947 }, { "epoch": 0.07255472217970305, "grad_norm": 0.627392918244427, "learning_rate": 9.962207707285118e-06, "loss": 1.2965, "step": 948 }, { "epoch": 0.07263125669677024, "grad_norm": 0.6385217687300322, "learning_rate": 9.962071611978744e-06, "loss": 1.5529, "step": 949 }, { "epoch": 0.07270779121383744, "grad_norm": 0.8710117545572564, "learning_rate": 9.961935273230844e-06, "loss": 1.6654, "step": 950 }, { "epoch": 0.07278432573090464, "grad_norm": 0.6555384824450418, "learning_rate": 9.9617986910498e-06, "loss": 1.5397, "step": 951 }, { "epoch": 0.07286086024797184, "grad_norm": 0.9441755815461472, "learning_rate": 9.961661865444e-06, "loss": 1.4475, "step": 952 }, { "epoch": 0.07293739476503903, "grad_norm": 1.165152304285993, "learning_rate": 9.961524796421851e-06, "loss": 1.3326, "step": 953 }, { "epoch": 0.07301392928210623, "grad_norm": 0.5535906637766458, "learning_rate": 9.961387483991775e-06, "loss": 1.3445, "step": 954 }, { "epoch": 0.07309046379917343, "grad_norm": 0.7138871197650903, "learning_rate": 9.961249928162211e-06, "loss": 1.5247, "step": 955 }, { "epoch": 0.07316699831624063, "grad_norm": 1.5039008075607763, "learning_rate": 9.96111212894161e-06, "loss": 1.3025, "step": 956 }, { "epoch": 0.07324353283330783, "grad_norm": 0.6979913827209157, "learning_rate": 9.960974086338438e-06, "loss": 1.6826, "step": 957 }, { "epoch": 0.07332006735037502, "grad_norm": 0.6189237269033024, "learning_rate": 9.960835800361176e-06, "loss": 1.1766, "step": 958 }, { "epoch": 0.07339660186744222, "grad_norm": 0.6108172392590948, "learning_rate": 9.960697271018323e-06, "loss": 1.3948, "step": 959 }, { "epoch": 0.07347313638450942, "grad_norm": 0.5986332662467829, "learning_rate": 9.96055849831839e-06, "loss": 1.4659, "step": 960 }, { "epoch": 0.07354967090157662, "grad_norm": 0.7867575983664218, "learning_rate": 9.9604194822699e-06, "loss": 1.6885, "step": 961 }, { "epoch": 0.07362620541864381, "grad_norm": 0.6705349593972433, "learning_rate": 9.9602802228814e-06, "loss": 1.3503, "step": 962 }, { "epoch": 0.07370273993571101, "grad_norm": 0.5710121054789417, "learning_rate": 9.960140720161445e-06, "loss": 1.3753, "step": 963 }, { "epoch": 0.07377927445277821, "grad_norm": 0.7564590114157599, "learning_rate": 9.960000974118604e-06, "loss": 1.4386, "step": 964 }, { "epoch": 0.0738558089698454, "grad_norm": 0.6723650430008026, "learning_rate": 9.959860984761465e-06, "loss": 1.615, "step": 965 }, { "epoch": 0.07393234348691259, "grad_norm": 0.6151823067894785, "learning_rate": 9.95972075209863e-06, "loss": 1.5337, "step": 966 }, { "epoch": 0.07400887800397979, "grad_norm": 0.7308392151176611, "learning_rate": 9.959580276138715e-06, "loss": 1.5502, "step": 967 }, { "epoch": 0.07408541252104699, "grad_norm": 0.8506064901498159, "learning_rate": 9.95943955689035e-06, "loss": 1.7709, "step": 968 }, { "epoch": 0.07416194703811418, "grad_norm": 0.8584970232875548, "learning_rate": 9.959298594362183e-06, "loss": 1.5881, "step": 969 }, { "epoch": 0.07423848155518138, "grad_norm": 0.6320223013063146, "learning_rate": 9.959157388562871e-06, "loss": 1.6163, "step": 970 }, { "epoch": 0.07431501607224858, "grad_norm": 0.7960836006527916, "learning_rate": 9.959015939501098e-06, "loss": 1.3506, "step": 971 }, { "epoch": 0.07439155058931578, "grad_norm": 0.9259138890924267, "learning_rate": 9.95887424718555e-06, "loss": 1.3193, "step": 972 }, { "epoch": 0.07446808510638298, "grad_norm": 0.6551027493781804, "learning_rate": 9.95873231162493e-06, "loss": 1.6214, "step": 973 }, { "epoch": 0.07454461962345017, "grad_norm": 0.5244053719527695, "learning_rate": 9.958590132827965e-06, "loss": 1.057, "step": 974 }, { "epoch": 0.07462115414051737, "grad_norm": 0.8313740871489422, "learning_rate": 9.958447710803388e-06, "loss": 1.7204, "step": 975 }, { "epoch": 0.07469768865758457, "grad_norm": 0.6441257661056092, "learning_rate": 9.95830504555995e-06, "loss": 1.441, "step": 976 }, { "epoch": 0.07477422317465177, "grad_norm": 0.709166810247056, "learning_rate": 9.958162137106418e-06, "loss": 1.7438, "step": 977 }, { "epoch": 0.07485075769171896, "grad_norm": 0.5469284660881109, "learning_rate": 9.95801898545157e-06, "loss": 1.3318, "step": 978 }, { "epoch": 0.07492729220878616, "grad_norm": 0.6375217305845663, "learning_rate": 9.957875590604205e-06, "loss": 1.363, "step": 979 }, { "epoch": 0.07500382672585336, "grad_norm": 0.6208778656740401, "learning_rate": 9.95773195257313e-06, "loss": 1.4009, "step": 980 }, { "epoch": 0.07508036124292056, "grad_norm": 0.613169109612027, "learning_rate": 9.957588071367172e-06, "loss": 1.3755, "step": 981 }, { "epoch": 0.07515689575998775, "grad_norm": 1.5212057356971649, "learning_rate": 9.957443946995174e-06, "loss": 1.5969, "step": 982 }, { "epoch": 0.07523343027705495, "grad_norm": 0.5828524039417812, "learning_rate": 9.957299579465989e-06, "loss": 1.1411, "step": 983 }, { "epoch": 0.07530996479412215, "grad_norm": 0.6073218708815911, "learning_rate": 9.957154968788488e-06, "loss": 1.5464, "step": 984 }, { "epoch": 0.07538649931118935, "grad_norm": 0.5855297630665558, "learning_rate": 9.957010114971555e-06, "loss": 1.2005, "step": 985 }, { "epoch": 0.07546303382825655, "grad_norm": 0.5722168802085099, "learning_rate": 9.956865018024092e-06, "loss": 1.4157, "step": 986 }, { "epoch": 0.07553956834532374, "grad_norm": 0.6029689523798507, "learning_rate": 9.956719677955013e-06, "loss": 1.3476, "step": 987 }, { "epoch": 0.07561610286239094, "grad_norm": 0.5617763554593896, "learning_rate": 9.956574094773249e-06, "loss": 1.2128, "step": 988 }, { "epoch": 0.07569263737945814, "grad_norm": 0.580468511021025, "learning_rate": 9.956428268487743e-06, "loss": 1.3957, "step": 989 }, { "epoch": 0.07576917189652534, "grad_norm": 0.773124632042539, "learning_rate": 9.95628219910746e-06, "loss": 1.8493, "step": 990 }, { "epoch": 0.07584570641359253, "grad_norm": 0.6978880410864485, "learning_rate": 9.956135886641368e-06, "loss": 1.7131, "step": 991 }, { "epoch": 0.07592224093065973, "grad_norm": 0.5614340076184756, "learning_rate": 9.955989331098463e-06, "loss": 1.2599, "step": 992 }, { "epoch": 0.07599877544772693, "grad_norm": 0.7535284772045969, "learning_rate": 9.955842532487747e-06, "loss": 1.8059, "step": 993 }, { "epoch": 0.07607530996479413, "grad_norm": 0.6172628475751759, "learning_rate": 9.955695490818242e-06, "loss": 1.5059, "step": 994 }, { "epoch": 0.07615184448186132, "grad_norm": 1.945876709637634, "learning_rate": 9.955548206098979e-06, "loss": 1.6508, "step": 995 }, { "epoch": 0.07622837899892852, "grad_norm": 0.5711035019250575, "learning_rate": 9.955400678339009e-06, "loss": 1.2939, "step": 996 }, { "epoch": 0.07630491351599572, "grad_norm": 0.6334450816188955, "learning_rate": 9.955252907547396e-06, "loss": 1.5932, "step": 997 }, { "epoch": 0.07638144803306292, "grad_norm": 0.6023154708797589, "learning_rate": 9.955104893733221e-06, "loss": 1.4297, "step": 998 }, { "epoch": 0.0764579825501301, "grad_norm": 0.6474915559567057, "learning_rate": 9.95495663690558e-06, "loss": 1.5364, "step": 999 }, { "epoch": 0.0765345170671973, "grad_norm": 1.1679326969448007, "learning_rate": 9.954808137073579e-06, "loss": 1.7706, "step": 1000 }, { "epoch": 0.0766110515842645, "grad_norm": 0.5333676014125461, "learning_rate": 9.954659394246341e-06, "loss": 1.2474, "step": 1001 }, { "epoch": 0.0766875861013317, "grad_norm": 0.6684291903503895, "learning_rate": 9.95451040843301e-06, "loss": 1.7271, "step": 1002 }, { "epoch": 0.07676412061839889, "grad_norm": 0.5951904382233838, "learning_rate": 9.954361179642738e-06, "loss": 1.3903, "step": 1003 }, { "epoch": 0.07684065513546609, "grad_norm": 0.6822836632923943, "learning_rate": 9.954211707884693e-06, "loss": 1.4258, "step": 1004 }, { "epoch": 0.07691718965253329, "grad_norm": 0.7469475872364464, "learning_rate": 9.95406199316806e-06, "loss": 1.3381, "step": 1005 }, { "epoch": 0.07699372416960049, "grad_norm": 0.590363935959304, "learning_rate": 9.953912035502037e-06, "loss": 1.2442, "step": 1006 }, { "epoch": 0.07707025868666768, "grad_norm": 0.7482822528730586, "learning_rate": 9.95376183489584e-06, "loss": 2.0366, "step": 1007 }, { "epoch": 0.07714679320373488, "grad_norm": 0.5708088941209815, "learning_rate": 9.953611391358696e-06, "loss": 1.3276, "step": 1008 }, { "epoch": 0.07722332772080208, "grad_norm": 0.5790592389946856, "learning_rate": 9.953460704899847e-06, "loss": 1.2638, "step": 1009 }, { "epoch": 0.07729986223786928, "grad_norm": 0.6857853100968089, "learning_rate": 9.953309775528557e-06, "loss": 1.5022, "step": 1010 }, { "epoch": 0.07737639675493647, "grad_norm": 0.6846639630180594, "learning_rate": 9.953158603254093e-06, "loss": 1.5389, "step": 1011 }, { "epoch": 0.07745293127200367, "grad_norm": 0.6790176730304266, "learning_rate": 9.95300718808575e-06, "loss": 1.6631, "step": 1012 }, { "epoch": 0.07752946578907087, "grad_norm": 1.6971018686854757, "learning_rate": 9.952855530032826e-06, "loss": 1.4983, "step": 1013 }, { "epoch": 0.07760600030613807, "grad_norm": 1.1179687953832518, "learning_rate": 9.952703629104642e-06, "loss": 1.5126, "step": 1014 }, { "epoch": 0.07768253482320527, "grad_norm": 0.6053908844676522, "learning_rate": 9.95255148531053e-06, "loss": 1.446, "step": 1015 }, { "epoch": 0.07775906934027246, "grad_norm": 0.4995583656571746, "learning_rate": 9.95239909865984e-06, "loss": 0.8732, "step": 1016 }, { "epoch": 0.07783560385733966, "grad_norm": 0.5563071149490703, "learning_rate": 9.952246469161934e-06, "loss": 1.2185, "step": 1017 }, { "epoch": 0.07791213837440686, "grad_norm": 0.6340412743341894, "learning_rate": 9.95209359682619e-06, "loss": 1.4256, "step": 1018 }, { "epoch": 0.07798867289147406, "grad_norm": 0.5719057229418726, "learning_rate": 9.951940481661999e-06, "loss": 1.4134, "step": 1019 }, { "epoch": 0.07806520740854125, "grad_norm": 0.8745541335629397, "learning_rate": 9.951787123678772e-06, "loss": 1.7062, "step": 1020 }, { "epoch": 0.07814174192560845, "grad_norm": 0.5928272617982239, "learning_rate": 9.95163352288593e-06, "loss": 1.3935, "step": 1021 }, { "epoch": 0.07821827644267565, "grad_norm": 0.7149650022784527, "learning_rate": 9.951479679292914e-06, "loss": 1.2713, "step": 1022 }, { "epoch": 0.07829481095974285, "grad_norm": 0.7210159121619327, "learning_rate": 9.951325592909171e-06, "loss": 1.5251, "step": 1023 }, { "epoch": 0.07837134547681004, "grad_norm": 2.908074443142246, "learning_rate": 9.95117126374417e-06, "loss": 1.5527, "step": 1024 }, { "epoch": 0.07844787999387724, "grad_norm": 0.8493017319135755, "learning_rate": 9.951016691807398e-06, "loss": 1.6233, "step": 1025 }, { "epoch": 0.07852441451094444, "grad_norm": 0.5632588191182499, "learning_rate": 9.95086187710835e-06, "loss": 1.2386, "step": 1026 }, { "epoch": 0.07860094902801164, "grad_norm": 0.6187901788411129, "learning_rate": 9.950706819656533e-06, "loss": 1.439, "step": 1027 }, { "epoch": 0.07867748354507884, "grad_norm": 0.5823961901162658, "learning_rate": 9.950551519461481e-06, "loss": 1.3037, "step": 1028 }, { "epoch": 0.07875401806214603, "grad_norm": 0.7534023542631006, "learning_rate": 9.950395976532735e-06, "loss": 1.6582, "step": 1029 }, { "epoch": 0.07883055257921323, "grad_norm": 0.6211280065509629, "learning_rate": 9.950240190879848e-06, "loss": 1.5672, "step": 1030 }, { "epoch": 0.07890708709628043, "grad_norm": 0.6425765674668426, "learning_rate": 9.950084162512398e-06, "loss": 1.4581, "step": 1031 }, { "epoch": 0.07898362161334763, "grad_norm": 0.6925100049061822, "learning_rate": 9.949927891439967e-06, "loss": 1.607, "step": 1032 }, { "epoch": 0.07906015613041481, "grad_norm": 0.7595373022807022, "learning_rate": 9.949771377672157e-06, "loss": 1.5096, "step": 1033 }, { "epoch": 0.07913669064748201, "grad_norm": 0.83722028523324, "learning_rate": 9.949614621218587e-06, "loss": 1.8011, "step": 1034 }, { "epoch": 0.0792132251645492, "grad_norm": 0.6907899956706465, "learning_rate": 9.949457622088888e-06, "loss": 1.6486, "step": 1035 }, { "epoch": 0.0792897596816164, "grad_norm": 0.5283381132415681, "learning_rate": 9.949300380292706e-06, "loss": 0.9786, "step": 1036 }, { "epoch": 0.0793662941986836, "grad_norm": 0.658214720547626, "learning_rate": 9.949142895839701e-06, "loss": 1.5328, "step": 1037 }, { "epoch": 0.0794428287157508, "grad_norm": 0.8478817710203205, "learning_rate": 9.948985168739554e-06, "loss": 1.9755, "step": 1038 }, { "epoch": 0.079519363232818, "grad_norm": 1.4427254403739187, "learning_rate": 9.948827199001949e-06, "loss": 1.5031, "step": 1039 }, { "epoch": 0.0795958977498852, "grad_norm": 0.627631649831169, "learning_rate": 9.948668986636599e-06, "loss": 1.1381, "step": 1040 }, { "epoch": 0.07967243226695239, "grad_norm": 0.6513222140435198, "learning_rate": 9.948510531653223e-06, "loss": 1.6242, "step": 1041 }, { "epoch": 0.07974896678401959, "grad_norm": 0.6973297812574696, "learning_rate": 9.948351834061553e-06, "loss": 1.5916, "step": 1042 }, { "epoch": 0.07982550130108679, "grad_norm": 0.5956261333860855, "learning_rate": 9.948192893871344e-06, "loss": 1.1633, "step": 1043 }, { "epoch": 0.07990203581815399, "grad_norm": 0.6825547979354047, "learning_rate": 9.948033711092362e-06, "loss": 1.5587, "step": 1044 }, { "epoch": 0.07997857033522118, "grad_norm": 1.2968954879936616, "learning_rate": 9.947874285734385e-06, "loss": 1.4635, "step": 1045 }, { "epoch": 0.08005510485228838, "grad_norm": 0.7251697049259326, "learning_rate": 9.947714617807212e-06, "loss": 1.6429, "step": 1046 }, { "epoch": 0.08013163936935558, "grad_norm": 0.6333560468861407, "learning_rate": 9.94755470732065e-06, "loss": 1.4294, "step": 1047 }, { "epoch": 0.08020817388642278, "grad_norm": 2.1601681957814245, "learning_rate": 9.947394554284526e-06, "loss": 2.0344, "step": 1048 }, { "epoch": 0.08028470840348997, "grad_norm": 0.5140067112114642, "learning_rate": 9.947234158708678e-06, "loss": 1.0586, "step": 1049 }, { "epoch": 0.08036124292055717, "grad_norm": 0.562430836110291, "learning_rate": 9.947073520602968e-06, "loss": 1.2466, "step": 1050 }, { "epoch": 0.08043777743762437, "grad_norm": 0.6824871406646882, "learning_rate": 9.946912639977257e-06, "loss": 1.6452, "step": 1051 }, { "epoch": 0.08051431195469157, "grad_norm": 0.59218583663795, "learning_rate": 9.946751516841438e-06, "loss": 1.3836, "step": 1052 }, { "epoch": 0.08059084647175876, "grad_norm": 0.744769332323476, "learning_rate": 9.946590151205405e-06, "loss": 1.8331, "step": 1053 }, { "epoch": 0.08066738098882596, "grad_norm": 0.6173201226506037, "learning_rate": 9.946428543079076e-06, "loss": 1.4055, "step": 1054 }, { "epoch": 0.08074391550589316, "grad_norm": 0.5208845976683911, "learning_rate": 9.946266692472379e-06, "loss": 1.2173, "step": 1055 }, { "epoch": 0.08082045002296036, "grad_norm": 0.6555111272059218, "learning_rate": 9.94610459939526e-06, "loss": 1.5379, "step": 1056 }, { "epoch": 0.08089698454002756, "grad_norm": 0.539818452635486, "learning_rate": 9.945942263857676e-06, "loss": 1.2387, "step": 1057 }, { "epoch": 0.08097351905709475, "grad_norm": 0.5974834710543152, "learning_rate": 9.945779685869604e-06, "loss": 1.3857, "step": 1058 }, { "epoch": 0.08105005357416195, "grad_norm": 0.8002822132751807, "learning_rate": 9.945616865441032e-06, "loss": 2.0809, "step": 1059 }, { "epoch": 0.08112658809122915, "grad_norm": 0.6372199003097274, "learning_rate": 9.945453802581967e-06, "loss": 1.4714, "step": 1060 }, { "epoch": 0.08120312260829635, "grad_norm": 0.5575358695858924, "learning_rate": 9.945290497302424e-06, "loss": 1.3153, "step": 1061 }, { "epoch": 0.08127965712536354, "grad_norm": 0.6441213342614732, "learning_rate": 9.945126949612439e-06, "loss": 1.3977, "step": 1062 }, { "epoch": 0.08135619164243074, "grad_norm": 0.6292866522423622, "learning_rate": 9.944963159522062e-06, "loss": 1.4144, "step": 1063 }, { "epoch": 0.08143272615949794, "grad_norm": 0.608103847046049, "learning_rate": 9.944799127041353e-06, "loss": 1.3576, "step": 1064 }, { "epoch": 0.08150926067656514, "grad_norm": 0.5549809670682626, "learning_rate": 9.944634852180393e-06, "loss": 1.3317, "step": 1065 }, { "epoch": 0.08158579519363234, "grad_norm": 0.5749067520264368, "learning_rate": 9.944470334949276e-06, "loss": 1.1488, "step": 1066 }, { "epoch": 0.08166232971069952, "grad_norm": 0.5527481710852837, "learning_rate": 9.944305575358111e-06, "loss": 1.1334, "step": 1067 }, { "epoch": 0.08173886422776672, "grad_norm": 0.7189096939829165, "learning_rate": 9.944140573417019e-06, "loss": 1.8373, "step": 1068 }, { "epoch": 0.08181539874483391, "grad_norm": 0.8343970798310545, "learning_rate": 9.943975329136139e-06, "loss": 1.4754, "step": 1069 }, { "epoch": 0.08189193326190111, "grad_norm": 0.5655930180577801, "learning_rate": 9.943809842525627e-06, "loss": 1.4325, "step": 1070 }, { "epoch": 0.08196846777896831, "grad_norm": 0.647235243447947, "learning_rate": 9.943644113595646e-06, "loss": 1.6915, "step": 1071 }, { "epoch": 0.08204500229603551, "grad_norm": 0.8677123355560282, "learning_rate": 9.943478142356383e-06, "loss": 1.4909, "step": 1072 }, { "epoch": 0.0821215368131027, "grad_norm": 0.6123814196354124, "learning_rate": 9.943311928818032e-06, "loss": 1.4443, "step": 1073 }, { "epoch": 0.0821980713301699, "grad_norm": 0.6167742225769007, "learning_rate": 9.943145472990809e-06, "loss": 1.5546, "step": 1074 }, { "epoch": 0.0822746058472371, "grad_norm": 0.6711200202881026, "learning_rate": 9.942978774884942e-06, "loss": 1.6377, "step": 1075 }, { "epoch": 0.0823511403643043, "grad_norm": 0.5218917040186817, "learning_rate": 9.94281183451067e-06, "loss": 1.2038, "step": 1076 }, { "epoch": 0.0824276748813715, "grad_norm": 0.5778890001237121, "learning_rate": 9.942644651878253e-06, "loss": 1.4764, "step": 1077 }, { "epoch": 0.0825042093984387, "grad_norm": 0.6356287020111566, "learning_rate": 9.942477226997961e-06, "loss": 1.5224, "step": 1078 }, { "epoch": 0.08258074391550589, "grad_norm": 0.6418587741576106, "learning_rate": 9.942309559880085e-06, "loss": 1.5568, "step": 1079 }, { "epoch": 0.08265727843257309, "grad_norm": 0.539253709881528, "learning_rate": 9.942141650534922e-06, "loss": 1.2191, "step": 1080 }, { "epoch": 0.08273381294964029, "grad_norm": 0.6234201343406232, "learning_rate": 9.941973498972793e-06, "loss": 1.335, "step": 1081 }, { "epoch": 0.08281034746670748, "grad_norm": 0.5172375818523512, "learning_rate": 9.941805105204025e-06, "loss": 0.9458, "step": 1082 }, { "epoch": 0.08288688198377468, "grad_norm": 0.5872585682759218, "learning_rate": 9.941636469238968e-06, "loss": 1.1743, "step": 1083 }, { "epoch": 0.08296341650084188, "grad_norm": 0.5422024412364421, "learning_rate": 9.941467591087986e-06, "loss": 1.1219, "step": 1084 }, { "epoch": 0.08303995101790908, "grad_norm": 0.7073550240062311, "learning_rate": 9.941298470761451e-06, "loss": 1.4993, "step": 1085 }, { "epoch": 0.08311648553497628, "grad_norm": 0.5691358877488627, "learning_rate": 9.941129108269755e-06, "loss": 1.2894, "step": 1086 }, { "epoch": 0.08319302005204347, "grad_norm": 0.6552143914466361, "learning_rate": 9.940959503623306e-06, "loss": 1.4598, "step": 1087 }, { "epoch": 0.08326955456911067, "grad_norm": 0.5510184066135605, "learning_rate": 9.940789656832524e-06, "loss": 1.2976, "step": 1088 }, { "epoch": 0.08334608908617787, "grad_norm": 0.5569695481122887, "learning_rate": 9.940619567907844e-06, "loss": 1.1264, "step": 1089 }, { "epoch": 0.08342262360324507, "grad_norm": 0.6001205482708065, "learning_rate": 9.940449236859717e-06, "loss": 1.3864, "step": 1090 }, { "epoch": 0.08349915812031226, "grad_norm": 0.6703771328058044, "learning_rate": 9.94027866369861e-06, "loss": 1.7648, "step": 1091 }, { "epoch": 0.08357569263737946, "grad_norm": 0.8965689454085151, "learning_rate": 9.940107848435004e-06, "loss": 1.5619, "step": 1092 }, { "epoch": 0.08365222715444666, "grad_norm": 0.5770163166821443, "learning_rate": 9.939936791079392e-06, "loss": 1.3691, "step": 1093 }, { "epoch": 0.08372876167151386, "grad_norm": 0.8326474379313283, "learning_rate": 9.939765491642285e-06, "loss": 1.7061, "step": 1094 }, { "epoch": 0.08380529618858105, "grad_norm": 0.6543248969250657, "learning_rate": 9.939593950134207e-06, "loss": 1.4915, "step": 1095 }, { "epoch": 0.08388183070564825, "grad_norm": 0.6188263855007177, "learning_rate": 9.939422166565704e-06, "loss": 1.5469, "step": 1096 }, { "epoch": 0.08395836522271545, "grad_norm": 0.6695880834437925, "learning_rate": 9.939250140947322e-06, "loss": 1.3547, "step": 1097 }, { "epoch": 0.08403489973978265, "grad_norm": 0.7201379784012408, "learning_rate": 9.939077873289637e-06, "loss": 1.6534, "step": 1098 }, { "epoch": 0.08411143425684985, "grad_norm": 0.780014862323792, "learning_rate": 9.938905363603232e-06, "loss": 1.3677, "step": 1099 }, { "epoch": 0.08418796877391704, "grad_norm": 0.5089503792528731, "learning_rate": 9.938732611898706e-06, "loss": 0.9067, "step": 1100 }, { "epoch": 0.08426450329098423, "grad_norm": 0.6626229756511384, "learning_rate": 9.938559618186674e-06, "loss": 1.4902, "step": 1101 }, { "epoch": 0.08434103780805143, "grad_norm": 0.6398969202954031, "learning_rate": 9.938386382477764e-06, "loss": 1.5077, "step": 1102 }, { "epoch": 0.08441757232511862, "grad_norm": 0.7314593398563481, "learning_rate": 9.938212904782622e-06, "loss": 1.5674, "step": 1103 }, { "epoch": 0.08449410684218582, "grad_norm": 0.599843519569803, "learning_rate": 9.938039185111907e-06, "loss": 1.3266, "step": 1104 }, { "epoch": 0.08457064135925302, "grad_norm": 0.5670442886437143, "learning_rate": 9.93786522347629e-06, "loss": 1.2598, "step": 1105 }, { "epoch": 0.08464717587632022, "grad_norm": 0.5749480842934815, "learning_rate": 9.937691019886463e-06, "loss": 1.3477, "step": 1106 }, { "epoch": 0.08472371039338741, "grad_norm": 0.6961965865706542, "learning_rate": 9.93751657435313e-06, "loss": 1.8622, "step": 1107 }, { "epoch": 0.08480024491045461, "grad_norm": 0.6323085750871227, "learning_rate": 9.937341886887006e-06, "loss": 1.5849, "step": 1108 }, { "epoch": 0.08487677942752181, "grad_norm": 0.7417686012827166, "learning_rate": 9.937166957498824e-06, "loss": 1.7742, "step": 1109 }, { "epoch": 0.084953313944589, "grad_norm": 0.6850647644066823, "learning_rate": 9.936991786199338e-06, "loss": 1.6433, "step": 1110 }, { "epoch": 0.0850298484616562, "grad_norm": 0.8994100950941287, "learning_rate": 9.936816372999305e-06, "loss": 1.5112, "step": 1111 }, { "epoch": 0.0851063829787234, "grad_norm": 0.6852186481875614, "learning_rate": 9.936640717909505e-06, "loss": 1.4311, "step": 1112 }, { "epoch": 0.0851829174957906, "grad_norm": 0.5831407547335783, "learning_rate": 9.936464820940733e-06, "loss": 1.2528, "step": 1113 }, { "epoch": 0.0852594520128578, "grad_norm": 0.5636653896541369, "learning_rate": 9.936288682103791e-06, "loss": 1.3916, "step": 1114 }, { "epoch": 0.085335986529925, "grad_norm": 0.5744813797683082, "learning_rate": 9.93611230140951e-06, "loss": 1.3099, "step": 1115 }, { "epoch": 0.08541252104699219, "grad_norm": 0.6438973556022218, "learning_rate": 9.935935678868718e-06, "loss": 1.6352, "step": 1116 }, { "epoch": 0.08548905556405939, "grad_norm": 0.5996696182564107, "learning_rate": 9.935758814492273e-06, "loss": 1.2651, "step": 1117 }, { "epoch": 0.08556559008112659, "grad_norm": 0.6676807147164441, "learning_rate": 9.935581708291043e-06, "loss": 1.6358, "step": 1118 }, { "epoch": 0.08564212459819379, "grad_norm": 0.646780924349035, "learning_rate": 9.935404360275904e-06, "loss": 1.5425, "step": 1119 }, { "epoch": 0.08571865911526098, "grad_norm": 1.8929376134624478, "learning_rate": 9.935226770457759e-06, "loss": 1.9168, "step": 1120 }, { "epoch": 0.08579519363232818, "grad_norm": 0.6301538708319764, "learning_rate": 9.935048938847516e-06, "loss": 1.4702, "step": 1121 }, { "epoch": 0.08587172814939538, "grad_norm": 0.6351487586092766, "learning_rate": 9.934870865456104e-06, "loss": 1.3958, "step": 1122 }, { "epoch": 0.08594826266646258, "grad_norm": 0.6159825758202696, "learning_rate": 9.934692550294461e-06, "loss": 1.2453, "step": 1123 }, { "epoch": 0.08602479718352977, "grad_norm": 0.8092977402884824, "learning_rate": 9.934513993373547e-06, "loss": 1.6281, "step": 1124 }, { "epoch": 0.08610133170059697, "grad_norm": 0.583897931556512, "learning_rate": 9.934335194704331e-06, "loss": 1.4188, "step": 1125 }, { "epoch": 0.08617786621766417, "grad_norm": 0.6492039396291244, "learning_rate": 9.934156154297799e-06, "loss": 1.6489, "step": 1126 }, { "epoch": 0.08625440073473137, "grad_norm": 0.5563121819857367, "learning_rate": 9.933976872164953e-06, "loss": 1.3613, "step": 1127 }, { "epoch": 0.08633093525179857, "grad_norm": 0.9544972372270001, "learning_rate": 9.933797348316805e-06, "loss": 1.5241, "step": 1128 }, { "epoch": 0.08640746976886576, "grad_norm": 0.5357141341388959, "learning_rate": 9.93361758276439e-06, "loss": 1.2559, "step": 1129 }, { "epoch": 0.08648400428593296, "grad_norm": 0.6285229935682782, "learning_rate": 9.933437575518752e-06, "loss": 1.5482, "step": 1130 }, { "epoch": 0.08656053880300016, "grad_norm": 0.587748254408307, "learning_rate": 9.93325732659095e-06, "loss": 1.3984, "step": 1131 }, { "epoch": 0.08663707332006736, "grad_norm": 0.7354673153455511, "learning_rate": 9.933076835992058e-06, "loss": 1.3575, "step": 1132 }, { "epoch": 0.08671360783713455, "grad_norm": 0.7197699429159818, "learning_rate": 9.932896103733169e-06, "loss": 1.873, "step": 1133 }, { "epoch": 0.08679014235420174, "grad_norm": 0.7242769825554235, "learning_rate": 9.932715129825387e-06, "loss": 1.6796, "step": 1134 }, { "epoch": 0.08686667687126894, "grad_norm": 0.5341031225393074, "learning_rate": 9.93253391427983e-06, "loss": 1.1274, "step": 1135 }, { "epoch": 0.08694321138833613, "grad_norm": 0.6437547850075904, "learning_rate": 9.932352457107632e-06, "loss": 1.3306, "step": 1136 }, { "epoch": 0.08701974590540333, "grad_norm": 0.7885517924977828, "learning_rate": 9.932170758319945e-06, "loss": 1.3788, "step": 1137 }, { "epoch": 0.08709628042247053, "grad_norm": 0.5348744396657064, "learning_rate": 9.931988817927933e-06, "loss": 1.2455, "step": 1138 }, { "epoch": 0.08717281493953773, "grad_norm": 0.6095850225878418, "learning_rate": 9.93180663594277e-06, "loss": 1.4119, "step": 1139 }, { "epoch": 0.08724934945660492, "grad_norm": 1.3152318768060929, "learning_rate": 9.931624212375657e-06, "loss": 1.4692, "step": 1140 }, { "epoch": 0.08732588397367212, "grad_norm": 0.6557306033037987, "learning_rate": 9.931441547237797e-06, "loss": 1.4689, "step": 1141 }, { "epoch": 0.08740241849073932, "grad_norm": 0.6246370231220177, "learning_rate": 9.931258640540415e-06, "loss": 1.3784, "step": 1142 }, { "epoch": 0.08747895300780652, "grad_norm": 0.7901574632752889, "learning_rate": 9.931075492294753e-06, "loss": 1.6961, "step": 1143 }, { "epoch": 0.08755548752487372, "grad_norm": 0.6312404957513544, "learning_rate": 9.930892102512059e-06, "loss": 1.451, "step": 1144 }, { "epoch": 0.08763202204194091, "grad_norm": 0.5770403407855558, "learning_rate": 9.930708471203603e-06, "loss": 1.4139, "step": 1145 }, { "epoch": 0.08770855655900811, "grad_norm": 2.725727351056535, "learning_rate": 9.930524598380667e-06, "loss": 1.6734, "step": 1146 }, { "epoch": 0.08778509107607531, "grad_norm": 0.5886109049008763, "learning_rate": 9.930340484054552e-06, "loss": 1.3193, "step": 1147 }, { "epoch": 0.0878616255931425, "grad_norm": 0.58955779587175, "learning_rate": 9.930156128236566e-06, "loss": 1.3299, "step": 1148 }, { "epoch": 0.0879381601102097, "grad_norm": 0.836024038865359, "learning_rate": 9.929971530938041e-06, "loss": 1.6003, "step": 1149 }, { "epoch": 0.0880146946272769, "grad_norm": 0.755830953620058, "learning_rate": 9.929786692170314e-06, "loss": 1.7693, "step": 1150 }, { "epoch": 0.0880912291443441, "grad_norm": 0.6662578475368441, "learning_rate": 9.929601611944746e-06, "loss": 1.7824, "step": 1151 }, { "epoch": 0.0881677636614113, "grad_norm": 0.7299348956078802, "learning_rate": 9.929416290272708e-06, "loss": 1.7546, "step": 1152 }, { "epoch": 0.0882442981784785, "grad_norm": 0.6448700536923, "learning_rate": 9.929230727165586e-06, "loss": 1.5167, "step": 1153 }, { "epoch": 0.08832083269554569, "grad_norm": 0.6375605763026958, "learning_rate": 9.929044922634783e-06, "loss": 1.5165, "step": 1154 }, { "epoch": 0.08839736721261289, "grad_norm": 0.5450800684164682, "learning_rate": 9.928858876691712e-06, "loss": 1.2263, "step": 1155 }, { "epoch": 0.08847390172968009, "grad_norm": 0.5649250152701499, "learning_rate": 9.92867258934781e-06, "loss": 1.3364, "step": 1156 }, { "epoch": 0.08855043624674729, "grad_norm": 0.510539591929836, "learning_rate": 9.928486060614516e-06, "loss": 1.0626, "step": 1157 }, { "epoch": 0.08862697076381448, "grad_norm": 0.6437177934541813, "learning_rate": 9.928299290503297e-06, "loss": 1.5343, "step": 1158 }, { "epoch": 0.08870350528088168, "grad_norm": 0.7190988122422529, "learning_rate": 9.928112279025626e-06, "loss": 1.5643, "step": 1159 }, { "epoch": 0.08878003979794888, "grad_norm": 0.6166705209910291, "learning_rate": 9.927925026192992e-06, "loss": 1.2722, "step": 1160 }, { "epoch": 0.08885657431501608, "grad_norm": 0.7436814566446178, "learning_rate": 9.927737532016903e-06, "loss": 1.4588, "step": 1161 }, { "epoch": 0.08893310883208327, "grad_norm": 0.6900166280572028, "learning_rate": 9.927549796508879e-06, "loss": 1.5806, "step": 1162 }, { "epoch": 0.08900964334915047, "grad_norm": 0.6014582834219908, "learning_rate": 9.927361819680455e-06, "loss": 1.2765, "step": 1163 }, { "epoch": 0.08908617786621767, "grad_norm": 0.506639681532466, "learning_rate": 9.927173601543178e-06, "loss": 1.1822, "step": 1164 }, { "epoch": 0.08916271238328487, "grad_norm": 0.7556686871234609, "learning_rate": 9.926985142108616e-06, "loss": 1.8213, "step": 1165 }, { "epoch": 0.08923924690035206, "grad_norm": 1.0201299816331544, "learning_rate": 9.926796441388349e-06, "loss": 1.7145, "step": 1166 }, { "epoch": 0.08931578141741926, "grad_norm": 0.6011818489909634, "learning_rate": 9.926607499393968e-06, "loss": 1.406, "step": 1167 }, { "epoch": 0.08939231593448645, "grad_norm": 0.6737485060544572, "learning_rate": 9.926418316137083e-06, "loss": 1.4198, "step": 1168 }, { "epoch": 0.08946885045155364, "grad_norm": 0.5920289256521023, "learning_rate": 9.92622889162932e-06, "loss": 1.4278, "step": 1169 }, { "epoch": 0.08954538496862084, "grad_norm": 0.6587693853488363, "learning_rate": 9.926039225882318e-06, "loss": 1.5406, "step": 1170 }, { "epoch": 0.08962191948568804, "grad_norm": 0.7125472567526306, "learning_rate": 9.925849318907728e-06, "loss": 1.8805, "step": 1171 }, { "epoch": 0.08969845400275524, "grad_norm": 0.8139369303174644, "learning_rate": 9.925659170717219e-06, "loss": 1.2929, "step": 1172 }, { "epoch": 0.08977498851982244, "grad_norm": 0.5925070926408192, "learning_rate": 9.925468781322477e-06, "loss": 1.4014, "step": 1173 }, { "epoch": 0.08985152303688963, "grad_norm": 0.6027979708918879, "learning_rate": 9.925278150735196e-06, "loss": 1.6028, "step": 1174 }, { "epoch": 0.08992805755395683, "grad_norm": 0.5735447282484196, "learning_rate": 9.925087278967093e-06, "loss": 1.3308, "step": 1175 }, { "epoch": 0.09000459207102403, "grad_norm": 0.515313628275048, "learning_rate": 9.924896166029894e-06, "loss": 1.1484, "step": 1176 }, { "epoch": 0.09008112658809123, "grad_norm": 0.5398346360382185, "learning_rate": 9.92470481193534e-06, "loss": 1.3145, "step": 1177 }, { "epoch": 0.09015766110515842, "grad_norm": 0.6509277853809792, "learning_rate": 9.924513216695189e-06, "loss": 1.5188, "step": 1178 }, { "epoch": 0.09023419562222562, "grad_norm": 0.651371561503844, "learning_rate": 9.924321380321215e-06, "loss": 1.1934, "step": 1179 }, { "epoch": 0.09031073013929282, "grad_norm": 0.6036548482070155, "learning_rate": 9.924129302825205e-06, "loss": 1.6144, "step": 1180 }, { "epoch": 0.09038726465636002, "grad_norm": 0.7206887421310554, "learning_rate": 9.923936984218957e-06, "loss": 1.5422, "step": 1181 }, { "epoch": 0.09046379917342721, "grad_norm": 0.7477142855428068, "learning_rate": 9.923744424514292e-06, "loss": 1.5435, "step": 1182 }, { "epoch": 0.09054033369049441, "grad_norm": 0.9302950322173703, "learning_rate": 9.92355162372304e-06, "loss": 1.6222, "step": 1183 }, { "epoch": 0.09061686820756161, "grad_norm": 0.743855925537777, "learning_rate": 9.923358581857048e-06, "loss": 1.753, "step": 1184 }, { "epoch": 0.09069340272462881, "grad_norm": 0.5901060391068014, "learning_rate": 9.923165298928174e-06, "loss": 1.5458, "step": 1185 }, { "epoch": 0.090769937241696, "grad_norm": 0.5137451327536121, "learning_rate": 9.922971774948297e-06, "loss": 1.1724, "step": 1186 }, { "epoch": 0.0908464717587632, "grad_norm": 0.6888468756398664, "learning_rate": 9.922778009929307e-06, "loss": 1.4149, "step": 1187 }, { "epoch": 0.0909230062758304, "grad_norm": 0.6950426752881382, "learning_rate": 9.922584003883109e-06, "loss": 1.5658, "step": 1188 }, { "epoch": 0.0909995407928976, "grad_norm": 0.5666299742545259, "learning_rate": 9.922389756821623e-06, "loss": 1.4233, "step": 1189 }, { "epoch": 0.0910760753099648, "grad_norm": 0.7489485861406814, "learning_rate": 9.922195268756784e-06, "loss": 1.6255, "step": 1190 }, { "epoch": 0.091152609827032, "grad_norm": 0.603583533910851, "learning_rate": 9.922000539700545e-06, "loss": 1.5185, "step": 1191 }, { "epoch": 0.09122914434409919, "grad_norm": 0.6137128950623923, "learning_rate": 9.921805569664866e-06, "loss": 1.3471, "step": 1192 }, { "epoch": 0.09130567886116639, "grad_norm": 0.5190079976895156, "learning_rate": 9.92161035866173e-06, "loss": 1.0865, "step": 1193 }, { "epoch": 0.09138221337823359, "grad_norm": 0.5937576162052196, "learning_rate": 9.92141490670313e-06, "loss": 1.4983, "step": 1194 }, { "epoch": 0.09145874789530078, "grad_norm": 0.9582760814200416, "learning_rate": 9.921219213801075e-06, "loss": 1.8511, "step": 1195 }, { "epoch": 0.09153528241236798, "grad_norm": 0.6543595850560079, "learning_rate": 9.92102327996759e-06, "loss": 1.6571, "step": 1196 }, { "epoch": 0.09161181692943518, "grad_norm": 1.0182584209900827, "learning_rate": 9.920827105214711e-06, "loss": 1.8377, "step": 1197 }, { "epoch": 0.09168835144650238, "grad_norm": 0.5957390611657422, "learning_rate": 9.920630689554496e-06, "loss": 1.4075, "step": 1198 }, { "epoch": 0.09176488596356958, "grad_norm": 1.9375786804354962, "learning_rate": 9.92043403299901e-06, "loss": 1.6011, "step": 1199 }, { "epoch": 0.09184142048063677, "grad_norm": 1.0357741629633275, "learning_rate": 9.920237135560337e-06, "loss": 1.4634, "step": 1200 }, { "epoch": 0.09191795499770397, "grad_norm": 0.6662930915947312, "learning_rate": 9.920039997250576e-06, "loss": 1.4688, "step": 1201 }, { "epoch": 0.09199448951477115, "grad_norm": 1.100344574855051, "learning_rate": 9.919842618081838e-06, "loss": 1.8842, "step": 1202 }, { "epoch": 0.09207102403183835, "grad_norm": 0.5439326741861918, "learning_rate": 9.919644998066251e-06, "loss": 1.3091, "step": 1203 }, { "epoch": 0.09214755854890555, "grad_norm": 0.5937531449094482, "learning_rate": 9.919447137215958e-06, "loss": 1.4076, "step": 1204 }, { "epoch": 0.09222409306597275, "grad_norm": 0.6312502850973113, "learning_rate": 9.919249035543117e-06, "loss": 1.681, "step": 1205 }, { "epoch": 0.09230062758303995, "grad_norm": 0.6133483847714536, "learning_rate": 9.919050693059899e-06, "loss": 1.478, "step": 1206 }, { "epoch": 0.09237716210010714, "grad_norm": 0.6486592941127897, "learning_rate": 9.91885210977849e-06, "loss": 1.1592, "step": 1207 }, { "epoch": 0.09245369661717434, "grad_norm": 0.5937216639779067, "learning_rate": 9.918653285711091e-06, "loss": 1.4457, "step": 1208 }, { "epoch": 0.09253023113424154, "grad_norm": 0.572505960441378, "learning_rate": 9.918454220869922e-06, "loss": 1.2739, "step": 1209 }, { "epoch": 0.09260676565130874, "grad_norm": 0.6215429941195307, "learning_rate": 9.918254915267211e-06, "loss": 1.4826, "step": 1210 }, { "epoch": 0.09268330016837593, "grad_norm": 0.8627497650437325, "learning_rate": 9.918055368915205e-06, "loss": 1.4959, "step": 1211 }, { "epoch": 0.09275983468544313, "grad_norm": 0.6236916364318618, "learning_rate": 9.917855581826163e-06, "loss": 1.406, "step": 1212 }, { "epoch": 0.09283636920251033, "grad_norm": 0.6634028535902391, "learning_rate": 9.917655554012362e-06, "loss": 1.6492, "step": 1213 }, { "epoch": 0.09291290371957753, "grad_norm": 0.6733593656839119, "learning_rate": 9.917455285486093e-06, "loss": 1.6586, "step": 1214 }, { "epoch": 0.09298943823664473, "grad_norm": 0.6265002937428896, "learning_rate": 9.91725477625966e-06, "loss": 1.3136, "step": 1215 }, { "epoch": 0.09306597275371192, "grad_norm": 0.6388865569548146, "learning_rate": 9.917054026345384e-06, "loss": 1.3762, "step": 1216 }, { "epoch": 0.09314250727077912, "grad_norm": 0.5672264579780392, "learning_rate": 9.916853035755596e-06, "loss": 1.2262, "step": 1217 }, { "epoch": 0.09321904178784632, "grad_norm": 0.6340563511405947, "learning_rate": 9.916651804502651e-06, "loss": 1.4962, "step": 1218 }, { "epoch": 0.09329557630491352, "grad_norm": 0.7843708662069072, "learning_rate": 9.91645033259891e-06, "loss": 1.5063, "step": 1219 }, { "epoch": 0.09337211082198071, "grad_norm": 0.8395858308447228, "learning_rate": 9.916248620056754e-06, "loss": 1.6718, "step": 1220 }, { "epoch": 0.09344864533904791, "grad_norm": 0.5749847726387682, "learning_rate": 9.916046666888574e-06, "loss": 1.2929, "step": 1221 }, { "epoch": 0.09352517985611511, "grad_norm": 0.6312006198352771, "learning_rate": 9.91584447310678e-06, "loss": 1.7028, "step": 1222 }, { "epoch": 0.09360171437318231, "grad_norm": 0.8826076316030786, "learning_rate": 9.915642038723795e-06, "loss": 1.4934, "step": 1223 }, { "epoch": 0.0936782488902495, "grad_norm": 0.6429943305176048, "learning_rate": 9.915439363752058e-06, "loss": 1.4647, "step": 1224 }, { "epoch": 0.0937547834073167, "grad_norm": 0.5868925815658103, "learning_rate": 9.915236448204023e-06, "loss": 1.3691, "step": 1225 }, { "epoch": 0.0938313179243839, "grad_norm": 0.6318860707932081, "learning_rate": 9.915033292092154e-06, "loss": 1.1522, "step": 1226 }, { "epoch": 0.0939078524414511, "grad_norm": 0.5066468705746675, "learning_rate": 9.914829895428937e-06, "loss": 1.1003, "step": 1227 }, { "epoch": 0.0939843869585183, "grad_norm": 0.5522004335946067, "learning_rate": 9.914626258226867e-06, "loss": 1.251, "step": 1228 }, { "epoch": 0.0940609214755855, "grad_norm": 0.6128313409058163, "learning_rate": 9.914422380498458e-06, "loss": 1.5535, "step": 1229 }, { "epoch": 0.09413745599265269, "grad_norm": 0.5716892207248564, "learning_rate": 9.914218262256237e-06, "loss": 1.3414, "step": 1230 }, { "epoch": 0.09421399050971989, "grad_norm": 0.7428032586322924, "learning_rate": 9.914013903512744e-06, "loss": 1.6686, "step": 1231 }, { "epoch": 0.09429052502678709, "grad_norm": 0.5785561429678214, "learning_rate": 9.913809304280534e-06, "loss": 1.0818, "step": 1232 }, { "epoch": 0.09436705954385428, "grad_norm": 0.5871705795138694, "learning_rate": 9.913604464572185e-06, "loss": 1.2813, "step": 1233 }, { "epoch": 0.09444359406092148, "grad_norm": 0.5954738269434018, "learning_rate": 9.913399384400273e-06, "loss": 1.4203, "step": 1234 }, { "epoch": 0.09452012857798868, "grad_norm": 0.6054107865952914, "learning_rate": 9.913194063777406e-06, "loss": 1.3768, "step": 1235 }, { "epoch": 0.09459666309505586, "grad_norm": 0.6493515153165154, "learning_rate": 9.9129885027162e-06, "loss": 1.4558, "step": 1236 }, { "epoch": 0.09467319761212306, "grad_norm": 0.849550821304108, "learning_rate": 9.91278270122928e-06, "loss": 1.8007, "step": 1237 }, { "epoch": 0.09474973212919026, "grad_norm": 0.8106552534181272, "learning_rate": 9.912576659329294e-06, "loss": 1.8589, "step": 1238 }, { "epoch": 0.09482626664625746, "grad_norm": 0.6479555678704916, "learning_rate": 9.912370377028903e-06, "loss": 1.5706, "step": 1239 }, { "epoch": 0.09490280116332465, "grad_norm": 0.5101379021467201, "learning_rate": 9.912163854340778e-06, "loss": 1.2831, "step": 1240 }, { "epoch": 0.09497933568039185, "grad_norm": 0.5418767954237864, "learning_rate": 9.911957091277614e-06, "loss": 1.2427, "step": 1241 }, { "epoch": 0.09505587019745905, "grad_norm": 0.6176370832660052, "learning_rate": 9.911750087852108e-06, "loss": 1.5524, "step": 1242 }, { "epoch": 0.09513240471452625, "grad_norm": 0.5145117662092639, "learning_rate": 9.911542844076985e-06, "loss": 1.1378, "step": 1243 }, { "epoch": 0.09520893923159345, "grad_norm": 0.6762098319778825, "learning_rate": 9.911335359964976e-06, "loss": 1.3671, "step": 1244 }, { "epoch": 0.09528547374866064, "grad_norm": 0.6227271354235063, "learning_rate": 9.911127635528828e-06, "loss": 1.5416, "step": 1245 }, { "epoch": 0.09536200826572784, "grad_norm": 0.5489582891652904, "learning_rate": 9.910919670781309e-06, "loss": 1.2637, "step": 1246 }, { "epoch": 0.09543854278279504, "grad_norm": 0.6521964939639443, "learning_rate": 9.910711465735193e-06, "loss": 1.5113, "step": 1247 }, { "epoch": 0.09551507729986224, "grad_norm": 0.6578220378660461, "learning_rate": 9.910503020403273e-06, "loss": 1.8643, "step": 1248 }, { "epoch": 0.09559161181692943, "grad_norm": 0.5726314296554736, "learning_rate": 9.910294334798359e-06, "loss": 1.1777, "step": 1249 }, { "epoch": 0.09566814633399663, "grad_norm": 0.6495303193794322, "learning_rate": 9.91008540893327e-06, "loss": 1.6435, "step": 1250 }, { "epoch": 0.09574468085106383, "grad_norm": 0.6463499264332869, "learning_rate": 9.909876242820845e-06, "loss": 1.6816, "step": 1251 }, { "epoch": 0.09582121536813103, "grad_norm": 0.7350006798185635, "learning_rate": 9.909666836473934e-06, "loss": 1.3413, "step": 1252 }, { "epoch": 0.09589774988519822, "grad_norm": 0.5830039780720724, "learning_rate": 9.909457189905406e-06, "loss": 1.4547, "step": 1253 }, { "epoch": 0.09597428440226542, "grad_norm": 0.8223986620895891, "learning_rate": 9.909247303128142e-06, "loss": 1.8792, "step": 1254 }, { "epoch": 0.09605081891933262, "grad_norm": 0.6186987643310328, "learning_rate": 9.909037176155037e-06, "loss": 1.4623, "step": 1255 }, { "epoch": 0.09612735343639982, "grad_norm": 0.5819027200243296, "learning_rate": 9.908826808999001e-06, "loss": 1.4189, "step": 1256 }, { "epoch": 0.09620388795346702, "grad_norm": 0.6091149303967759, "learning_rate": 9.908616201672962e-06, "loss": 1.5078, "step": 1257 }, { "epoch": 0.09628042247053421, "grad_norm": 0.9252703160932325, "learning_rate": 9.908405354189858e-06, "loss": 1.3688, "step": 1258 }, { "epoch": 0.09635695698760141, "grad_norm": 0.8224722374130923, "learning_rate": 9.908194266562645e-06, "loss": 1.6795, "step": 1259 }, { "epoch": 0.09643349150466861, "grad_norm": 0.5290603635766149, "learning_rate": 9.907982938804295e-06, "loss": 1.2629, "step": 1260 }, { "epoch": 0.0965100260217358, "grad_norm": 1.0848096874606379, "learning_rate": 9.907771370927789e-06, "loss": 1.611, "step": 1261 }, { "epoch": 0.096586560538803, "grad_norm": 0.6917726251339996, "learning_rate": 9.907559562946128e-06, "loss": 1.2921, "step": 1262 }, { "epoch": 0.0966630950558702, "grad_norm": 0.587381131199829, "learning_rate": 9.907347514872326e-06, "loss": 1.4717, "step": 1263 }, { "epoch": 0.0967396295729374, "grad_norm": 0.6837188151488521, "learning_rate": 9.907135226719411e-06, "loss": 1.5903, "step": 1264 }, { "epoch": 0.0968161640900046, "grad_norm": 0.6423769441044357, "learning_rate": 9.90692269850043e-06, "loss": 1.2949, "step": 1265 }, { "epoch": 0.0968926986070718, "grad_norm": 0.5476839085635586, "learning_rate": 9.906709930228436e-06, "loss": 1.1303, "step": 1266 }, { "epoch": 0.09696923312413899, "grad_norm": 0.594762044044822, "learning_rate": 9.906496921916508e-06, "loss": 1.3444, "step": 1267 }, { "epoch": 0.09704576764120619, "grad_norm": 0.6356836019117391, "learning_rate": 9.906283673577729e-06, "loss": 1.5491, "step": 1268 }, { "epoch": 0.09712230215827339, "grad_norm": 0.6890055104877891, "learning_rate": 9.906070185225204e-06, "loss": 1.6447, "step": 1269 }, { "epoch": 0.09719883667534057, "grad_norm": 0.6225273264673353, "learning_rate": 9.905856456872049e-06, "loss": 1.5237, "step": 1270 }, { "epoch": 0.09727537119240777, "grad_norm": 0.654631527011944, "learning_rate": 9.905642488531398e-06, "loss": 1.5166, "step": 1271 }, { "epoch": 0.09735190570947497, "grad_norm": 0.7528083636767023, "learning_rate": 9.905428280216397e-06, "loss": 1.8726, "step": 1272 }, { "epoch": 0.09742844022654217, "grad_norm": 0.8111736567609211, "learning_rate": 9.905213831940208e-06, "loss": 1.7168, "step": 1273 }, { "epoch": 0.09750497474360936, "grad_norm": 0.5929579895638721, "learning_rate": 9.904999143716005e-06, "loss": 1.3529, "step": 1274 }, { "epoch": 0.09758150926067656, "grad_norm": 0.8044353203617033, "learning_rate": 9.904784215556981e-06, "loss": 1.6024, "step": 1275 }, { "epoch": 0.09765804377774376, "grad_norm": 0.5827484301947438, "learning_rate": 9.904569047476344e-06, "loss": 1.49, "step": 1276 }, { "epoch": 0.09773457829481096, "grad_norm": 0.6191178648579287, "learning_rate": 9.904353639487311e-06, "loss": 1.412, "step": 1277 }, { "epoch": 0.09781111281187815, "grad_norm": 0.5952836456782931, "learning_rate": 9.90413799160312e-06, "loss": 1.4457, "step": 1278 }, { "epoch": 0.09788764732894535, "grad_norm": 0.7776899211307123, "learning_rate": 9.903922103837017e-06, "loss": 1.7504, "step": 1279 }, { "epoch": 0.09796418184601255, "grad_norm": 0.6349638917786288, "learning_rate": 9.90370597620227e-06, "loss": 1.5692, "step": 1280 }, { "epoch": 0.09804071636307975, "grad_norm": 0.9083914815476082, "learning_rate": 9.90348960871216e-06, "loss": 1.4584, "step": 1281 }, { "epoch": 0.09811725088014694, "grad_norm": 0.5021661456784867, "learning_rate": 9.903273001379976e-06, "loss": 1.1764, "step": 1282 }, { "epoch": 0.09819378539721414, "grad_norm": 0.7423066274720433, "learning_rate": 9.903056154219033e-06, "loss": 1.2198, "step": 1283 }, { "epoch": 0.09827031991428134, "grad_norm": 0.5034765395253352, "learning_rate": 9.902839067242652e-06, "loss": 1.0716, "step": 1284 }, { "epoch": 0.09834685443134854, "grad_norm": 0.678662000479725, "learning_rate": 9.90262174046417e-06, "loss": 1.4778, "step": 1285 }, { "epoch": 0.09842338894841574, "grad_norm": 0.6524786802280258, "learning_rate": 9.902404173896941e-06, "loss": 1.5571, "step": 1286 }, { "epoch": 0.09849992346548293, "grad_norm": 0.6144132880121096, "learning_rate": 9.902186367554335e-06, "loss": 1.4093, "step": 1287 }, { "epoch": 0.09857645798255013, "grad_norm": 0.5617918481510995, "learning_rate": 9.901968321449734e-06, "loss": 1.3332, "step": 1288 }, { "epoch": 0.09865299249961733, "grad_norm": 0.6232101791259655, "learning_rate": 9.901750035596533e-06, "loss": 1.3804, "step": 1289 }, { "epoch": 0.09872952701668453, "grad_norm": 1.4665354166626932, "learning_rate": 9.901531510008146e-06, "loss": 1.5583, "step": 1290 }, { "epoch": 0.09880606153375172, "grad_norm": 0.6914360391438907, "learning_rate": 9.901312744698e-06, "loss": 1.3336, "step": 1291 }, { "epoch": 0.09888259605081892, "grad_norm": 0.5585380189260882, "learning_rate": 9.901093739679536e-06, "loss": 1.2054, "step": 1292 }, { "epoch": 0.09895913056788612, "grad_norm": 0.535572834142885, "learning_rate": 9.900874494966211e-06, "loss": 1.1809, "step": 1293 }, { "epoch": 0.09903566508495332, "grad_norm": 0.607674124181319, "learning_rate": 9.900655010571495e-06, "loss": 1.4608, "step": 1294 }, { "epoch": 0.09911219960202051, "grad_norm": 0.6466968907364921, "learning_rate": 9.900435286508874e-06, "loss": 1.385, "step": 1295 }, { "epoch": 0.09918873411908771, "grad_norm": 0.6812982757963199, "learning_rate": 9.90021532279185e-06, "loss": 1.4298, "step": 1296 }, { "epoch": 0.09926526863615491, "grad_norm": 0.6186676487862617, "learning_rate": 9.899995119433939e-06, "loss": 1.3745, "step": 1297 }, { "epoch": 0.09934180315322211, "grad_norm": 0.8684824621455206, "learning_rate": 9.899774676448665e-06, "loss": 1.7378, "step": 1298 }, { "epoch": 0.0994183376702893, "grad_norm": 0.5958288773357319, "learning_rate": 9.89955399384958e-06, "loss": 1.3015, "step": 1299 }, { "epoch": 0.0994948721873565, "grad_norm": 0.6378264125271208, "learning_rate": 9.89933307165024e-06, "loss": 1.3811, "step": 1300 }, { "epoch": 0.0995714067044237, "grad_norm": 0.6145163114281547, "learning_rate": 9.899111909864218e-06, "loss": 1.3716, "step": 1301 }, { "epoch": 0.0996479412214909, "grad_norm": 0.7177599190825809, "learning_rate": 9.898890508505104e-06, "loss": 1.7793, "step": 1302 }, { "epoch": 0.0997244757385581, "grad_norm": 1.7416438675436723, "learning_rate": 9.898668867586501e-06, "loss": 1.6464, "step": 1303 }, { "epoch": 0.09980101025562528, "grad_norm": 0.6650631159649323, "learning_rate": 9.89844698712203e-06, "loss": 1.5936, "step": 1304 }, { "epoch": 0.09987754477269248, "grad_norm": 0.5737171655776327, "learning_rate": 9.898224867125321e-06, "loss": 1.3462, "step": 1305 }, { "epoch": 0.09995407928975968, "grad_norm": 0.7105235620903199, "learning_rate": 9.898002507610023e-06, "loss": 1.5059, "step": 1306 }, { "epoch": 0.10003061380682687, "grad_norm": 0.612288783225743, "learning_rate": 9.897779908589797e-06, "loss": 1.4534, "step": 1307 }, { "epoch": 0.10010714832389407, "grad_norm": 0.5099859540733034, "learning_rate": 9.897557070078323e-06, "loss": 1.1676, "step": 1308 }, { "epoch": 0.10018368284096127, "grad_norm": 0.6228311738220335, "learning_rate": 9.89733399208929e-06, "loss": 1.338, "step": 1309 }, { "epoch": 0.10026021735802847, "grad_norm": 0.6134532338529213, "learning_rate": 9.897110674636406e-06, "loss": 1.3984, "step": 1310 }, { "epoch": 0.10033675187509566, "grad_norm": 0.5877604395205577, "learning_rate": 9.896887117733393e-06, "loss": 1.3399, "step": 1311 }, { "epoch": 0.10041328639216286, "grad_norm": 0.8283432649092556, "learning_rate": 9.896663321393984e-06, "loss": 1.7535, "step": 1312 }, { "epoch": 0.10048982090923006, "grad_norm": 0.6287269081025334, "learning_rate": 9.896439285631933e-06, "loss": 1.7027, "step": 1313 }, { "epoch": 0.10056635542629726, "grad_norm": 0.7205144445167886, "learning_rate": 9.896215010461004e-06, "loss": 1.6082, "step": 1314 }, { "epoch": 0.10064288994336446, "grad_norm": 1.23304172471821, "learning_rate": 9.895990495894979e-06, "loss": 1.5512, "step": 1315 }, { "epoch": 0.10071942446043165, "grad_norm": 0.5394667975321824, "learning_rate": 9.895765741947648e-06, "loss": 1.1169, "step": 1316 }, { "epoch": 0.10079595897749885, "grad_norm": 0.6551944346555991, "learning_rate": 9.895540748632825e-06, "loss": 1.2805, "step": 1317 }, { "epoch": 0.10087249349456605, "grad_norm": 0.5438146024761082, "learning_rate": 9.895315515964334e-06, "loss": 1.1608, "step": 1318 }, { "epoch": 0.10094902801163325, "grad_norm": 1.962401086377171, "learning_rate": 9.895090043956011e-06, "loss": 1.6801, "step": 1319 }, { "epoch": 0.10102556252870044, "grad_norm": 0.5957916641472351, "learning_rate": 9.894864332621713e-06, "loss": 1.4095, "step": 1320 }, { "epoch": 0.10110209704576764, "grad_norm": 0.6592038717313577, "learning_rate": 9.894638381975305e-06, "loss": 1.6505, "step": 1321 }, { "epoch": 0.10117863156283484, "grad_norm": 0.77472031536927, "learning_rate": 9.894412192030674e-06, "loss": 1.4497, "step": 1322 }, { "epoch": 0.10125516607990204, "grad_norm": 0.6304157446622333, "learning_rate": 9.894185762801716e-06, "loss": 1.6355, "step": 1323 }, { "epoch": 0.10133170059696923, "grad_norm": 0.5519252138188507, "learning_rate": 9.893959094302343e-06, "loss": 1.1563, "step": 1324 }, { "epoch": 0.10140823511403643, "grad_norm": 0.5512260092191017, "learning_rate": 9.893732186546481e-06, "loss": 1.2587, "step": 1325 }, { "epoch": 0.10148476963110363, "grad_norm": 0.5480379998232404, "learning_rate": 9.893505039548075e-06, "loss": 1.3549, "step": 1326 }, { "epoch": 0.10156130414817083, "grad_norm": 0.6257483873272291, "learning_rate": 9.89327765332108e-06, "loss": 1.3838, "step": 1327 }, { "epoch": 0.10163783866523803, "grad_norm": 0.4924535684300365, "learning_rate": 9.893050027879468e-06, "loss": 1.0316, "step": 1328 }, { "epoch": 0.10171437318230522, "grad_norm": 0.609401076841166, "learning_rate": 9.892822163237223e-06, "loss": 1.5703, "step": 1329 }, { "epoch": 0.10179090769937242, "grad_norm": 0.5866920516171901, "learning_rate": 9.892594059408347e-06, "loss": 1.4589, "step": 1330 }, { "epoch": 0.10186744221643962, "grad_norm": 0.6924625801279216, "learning_rate": 9.892365716406855e-06, "loss": 1.6477, "step": 1331 }, { "epoch": 0.10194397673350682, "grad_norm": 0.627573073405041, "learning_rate": 9.892137134246778e-06, "loss": 1.5047, "step": 1332 }, { "epoch": 0.10202051125057401, "grad_norm": 1.2555863144742225, "learning_rate": 9.891908312942161e-06, "loss": 1.2745, "step": 1333 }, { "epoch": 0.10209704576764121, "grad_norm": 0.5344798564506086, "learning_rate": 9.891679252507062e-06, "loss": 1.0673, "step": 1334 }, { "epoch": 0.10217358028470841, "grad_norm": 0.7079173970489147, "learning_rate": 9.891449952955555e-06, "loss": 1.5833, "step": 1335 }, { "epoch": 0.10225011480177561, "grad_norm": 0.638421906791813, "learning_rate": 9.89122041430173e-06, "loss": 1.5629, "step": 1336 }, { "epoch": 0.1023266493188428, "grad_norm": 0.6097572581583378, "learning_rate": 9.890990636559692e-06, "loss": 1.4443, "step": 1337 }, { "epoch": 0.10240318383590999, "grad_norm": 0.6702639371692175, "learning_rate": 9.890760619743556e-06, "loss": 1.5283, "step": 1338 }, { "epoch": 0.10247971835297719, "grad_norm": 0.5573601073724599, "learning_rate": 9.890530363867456e-06, "loss": 1.2861, "step": 1339 }, { "epoch": 0.10255625287004438, "grad_norm": 0.6178624439760738, "learning_rate": 9.890299868945539e-06, "loss": 1.2503, "step": 1340 }, { "epoch": 0.10263278738711158, "grad_norm": 0.6244774750816602, "learning_rate": 9.89006913499197e-06, "loss": 1.4724, "step": 1341 }, { "epoch": 0.10270932190417878, "grad_norm": 0.7844414013833904, "learning_rate": 9.889838162020925e-06, "loss": 1.4004, "step": 1342 }, { "epoch": 0.10278585642124598, "grad_norm": 0.6338966000116875, "learning_rate": 9.889606950046593e-06, "loss": 1.5475, "step": 1343 }, { "epoch": 0.10286239093831318, "grad_norm": 0.5996771441330753, "learning_rate": 9.889375499083182e-06, "loss": 1.2248, "step": 1344 }, { "epoch": 0.10293892545538037, "grad_norm": 0.5418600831375955, "learning_rate": 9.889143809144917e-06, "loss": 1.0851, "step": 1345 }, { "epoch": 0.10301545997244757, "grad_norm": 0.7111919156358758, "learning_rate": 9.888911880246025e-06, "loss": 1.5551, "step": 1346 }, { "epoch": 0.10309199448951477, "grad_norm": 0.5936554145370959, "learning_rate": 9.888679712400765e-06, "loss": 1.4665, "step": 1347 }, { "epoch": 0.10316852900658197, "grad_norm": 0.9719197976001102, "learning_rate": 9.888447305623396e-06, "loss": 1.7177, "step": 1348 }, { "epoch": 0.10324506352364916, "grad_norm": 0.6752004733526072, "learning_rate": 9.888214659928204e-06, "loss": 1.5381, "step": 1349 }, { "epoch": 0.10332159804071636, "grad_norm": 0.6100361779837382, "learning_rate": 9.887981775329476e-06, "loss": 1.582, "step": 1350 }, { "epoch": 0.10339813255778356, "grad_norm": 0.7202571982126982, "learning_rate": 9.887748651841526e-06, "loss": 1.3316, "step": 1351 }, { "epoch": 0.10347466707485076, "grad_norm": 0.7562426571316024, "learning_rate": 9.887515289478677e-06, "loss": 1.5457, "step": 1352 }, { "epoch": 0.10355120159191795, "grad_norm": 0.5317341057219103, "learning_rate": 9.887281688255268e-06, "loss": 1.1172, "step": 1353 }, { "epoch": 0.10362773610898515, "grad_norm": 0.7219552746593482, "learning_rate": 9.88704784818565e-06, "loss": 1.3898, "step": 1354 }, { "epoch": 0.10370427062605235, "grad_norm": 0.6175955097841548, "learning_rate": 9.886813769284193e-06, "loss": 1.287, "step": 1355 }, { "epoch": 0.10378080514311955, "grad_norm": 0.6595225474392614, "learning_rate": 9.886579451565277e-06, "loss": 1.5597, "step": 1356 }, { "epoch": 0.10385733966018675, "grad_norm": 0.6541092793597979, "learning_rate": 9.886344895043304e-06, "loss": 1.5676, "step": 1357 }, { "epoch": 0.10393387417725394, "grad_norm": 0.6382805073676281, "learning_rate": 9.88611009973268e-06, "loss": 1.5208, "step": 1358 }, { "epoch": 0.10401040869432114, "grad_norm": 0.7367918627978244, "learning_rate": 9.885875065647835e-06, "loss": 1.7307, "step": 1359 }, { "epoch": 0.10408694321138834, "grad_norm": 0.6087016204495476, "learning_rate": 9.88563979280321e-06, "loss": 1.3312, "step": 1360 }, { "epoch": 0.10416347772845554, "grad_norm": 0.7085479211792726, "learning_rate": 9.88540428121326e-06, "loss": 1.7779, "step": 1361 }, { "epoch": 0.10424001224552273, "grad_norm": 0.7559515733633371, "learning_rate": 9.885168530892456e-06, "loss": 1.7082, "step": 1362 }, { "epoch": 0.10431654676258993, "grad_norm": 0.5072026153804172, "learning_rate": 9.88493254185528e-06, "loss": 1.2752, "step": 1363 }, { "epoch": 0.10439308127965713, "grad_norm": 0.6673744458454959, "learning_rate": 9.884696314116239e-06, "loss": 1.6639, "step": 1364 }, { "epoch": 0.10446961579672433, "grad_norm": 0.6696154482922574, "learning_rate": 9.884459847689841e-06, "loss": 1.9056, "step": 1365 }, { "epoch": 0.10454615031379152, "grad_norm": 0.5759705869080833, "learning_rate": 9.88422314259062e-06, "loss": 1.3538, "step": 1366 }, { "epoch": 0.10462268483085872, "grad_norm": 0.6564923139215048, "learning_rate": 9.883986198833114e-06, "loss": 1.5535, "step": 1367 }, { "epoch": 0.10469921934792592, "grad_norm": 0.5064713069496345, "learning_rate": 9.883749016431888e-06, "loss": 1.2375, "step": 1368 }, { "epoch": 0.10477575386499312, "grad_norm": 0.6165993693614205, "learning_rate": 9.88351159540151e-06, "loss": 1.3406, "step": 1369 }, { "epoch": 0.10485228838206032, "grad_norm": 1.056970338948858, "learning_rate": 9.883273935756572e-06, "loss": 1.3433, "step": 1370 }, { "epoch": 0.1049288228991275, "grad_norm": 0.6318741021284208, "learning_rate": 9.883036037511674e-06, "loss": 1.4759, "step": 1371 }, { "epoch": 0.1050053574161947, "grad_norm": 1.4930338733915487, "learning_rate": 9.882797900681434e-06, "loss": 1.6733, "step": 1372 }, { "epoch": 0.1050818919332619, "grad_norm": 0.6387009000415549, "learning_rate": 9.882559525280484e-06, "loss": 1.3644, "step": 1373 }, { "epoch": 0.10515842645032909, "grad_norm": 0.7548189471989348, "learning_rate": 9.882320911323468e-06, "loss": 1.9001, "step": 1374 }, { "epoch": 0.10523496096739629, "grad_norm": 0.6935637250551644, "learning_rate": 9.882082058825051e-06, "loss": 1.2829, "step": 1375 }, { "epoch": 0.10531149548446349, "grad_norm": 0.5183146110234294, "learning_rate": 9.881842967799907e-06, "loss": 1.1201, "step": 1376 }, { "epoch": 0.10538803000153069, "grad_norm": 0.6214080272579884, "learning_rate": 9.881603638262726e-06, "loss": 1.3189, "step": 1377 }, { "epoch": 0.10546456451859788, "grad_norm": 0.6448969454082847, "learning_rate": 9.881364070228215e-06, "loss": 1.5686, "step": 1378 }, { "epoch": 0.10554109903566508, "grad_norm": 0.5992424129395425, "learning_rate": 9.881124263711092e-06, "loss": 1.3376, "step": 1379 }, { "epoch": 0.10561763355273228, "grad_norm": 0.7450186939625929, "learning_rate": 9.880884218726091e-06, "loss": 1.329, "step": 1380 }, { "epoch": 0.10569416806979948, "grad_norm": 0.6461924627303076, "learning_rate": 9.880643935287962e-06, "loss": 1.3619, "step": 1381 }, { "epoch": 0.10577070258686667, "grad_norm": 0.5848858635675818, "learning_rate": 9.88040341341147e-06, "loss": 1.4029, "step": 1382 }, { "epoch": 0.10584723710393387, "grad_norm": 0.6670632248839098, "learning_rate": 9.88016265311139e-06, "loss": 1.4442, "step": 1383 }, { "epoch": 0.10592377162100107, "grad_norm": 0.8913095185666363, "learning_rate": 9.87992165440252e-06, "loss": 1.8457, "step": 1384 }, { "epoch": 0.10600030613806827, "grad_norm": 0.9753188788011311, "learning_rate": 9.879680417299662e-06, "loss": 1.2731, "step": 1385 }, { "epoch": 0.10607684065513547, "grad_norm": 0.6736483296418627, "learning_rate": 9.879438941817641e-06, "loss": 1.4388, "step": 1386 }, { "epoch": 0.10615337517220266, "grad_norm": 0.6457710430065126, "learning_rate": 9.879197227971296e-06, "loss": 1.3585, "step": 1387 }, { "epoch": 0.10622990968926986, "grad_norm": 0.6136005016181761, "learning_rate": 9.878955275775478e-06, "loss": 1.4071, "step": 1388 }, { "epoch": 0.10630644420633706, "grad_norm": 0.6868277570279216, "learning_rate": 9.878713085245048e-06, "loss": 1.6387, "step": 1389 }, { "epoch": 0.10638297872340426, "grad_norm": 0.8975128882749733, "learning_rate": 9.878470656394892e-06, "loss": 1.5107, "step": 1390 }, { "epoch": 0.10645951324047145, "grad_norm": 0.6375689662557605, "learning_rate": 9.878227989239906e-06, "loss": 1.0588, "step": 1391 }, { "epoch": 0.10653604775753865, "grad_norm": 0.8198300474977291, "learning_rate": 9.877985083795e-06, "loss": 1.8552, "step": 1392 }, { "epoch": 0.10661258227460585, "grad_norm": 3.906659727004178, "learning_rate": 9.877741940075094e-06, "loss": 2.1207, "step": 1393 }, { "epoch": 0.10668911679167305, "grad_norm": 0.6831620078407595, "learning_rate": 9.877498558095133e-06, "loss": 1.5391, "step": 1394 }, { "epoch": 0.10676565130874024, "grad_norm": 0.6117235548361343, "learning_rate": 9.877254937870069e-06, "loss": 1.5424, "step": 1395 }, { "epoch": 0.10684218582580744, "grad_norm": 0.681006635116301, "learning_rate": 9.87701107941487e-06, "loss": 1.6664, "step": 1396 }, { "epoch": 0.10691872034287464, "grad_norm": 0.5909856849752584, "learning_rate": 9.87676698274452e-06, "loss": 1.4002, "step": 1397 }, { "epoch": 0.10699525485994184, "grad_norm": 0.7241271299911893, "learning_rate": 9.876522647874018e-06, "loss": 1.4564, "step": 1398 }, { "epoch": 0.10707178937700904, "grad_norm": 0.5531784089734553, "learning_rate": 9.876278074818375e-06, "loss": 1.238, "step": 1399 }, { "epoch": 0.10714832389407623, "grad_norm": 0.5495630751711033, "learning_rate": 9.87603326359262e-06, "loss": 1.2906, "step": 1400 }, { "epoch": 0.10722485841114343, "grad_norm": 0.7029898806775431, "learning_rate": 9.875788214211795e-06, "loss": 1.6567, "step": 1401 }, { "epoch": 0.10730139292821063, "grad_norm": 0.591110282422075, "learning_rate": 9.875542926690955e-06, "loss": 1.4252, "step": 1402 }, { "epoch": 0.10737792744527783, "grad_norm": 0.701216614117421, "learning_rate": 9.875297401045171e-06, "loss": 1.8121, "step": 1403 }, { "epoch": 0.10745446196234502, "grad_norm": 0.6058827631887429, "learning_rate": 9.875051637289531e-06, "loss": 1.3956, "step": 1404 }, { "epoch": 0.10753099647941221, "grad_norm": 0.5325479277132245, "learning_rate": 9.874805635439132e-06, "loss": 1.3189, "step": 1405 }, { "epoch": 0.1076075309964794, "grad_norm": 0.6548233715753394, "learning_rate": 9.874559395509095e-06, "loss": 1.6483, "step": 1406 }, { "epoch": 0.1076840655135466, "grad_norm": 0.6963892018876171, "learning_rate": 9.874312917514544e-06, "loss": 1.622, "step": 1407 }, { "epoch": 0.1077606000306138, "grad_norm": 0.5346779936055962, "learning_rate": 9.874066201470625e-06, "loss": 1.1564, "step": 1408 }, { "epoch": 0.107837134547681, "grad_norm": 0.828872419259968, "learning_rate": 9.873819247392498e-06, "loss": 1.5868, "step": 1409 }, { "epoch": 0.1079136690647482, "grad_norm": 0.8422388411585074, "learning_rate": 9.873572055295336e-06, "loss": 1.5497, "step": 1410 }, { "epoch": 0.1079902035818154, "grad_norm": 0.5832741168712474, "learning_rate": 9.873324625194325e-06, "loss": 1.2352, "step": 1411 }, { "epoch": 0.10806673809888259, "grad_norm": 0.6066472186302868, "learning_rate": 9.873076957104672e-06, "loss": 1.2578, "step": 1412 }, { "epoch": 0.10814327261594979, "grad_norm": 0.6287557718953043, "learning_rate": 9.872829051041593e-06, "loss": 1.5288, "step": 1413 }, { "epoch": 0.10821980713301699, "grad_norm": 0.7847933401171231, "learning_rate": 9.872580907020318e-06, "loss": 1.3832, "step": 1414 }, { "epoch": 0.10829634165008419, "grad_norm": 0.699088706119871, "learning_rate": 9.872332525056098e-06, "loss": 1.7481, "step": 1415 }, { "epoch": 0.10837287616715138, "grad_norm": 0.6728496958476103, "learning_rate": 9.87208390516419e-06, "loss": 1.7694, "step": 1416 }, { "epoch": 0.10844941068421858, "grad_norm": 0.604759082434099, "learning_rate": 9.87183504735987e-06, "loss": 1.5655, "step": 1417 }, { "epoch": 0.10852594520128578, "grad_norm": 0.6623019145648811, "learning_rate": 9.871585951658431e-06, "loss": 1.5367, "step": 1418 }, { "epoch": 0.10860247971835298, "grad_norm": 0.7637516408507501, "learning_rate": 9.871336618075176e-06, "loss": 1.5478, "step": 1419 }, { "epoch": 0.10867901423542017, "grad_norm": 0.5974419918529007, "learning_rate": 9.871087046625429e-06, "loss": 1.3671, "step": 1420 }, { "epoch": 0.10875554875248737, "grad_norm": 0.5742574535960733, "learning_rate": 9.87083723732452e-06, "loss": 1.3715, "step": 1421 }, { "epoch": 0.10883208326955457, "grad_norm": 0.5710591297821227, "learning_rate": 9.8705871901878e-06, "loss": 1.1896, "step": 1422 }, { "epoch": 0.10890861778662177, "grad_norm": 0.705794066780892, "learning_rate": 9.870336905230632e-06, "loss": 1.8675, "step": 1423 }, { "epoch": 0.10898515230368896, "grad_norm": 0.7309622399137209, "learning_rate": 9.870086382468395e-06, "loss": 1.6109, "step": 1424 }, { "epoch": 0.10906168682075616, "grad_norm": 0.5705077456225466, "learning_rate": 9.86983562191648e-06, "loss": 1.3304, "step": 1425 }, { "epoch": 0.10913822133782336, "grad_norm": 0.5412907439440369, "learning_rate": 9.869584623590297e-06, "loss": 1.2162, "step": 1426 }, { "epoch": 0.10921475585489056, "grad_norm": 0.5508375402730766, "learning_rate": 9.869333387505265e-06, "loss": 1.2251, "step": 1427 }, { "epoch": 0.10929129037195776, "grad_norm": 0.6011288081861202, "learning_rate": 9.869081913676825e-06, "loss": 1.4305, "step": 1428 }, { "epoch": 0.10936782488902495, "grad_norm": 0.7108970919216423, "learning_rate": 9.868830202120425e-06, "loss": 1.5696, "step": 1429 }, { "epoch": 0.10944435940609215, "grad_norm": 0.9184361124795136, "learning_rate": 9.86857825285153e-06, "loss": 1.8413, "step": 1430 }, { "epoch": 0.10952089392315935, "grad_norm": 0.5405635668698171, "learning_rate": 9.868326065885625e-06, "loss": 1.1466, "step": 1431 }, { "epoch": 0.10959742844022655, "grad_norm": 1.6640416614296176, "learning_rate": 9.868073641238201e-06, "loss": 1.5942, "step": 1432 }, { "epoch": 0.10967396295729374, "grad_norm": 0.7519116641553926, "learning_rate": 9.86782097892477e-06, "loss": 1.6951, "step": 1433 }, { "epoch": 0.10975049747436094, "grad_norm": 0.5238855507717673, "learning_rate": 9.867568078960854e-06, "loss": 1.0138, "step": 1434 }, { "epoch": 0.10982703199142814, "grad_norm": 0.7785020891606287, "learning_rate": 9.867314941361993e-06, "loss": 1.81, "step": 1435 }, { "epoch": 0.10990356650849534, "grad_norm": 0.5692828364655482, "learning_rate": 9.867061566143745e-06, "loss": 1.371, "step": 1436 }, { "epoch": 0.10998010102556253, "grad_norm": 0.5524615060669832, "learning_rate": 9.866807953321668e-06, "loss": 1.2157, "step": 1437 }, { "epoch": 0.11005663554262973, "grad_norm": 1.2601152655850285, "learning_rate": 9.866554102911356e-06, "loss": 1.5858, "step": 1438 }, { "epoch": 0.11013317005969692, "grad_norm": 0.5966545089302574, "learning_rate": 9.866300014928399e-06, "loss": 1.382, "step": 1439 }, { "epoch": 0.11020970457676411, "grad_norm": 0.5640833801637868, "learning_rate": 9.866045689388411e-06, "loss": 1.1556, "step": 1440 }, { "epoch": 0.11028623909383131, "grad_norm": 0.6376419897337215, "learning_rate": 9.865791126307019e-06, "loss": 1.533, "step": 1441 }, { "epoch": 0.11036277361089851, "grad_norm": 0.5657568365343411, "learning_rate": 9.865536325699863e-06, "loss": 1.2913, "step": 1442 }, { "epoch": 0.11043930812796571, "grad_norm": 0.6291374827827931, "learning_rate": 9.8652812875826e-06, "loss": 1.4888, "step": 1443 }, { "epoch": 0.1105158426450329, "grad_norm": 0.8257814129955994, "learning_rate": 9.865026011970899e-06, "loss": 2.0336, "step": 1444 }, { "epoch": 0.1105923771621001, "grad_norm": 0.5812366064551528, "learning_rate": 9.864770498880446e-06, "loss": 1.374, "step": 1445 }, { "epoch": 0.1106689116791673, "grad_norm": 0.6612302445672437, "learning_rate": 9.864514748326941e-06, "loss": 1.6041, "step": 1446 }, { "epoch": 0.1107454461962345, "grad_norm": 0.6432504129055537, "learning_rate": 9.864258760326096e-06, "loss": 1.4847, "step": 1447 }, { "epoch": 0.1108219807133017, "grad_norm": 0.6224917480427001, "learning_rate": 9.86400253489364e-06, "loss": 1.4751, "step": 1448 }, { "epoch": 0.1108985152303689, "grad_norm": 0.58837829577906, "learning_rate": 9.86374607204532e-06, "loss": 1.4149, "step": 1449 }, { "epoch": 0.11097504974743609, "grad_norm": 0.6248268397500575, "learning_rate": 9.863489371796889e-06, "loss": 1.5487, "step": 1450 }, { "epoch": 0.11105158426450329, "grad_norm": 0.5672189828844068, "learning_rate": 9.863232434164122e-06, "loss": 1.2953, "step": 1451 }, { "epoch": 0.11112811878157049, "grad_norm": 0.70328208830691, "learning_rate": 9.862975259162805e-06, "loss": 1.7067, "step": 1452 }, { "epoch": 0.11120465329863768, "grad_norm": 0.6086848104658051, "learning_rate": 9.862717846808739e-06, "loss": 1.6785, "step": 1453 }, { "epoch": 0.11128118781570488, "grad_norm": 0.7327106563751401, "learning_rate": 9.862460197117743e-06, "loss": 1.5872, "step": 1454 }, { "epoch": 0.11135772233277208, "grad_norm": 0.5564506076679293, "learning_rate": 9.862202310105646e-06, "loss": 1.4203, "step": 1455 }, { "epoch": 0.11143425684983928, "grad_norm": 0.5352858203487485, "learning_rate": 9.86194418578829e-06, "loss": 1.194, "step": 1456 }, { "epoch": 0.11151079136690648, "grad_norm": 0.5810716873658701, "learning_rate": 9.861685824181541e-06, "loss": 1.2806, "step": 1457 }, { "epoch": 0.11158732588397367, "grad_norm": 0.6705527306845341, "learning_rate": 9.861427225301271e-06, "loss": 1.5913, "step": 1458 }, { "epoch": 0.11166386040104087, "grad_norm": 2.607754370291728, "learning_rate": 9.861168389163367e-06, "loss": 1.5657, "step": 1459 }, { "epoch": 0.11174039491810807, "grad_norm": 0.6823418195002388, "learning_rate": 9.860909315783736e-06, "loss": 1.6499, "step": 1460 }, { "epoch": 0.11181692943517527, "grad_norm": 0.6573453046487573, "learning_rate": 9.860650005178294e-06, "loss": 1.573, "step": 1461 }, { "epoch": 0.11189346395224246, "grad_norm": 0.7246981473551044, "learning_rate": 9.860390457362976e-06, "loss": 1.5975, "step": 1462 }, { "epoch": 0.11196999846930966, "grad_norm": 0.607661112897013, "learning_rate": 9.860130672353724e-06, "loss": 1.3662, "step": 1463 }, { "epoch": 0.11204653298637686, "grad_norm": 0.636056646376562, "learning_rate": 9.859870650166508e-06, "loss": 1.6152, "step": 1464 }, { "epoch": 0.11212306750344406, "grad_norm": 0.684981291610652, "learning_rate": 9.859610390817299e-06, "loss": 1.6972, "step": 1465 }, { "epoch": 0.11219960202051125, "grad_norm": 1.7518202604042317, "learning_rate": 9.85934989432209e-06, "loss": 2.0819, "step": 1466 }, { "epoch": 0.11227613653757845, "grad_norm": 1.3631881352965758, "learning_rate": 9.859089160696887e-06, "loss": 1.4327, "step": 1467 }, { "epoch": 0.11235267105464565, "grad_norm": 0.5431798821975453, "learning_rate": 9.858828189957707e-06, "loss": 1.2312, "step": 1468 }, { "epoch": 0.11242920557171285, "grad_norm": 0.5958266535814777, "learning_rate": 9.858566982120588e-06, "loss": 1.469, "step": 1469 }, { "epoch": 0.11250574008878005, "grad_norm": 0.5848567719749532, "learning_rate": 9.85830553720158e-06, "loss": 1.4237, "step": 1470 }, { "epoch": 0.11258227460584724, "grad_norm": 0.62811262043521, "learning_rate": 9.858043855216745e-06, "loss": 1.2855, "step": 1471 }, { "epoch": 0.11265880912291444, "grad_norm": 0.8066920233199664, "learning_rate": 9.857781936182162e-06, "loss": 1.49, "step": 1472 }, { "epoch": 0.11273534363998162, "grad_norm": 0.7111432281940953, "learning_rate": 9.857519780113924e-06, "loss": 1.6312, "step": 1473 }, { "epoch": 0.11281187815704882, "grad_norm": 0.7445866433505202, "learning_rate": 9.85725738702814e-06, "loss": 1.6313, "step": 1474 }, { "epoch": 0.11288841267411602, "grad_norm": 0.5639161198105391, "learning_rate": 9.856994756940931e-06, "loss": 1.2152, "step": 1475 }, { "epoch": 0.11296494719118322, "grad_norm": 0.6927035811853912, "learning_rate": 9.856731889868433e-06, "loss": 1.772, "step": 1476 }, { "epoch": 0.11304148170825042, "grad_norm": 0.8867455597674815, "learning_rate": 9.856468785826797e-06, "loss": 1.317, "step": 1477 }, { "epoch": 0.11311801622531761, "grad_norm": 0.6383954578964102, "learning_rate": 9.856205444832193e-06, "loss": 1.4771, "step": 1478 }, { "epoch": 0.11319455074238481, "grad_norm": 0.7109669703670516, "learning_rate": 9.855941866900798e-06, "loss": 1.8237, "step": 1479 }, { "epoch": 0.11327108525945201, "grad_norm": 0.7300806414623805, "learning_rate": 9.855678052048807e-06, "loss": 1.4526, "step": 1480 }, { "epoch": 0.1133476197765192, "grad_norm": 0.7822528400640982, "learning_rate": 9.855414000292428e-06, "loss": 1.558, "step": 1481 }, { "epoch": 0.1134241542935864, "grad_norm": 0.5880500315452337, "learning_rate": 9.85514971164789e-06, "loss": 1.4351, "step": 1482 }, { "epoch": 0.1135006888106536, "grad_norm": 0.8531207120712916, "learning_rate": 9.854885186131431e-06, "loss": 1.6259, "step": 1483 }, { "epoch": 0.1135772233277208, "grad_norm": 0.7704959266804765, "learning_rate": 9.8546204237593e-06, "loss": 1.4284, "step": 1484 }, { "epoch": 0.113653757844788, "grad_norm": 0.7607091048602481, "learning_rate": 9.854355424547765e-06, "loss": 1.5718, "step": 1485 }, { "epoch": 0.1137302923618552, "grad_norm": 0.7766050738538113, "learning_rate": 9.854090188513112e-06, "loss": 1.6271, "step": 1486 }, { "epoch": 0.11380682687892239, "grad_norm": 0.6427014765161918, "learning_rate": 9.853824715671638e-06, "loss": 1.482, "step": 1487 }, { "epoch": 0.11388336139598959, "grad_norm": 0.9973933008990907, "learning_rate": 9.853559006039651e-06, "loss": 1.6028, "step": 1488 }, { "epoch": 0.11395989591305679, "grad_norm": 0.5455778762061831, "learning_rate": 9.853293059633478e-06, "loss": 1.3105, "step": 1489 }, { "epoch": 0.11403643043012399, "grad_norm": 0.7342654732853602, "learning_rate": 9.853026876469463e-06, "loss": 1.732, "step": 1490 }, { "epoch": 0.11411296494719118, "grad_norm": 0.6897059133103598, "learning_rate": 9.852760456563956e-06, "loss": 1.8193, "step": 1491 }, { "epoch": 0.11418949946425838, "grad_norm": 0.7101888424492946, "learning_rate": 9.85249379993333e-06, "loss": 1.5147, "step": 1492 }, { "epoch": 0.11426603398132558, "grad_norm": 0.7330395281906953, "learning_rate": 9.852226906593966e-06, "loss": 1.0868, "step": 1493 }, { "epoch": 0.11434256849839278, "grad_norm": 1.167907580060761, "learning_rate": 9.851959776562267e-06, "loss": 1.6226, "step": 1494 }, { "epoch": 0.11441910301545997, "grad_norm": 0.6044707965481741, "learning_rate": 9.851692409854644e-06, "loss": 1.4214, "step": 1495 }, { "epoch": 0.11449563753252717, "grad_norm": 0.6575822507983973, "learning_rate": 9.851424806487526e-06, "loss": 1.3986, "step": 1496 }, { "epoch": 0.11457217204959437, "grad_norm": 0.6775602526783033, "learning_rate": 9.851156966477352e-06, "loss": 1.6425, "step": 1497 }, { "epoch": 0.11464870656666157, "grad_norm": 0.6736002718307782, "learning_rate": 9.850888889840583e-06, "loss": 1.483, "step": 1498 }, { "epoch": 0.11472524108372877, "grad_norm": 0.7801896606430367, "learning_rate": 9.850620576593687e-06, "loss": 1.9565, "step": 1499 }, { "epoch": 0.11480177560079596, "grad_norm": 0.5472840706863784, "learning_rate": 9.850352026753153e-06, "loss": 1.0843, "step": 1500 }, { "epoch": 0.11487831011786316, "grad_norm": 0.602969139907027, "learning_rate": 9.85008324033548e-06, "loss": 1.2333, "step": 1501 }, { "epoch": 0.11495484463493036, "grad_norm": 0.5502835856830561, "learning_rate": 9.849814217357182e-06, "loss": 1.3581, "step": 1502 }, { "epoch": 0.11503137915199756, "grad_norm": 0.6458987801237411, "learning_rate": 9.84954495783479e-06, "loss": 1.6249, "step": 1503 }, { "epoch": 0.11510791366906475, "grad_norm": 0.7182278479010711, "learning_rate": 9.84927546178485e-06, "loss": 1.5842, "step": 1504 }, { "epoch": 0.11518444818613195, "grad_norm": 0.557847760144535, "learning_rate": 9.849005729223916e-06, "loss": 1.1888, "step": 1505 }, { "epoch": 0.11526098270319915, "grad_norm": 0.6118642419504056, "learning_rate": 9.848735760168565e-06, "loss": 1.4843, "step": 1506 }, { "epoch": 0.11533751722026633, "grad_norm": 0.6773128062487713, "learning_rate": 9.848465554635383e-06, "loss": 1.7201, "step": 1507 }, { "epoch": 0.11541405173733353, "grad_norm": 0.5677236161152028, "learning_rate": 9.848195112640972e-06, "loss": 1.1779, "step": 1508 }, { "epoch": 0.11549058625440073, "grad_norm": 0.5788631051525677, "learning_rate": 9.84792443420195e-06, "loss": 1.2191, "step": 1509 }, { "epoch": 0.11556712077146793, "grad_norm": 0.5665573315035723, "learning_rate": 9.847653519334948e-06, "loss": 1.3324, "step": 1510 }, { "epoch": 0.11564365528853512, "grad_norm": 0.5754967600921174, "learning_rate": 9.847382368056611e-06, "loss": 1.4431, "step": 1511 }, { "epoch": 0.11572018980560232, "grad_norm": 0.7750550737724546, "learning_rate": 9.847110980383599e-06, "loss": 1.685, "step": 1512 }, { "epoch": 0.11579672432266952, "grad_norm": 0.6919335081411488, "learning_rate": 9.846839356332589e-06, "loss": 1.5435, "step": 1513 }, { "epoch": 0.11587325883973672, "grad_norm": 0.6414813406651197, "learning_rate": 9.846567495920268e-06, "loss": 1.6341, "step": 1514 }, { "epoch": 0.11594979335680392, "grad_norm": 0.6673041173392908, "learning_rate": 9.846295399163342e-06, "loss": 1.5287, "step": 1515 }, { "epoch": 0.11602632787387111, "grad_norm": 0.6413162376874867, "learning_rate": 9.846023066078529e-06, "loss": 1.5918, "step": 1516 }, { "epoch": 0.11610286239093831, "grad_norm": 0.6082402158853479, "learning_rate": 9.84575049668256e-06, "loss": 1.6454, "step": 1517 }, { "epoch": 0.11617939690800551, "grad_norm": 0.6413758210549937, "learning_rate": 9.845477690992185e-06, "loss": 1.665, "step": 1518 }, { "epoch": 0.1162559314250727, "grad_norm": 0.6682614382853791, "learning_rate": 9.845204649024164e-06, "loss": 1.6043, "step": 1519 }, { "epoch": 0.1163324659421399, "grad_norm": 0.5784350442395414, "learning_rate": 9.844931370795274e-06, "loss": 1.3662, "step": 1520 }, { "epoch": 0.1164090004592071, "grad_norm": 0.6614231963327297, "learning_rate": 9.844657856322307e-06, "loss": 1.3317, "step": 1521 }, { "epoch": 0.1164855349762743, "grad_norm": 0.5658400267712336, "learning_rate": 9.844384105622067e-06, "loss": 1.1848, "step": 1522 }, { "epoch": 0.1165620694933415, "grad_norm": 0.6879039114131242, "learning_rate": 9.844110118711378e-06, "loss": 1.4853, "step": 1523 }, { "epoch": 0.1166386040104087, "grad_norm": 0.5907313525182939, "learning_rate": 9.84383589560707e-06, "loss": 1.4508, "step": 1524 }, { "epoch": 0.11671513852747589, "grad_norm": 0.6346643903358392, "learning_rate": 9.843561436325992e-06, "loss": 1.4801, "step": 1525 }, { "epoch": 0.11679167304454309, "grad_norm": 0.5898695099996862, "learning_rate": 9.843286740885009e-06, "loss": 1.1879, "step": 1526 }, { "epoch": 0.11686820756161029, "grad_norm": 0.6210688682444905, "learning_rate": 9.843011809301003e-06, "loss": 1.3679, "step": 1527 }, { "epoch": 0.11694474207867749, "grad_norm": 0.6285959895388391, "learning_rate": 9.84273664159086e-06, "loss": 1.5024, "step": 1528 }, { "epoch": 0.11702127659574468, "grad_norm": 0.5861878004126382, "learning_rate": 9.842461237771492e-06, "loss": 1.3844, "step": 1529 }, { "epoch": 0.11709781111281188, "grad_norm": 0.5812032367395727, "learning_rate": 9.842185597859818e-06, "loss": 1.2037, "step": 1530 }, { "epoch": 0.11717434562987908, "grad_norm": 0.5912319070797905, "learning_rate": 9.841909721872775e-06, "loss": 1.3881, "step": 1531 }, { "epoch": 0.11725088014694628, "grad_norm": 0.9206613896512599, "learning_rate": 9.841633609827313e-06, "loss": 1.5025, "step": 1532 }, { "epoch": 0.11732741466401347, "grad_norm": 0.5975184930015812, "learning_rate": 9.841357261740398e-06, "loss": 1.4148, "step": 1533 }, { "epoch": 0.11740394918108067, "grad_norm": 0.8019379719503001, "learning_rate": 9.841080677629009e-06, "loss": 1.9807, "step": 1534 }, { "epoch": 0.11748048369814787, "grad_norm": 0.6625104045052832, "learning_rate": 9.840803857510142e-06, "loss": 1.6439, "step": 1535 }, { "epoch": 0.11755701821521507, "grad_norm": 0.6475391177177918, "learning_rate": 9.840526801400803e-06, "loss": 1.6774, "step": 1536 }, { "epoch": 0.11763355273228226, "grad_norm": 0.6686746749343511, "learning_rate": 9.840249509318016e-06, "loss": 1.5223, "step": 1537 }, { "epoch": 0.11771008724934946, "grad_norm": 0.6634690235329115, "learning_rate": 9.83997198127882e-06, "loss": 1.6297, "step": 1538 }, { "epoch": 0.11778662176641666, "grad_norm": 0.5300666384686722, "learning_rate": 9.839694217300265e-06, "loss": 1.1875, "step": 1539 }, { "epoch": 0.11786315628348386, "grad_norm": 0.5482599309069592, "learning_rate": 9.83941621739942e-06, "loss": 0.9931, "step": 1540 }, { "epoch": 0.11793969080055104, "grad_norm": 0.6481601876210656, "learning_rate": 9.839137981593365e-06, "loss": 1.3929, "step": 1541 }, { "epoch": 0.11801622531761824, "grad_norm": 0.580551219884453, "learning_rate": 9.838859509899195e-06, "loss": 1.4044, "step": 1542 }, { "epoch": 0.11809275983468544, "grad_norm": 0.6909207270251271, "learning_rate": 9.83858080233402e-06, "loss": 1.677, "step": 1543 }, { "epoch": 0.11816929435175264, "grad_norm": 0.5631880358919428, "learning_rate": 9.838301858914966e-06, "loss": 1.1803, "step": 1544 }, { "epoch": 0.11824582886881983, "grad_norm": 0.7634733953950377, "learning_rate": 9.838022679659172e-06, "loss": 1.4957, "step": 1545 }, { "epoch": 0.11832236338588703, "grad_norm": 1.5665583811299026, "learning_rate": 9.83774326458379e-06, "loss": 1.8628, "step": 1546 }, { "epoch": 0.11839889790295423, "grad_norm": 0.5081763834529044, "learning_rate": 9.83746361370599e-06, "loss": 1.1968, "step": 1547 }, { "epoch": 0.11847543242002143, "grad_norm": 0.6929203315461855, "learning_rate": 9.837183727042954e-06, "loss": 1.5689, "step": 1548 }, { "epoch": 0.11855196693708862, "grad_norm": 0.8003688286623453, "learning_rate": 9.836903604611878e-06, "loss": 1.4245, "step": 1549 }, { "epoch": 0.11862850145415582, "grad_norm": 0.6331351687176389, "learning_rate": 9.836623246429975e-06, "loss": 1.5908, "step": 1550 }, { "epoch": 0.11870503597122302, "grad_norm": 0.6724590854100472, "learning_rate": 9.83634265251447e-06, "loss": 1.6558, "step": 1551 }, { "epoch": 0.11878157048829022, "grad_norm": 0.6346907159264192, "learning_rate": 9.836061822882604e-06, "loss": 1.4711, "step": 1552 }, { "epoch": 0.11885810500535741, "grad_norm": 0.5482560864889687, "learning_rate": 9.835780757551631e-06, "loss": 1.1882, "step": 1553 }, { "epoch": 0.11893463952242461, "grad_norm": 0.5557833539886258, "learning_rate": 9.835499456538822e-06, "loss": 1.31, "step": 1554 }, { "epoch": 0.11901117403949181, "grad_norm": 0.655912340360435, "learning_rate": 9.83521791986146e-06, "loss": 1.6151, "step": 1555 }, { "epoch": 0.11908770855655901, "grad_norm": 0.5239660721390661, "learning_rate": 9.834936147536847e-06, "loss": 1.1895, "step": 1556 }, { "epoch": 0.1191642430736262, "grad_norm": 0.5971867642211468, "learning_rate": 9.83465413958229e-06, "loss": 1.3285, "step": 1557 }, { "epoch": 0.1192407775906934, "grad_norm": 1.1958475004982494, "learning_rate": 9.83437189601512e-06, "loss": 1.1546, "step": 1558 }, { "epoch": 0.1193173121077606, "grad_norm": 0.7135235017887861, "learning_rate": 9.834089416852677e-06, "loss": 1.3961, "step": 1559 }, { "epoch": 0.1193938466248278, "grad_norm": 0.4935145425384386, "learning_rate": 9.833806702112319e-06, "loss": 0.9363, "step": 1560 }, { "epoch": 0.119470381141895, "grad_norm": 0.6066475495637798, "learning_rate": 9.833523751811417e-06, "loss": 1.4063, "step": 1561 }, { "epoch": 0.1195469156589622, "grad_norm": 0.593050326159162, "learning_rate": 9.833240565967356e-06, "loss": 1.4378, "step": 1562 }, { "epoch": 0.11962345017602939, "grad_norm": 0.5940389368545609, "learning_rate": 9.832957144597534e-06, "loss": 1.3814, "step": 1563 }, { "epoch": 0.11969998469309659, "grad_norm": 0.6746156987180951, "learning_rate": 9.832673487719368e-06, "loss": 1.6024, "step": 1564 }, { "epoch": 0.11977651921016379, "grad_norm": 0.5252823921737518, "learning_rate": 9.832389595350284e-06, "loss": 1.1808, "step": 1565 }, { "epoch": 0.11985305372723098, "grad_norm": 1.0951961058665072, "learning_rate": 9.832105467507728e-06, "loss": 1.6378, "step": 1566 }, { "epoch": 0.11992958824429818, "grad_norm": 0.5575935641994689, "learning_rate": 9.831821104209156e-06, "loss": 1.3151, "step": 1567 }, { "epoch": 0.12000612276136538, "grad_norm": 0.6234599247048639, "learning_rate": 9.83153650547204e-06, "loss": 1.3459, "step": 1568 }, { "epoch": 0.12008265727843258, "grad_norm": 0.5710091493784336, "learning_rate": 9.831251671313866e-06, "loss": 1.4021, "step": 1569 }, { "epoch": 0.12015919179549978, "grad_norm": 0.8137664475116269, "learning_rate": 9.830966601752138e-06, "loss": 1.2411, "step": 1570 }, { "epoch": 0.12023572631256697, "grad_norm": 0.6506247251759193, "learning_rate": 9.830681296804368e-06, "loss": 1.4599, "step": 1571 }, { "epoch": 0.12031226082963417, "grad_norm": 0.7377515354948255, "learning_rate": 9.830395756488088e-06, "loss": 1.5335, "step": 1572 }, { "epoch": 0.12038879534670137, "grad_norm": 0.6082445491157211, "learning_rate": 9.830109980820843e-06, "loss": 1.5243, "step": 1573 }, { "epoch": 0.12046532986376857, "grad_norm": 0.6211403670941337, "learning_rate": 9.82982396982019e-06, "loss": 1.5949, "step": 1574 }, { "epoch": 0.12054186438083575, "grad_norm": 0.598261032195583, "learning_rate": 9.829537723503704e-06, "loss": 1.3636, "step": 1575 }, { "epoch": 0.12061839889790295, "grad_norm": 0.6194620786347764, "learning_rate": 9.829251241888971e-06, "loss": 1.5028, "step": 1576 }, { "epoch": 0.12069493341497015, "grad_norm": 0.671556084580693, "learning_rate": 9.828964524993596e-06, "loss": 1.6882, "step": 1577 }, { "epoch": 0.12077146793203734, "grad_norm": 0.6079315096621102, "learning_rate": 9.828677572835192e-06, "loss": 1.4642, "step": 1578 }, { "epoch": 0.12084800244910454, "grad_norm": 0.5857729415267456, "learning_rate": 9.828390385431394e-06, "loss": 1.3313, "step": 1579 }, { "epoch": 0.12092453696617174, "grad_norm": 0.590934304197789, "learning_rate": 9.828102962799846e-06, "loss": 1.4379, "step": 1580 }, { "epoch": 0.12100107148323894, "grad_norm": 0.5774937476297756, "learning_rate": 9.827815304958207e-06, "loss": 1.5027, "step": 1581 }, { "epoch": 0.12107760600030613, "grad_norm": 0.6366272914987494, "learning_rate": 9.827527411924154e-06, "loss": 1.737, "step": 1582 }, { "epoch": 0.12115414051737333, "grad_norm": 0.6037100197704807, "learning_rate": 9.827239283715374e-06, "loss": 1.5113, "step": 1583 }, { "epoch": 0.12123067503444053, "grad_norm": 0.6010988185286739, "learning_rate": 9.82695092034957e-06, "loss": 1.6368, "step": 1584 }, { "epoch": 0.12130720955150773, "grad_norm": 0.6166634622861618, "learning_rate": 9.826662321844463e-06, "loss": 1.6498, "step": 1585 }, { "epoch": 0.12138374406857493, "grad_norm": 0.6510737313039064, "learning_rate": 9.826373488217782e-06, "loss": 1.6042, "step": 1586 }, { "epoch": 0.12146027858564212, "grad_norm": 0.9452004282421013, "learning_rate": 9.826084419487277e-06, "loss": 1.6977, "step": 1587 }, { "epoch": 0.12153681310270932, "grad_norm": 0.5652296206201423, "learning_rate": 9.825795115670705e-06, "loss": 1.2716, "step": 1588 }, { "epoch": 0.12161334761977652, "grad_norm": 0.5539798240626803, "learning_rate": 9.825505576785847e-06, "loss": 1.2155, "step": 1589 }, { "epoch": 0.12168988213684372, "grad_norm": 0.5942111035731236, "learning_rate": 9.825215802850487e-06, "loss": 1.4481, "step": 1590 }, { "epoch": 0.12176641665391091, "grad_norm": 0.8165899387415783, "learning_rate": 9.824925793882436e-06, "loss": 1.7814, "step": 1591 }, { "epoch": 0.12184295117097811, "grad_norm": 0.5487536939685576, "learning_rate": 9.824635549899508e-06, "loss": 1.3429, "step": 1592 }, { "epoch": 0.12191948568804531, "grad_norm": 0.5570201137805009, "learning_rate": 9.82434507091954e-06, "loss": 1.2566, "step": 1593 }, { "epoch": 0.12199602020511251, "grad_norm": 0.6218939669841798, "learning_rate": 9.824054356960376e-06, "loss": 1.2774, "step": 1594 }, { "epoch": 0.1220725547221797, "grad_norm": 0.6009935696964231, "learning_rate": 9.823763408039882e-06, "loss": 1.4351, "step": 1595 }, { "epoch": 0.1221490892392469, "grad_norm": 1.0733426060056703, "learning_rate": 9.823472224175933e-06, "loss": 1.8409, "step": 1596 }, { "epoch": 0.1222256237563141, "grad_norm": 0.5340099207831515, "learning_rate": 9.823180805386419e-06, "loss": 1.067, "step": 1597 }, { "epoch": 0.1223021582733813, "grad_norm": 0.6281807902255763, "learning_rate": 9.822889151689249e-06, "loss": 1.6217, "step": 1598 }, { "epoch": 0.1223786927904485, "grad_norm": 0.8704245347601473, "learning_rate": 9.82259726310234e-06, "loss": 1.6046, "step": 1599 }, { "epoch": 0.1224552273075157, "grad_norm": 0.8327172352757866, "learning_rate": 9.822305139643628e-06, "loss": 1.426, "step": 1600 }, { "epoch": 0.12253176182458289, "grad_norm": 0.6917658170378104, "learning_rate": 9.822012781331064e-06, "loss": 1.6751, "step": 1601 }, { "epoch": 0.12260829634165009, "grad_norm": 0.7175573814899918, "learning_rate": 9.821720188182607e-06, "loss": 1.3685, "step": 1602 }, { "epoch": 0.12268483085871729, "grad_norm": 0.5888364998361282, "learning_rate": 9.821427360216237e-06, "loss": 1.5031, "step": 1603 }, { "epoch": 0.12276136537578448, "grad_norm": 2.1171299294162735, "learning_rate": 9.821134297449947e-06, "loss": 1.4469, "step": 1604 }, { "epoch": 0.12283789989285168, "grad_norm": 0.7589168158884052, "learning_rate": 9.820840999901742e-06, "loss": 1.2879, "step": 1605 }, { "epoch": 0.12291443440991888, "grad_norm": 0.6664403345520435, "learning_rate": 9.820547467589645e-06, "loss": 1.7194, "step": 1606 }, { "epoch": 0.12299096892698608, "grad_norm": 0.6293746937351111, "learning_rate": 9.820253700531689e-06, "loss": 1.5582, "step": 1607 }, { "epoch": 0.12306750344405326, "grad_norm": 0.5490796141588427, "learning_rate": 9.819959698745925e-06, "loss": 1.3321, "step": 1608 }, { "epoch": 0.12314403796112046, "grad_norm": 0.5121069997457722, "learning_rate": 9.819665462250419e-06, "loss": 1.1844, "step": 1609 }, { "epoch": 0.12322057247818766, "grad_norm": 0.822867479335361, "learning_rate": 9.819370991063247e-06, "loss": 1.4622, "step": 1610 }, { "epoch": 0.12329710699525485, "grad_norm": 0.5483513925958221, "learning_rate": 9.819076285202504e-06, "loss": 1.1325, "step": 1611 }, { "epoch": 0.12337364151232205, "grad_norm": 0.5898966869752146, "learning_rate": 9.818781344686297e-06, "loss": 1.1417, "step": 1612 }, { "epoch": 0.12345017602938925, "grad_norm": 0.6004132675943137, "learning_rate": 9.818486169532748e-06, "loss": 1.1749, "step": 1613 }, { "epoch": 0.12352671054645645, "grad_norm": 0.7208869160347937, "learning_rate": 9.818190759759993e-06, "loss": 1.7492, "step": 1614 }, { "epoch": 0.12360324506352365, "grad_norm": 0.6087569409373672, "learning_rate": 9.817895115386185e-06, "loss": 1.3358, "step": 1615 }, { "epoch": 0.12367977958059084, "grad_norm": 0.6047806356412369, "learning_rate": 9.817599236429489e-06, "loss": 1.4901, "step": 1616 }, { "epoch": 0.12375631409765804, "grad_norm": 0.7032890360885176, "learning_rate": 9.817303122908082e-06, "loss": 1.6664, "step": 1617 }, { "epoch": 0.12383284861472524, "grad_norm": 0.7035945643436682, "learning_rate": 9.817006774840159e-06, "loss": 1.6963, "step": 1618 }, { "epoch": 0.12390938313179244, "grad_norm": 0.6618419286624427, "learning_rate": 9.816710192243929e-06, "loss": 1.6141, "step": 1619 }, { "epoch": 0.12398591764885963, "grad_norm": 0.5127824783047678, "learning_rate": 9.816413375137615e-06, "loss": 1.1715, "step": 1620 }, { "epoch": 0.12406245216592683, "grad_norm": 0.6786123142879565, "learning_rate": 9.816116323539455e-06, "loss": 1.7331, "step": 1621 }, { "epoch": 0.12413898668299403, "grad_norm": 0.6260923873823171, "learning_rate": 9.8158190374677e-06, "loss": 1.6772, "step": 1622 }, { "epoch": 0.12421552120006123, "grad_norm": 0.6017520450959015, "learning_rate": 9.815521516940617e-06, "loss": 1.3059, "step": 1623 }, { "epoch": 0.12429205571712842, "grad_norm": 0.9393709002260363, "learning_rate": 9.815223761976484e-06, "loss": 1.8881, "step": 1624 }, { "epoch": 0.12436859023419562, "grad_norm": 0.5976829279369659, "learning_rate": 9.814925772593598e-06, "loss": 1.1859, "step": 1625 }, { "epoch": 0.12444512475126282, "grad_norm": 1.4009081595617252, "learning_rate": 9.81462754881027e-06, "loss": 1.2193, "step": 1626 }, { "epoch": 0.12452165926833002, "grad_norm": 0.5977515274671048, "learning_rate": 9.81432909064482e-06, "loss": 1.4447, "step": 1627 }, { "epoch": 0.12459819378539722, "grad_norm": 0.639410735126776, "learning_rate": 9.814030398115589e-06, "loss": 1.5203, "step": 1628 }, { "epoch": 0.12467472830246441, "grad_norm": 0.6584136870639301, "learning_rate": 9.813731471240928e-06, "loss": 1.4674, "step": 1629 }, { "epoch": 0.12475126281953161, "grad_norm": 0.8554473070012216, "learning_rate": 9.813432310039205e-06, "loss": 1.7321, "step": 1630 }, { "epoch": 0.12482779733659881, "grad_norm": 1.2282969378348079, "learning_rate": 9.813132914528798e-06, "loss": 1.5891, "step": 1631 }, { "epoch": 0.124904331853666, "grad_norm": 0.6098560706766248, "learning_rate": 9.81283328472811e-06, "loss": 1.3775, "step": 1632 }, { "epoch": 0.1249808663707332, "grad_norm": 0.6351959542497179, "learning_rate": 9.812533420655544e-06, "loss": 1.4266, "step": 1633 }, { "epoch": 0.1250574008878004, "grad_norm": 0.7005952163486368, "learning_rate": 9.812233322329529e-06, "loss": 1.6268, "step": 1634 }, { "epoch": 0.12513393540486759, "grad_norm": 0.772877373903333, "learning_rate": 9.8119329897685e-06, "loss": 1.7874, "step": 1635 }, { "epoch": 0.1252104699219348, "grad_norm": 0.5344196665729873, "learning_rate": 9.811632422990914e-06, "loss": 1.1885, "step": 1636 }, { "epoch": 0.12528700443900198, "grad_norm": 0.55926193890418, "learning_rate": 9.811331622015238e-06, "loss": 1.3612, "step": 1637 }, { "epoch": 0.1253635389560692, "grad_norm": 0.5396676263892376, "learning_rate": 9.811030586859952e-06, "loss": 1.1659, "step": 1638 }, { "epoch": 0.12544007347313638, "grad_norm": 0.6808415684013769, "learning_rate": 9.810729317543555e-06, "loss": 1.7158, "step": 1639 }, { "epoch": 0.1255166079902036, "grad_norm": 0.6192650128607009, "learning_rate": 9.810427814084557e-06, "loss": 1.6141, "step": 1640 }, { "epoch": 0.12559314250727077, "grad_norm": 1.0307701074096849, "learning_rate": 9.810126076501482e-06, "loss": 1.512, "step": 1641 }, { "epoch": 0.12566967702433798, "grad_norm": 0.6345404384627449, "learning_rate": 9.809824104812871e-06, "loss": 1.414, "step": 1642 }, { "epoch": 0.12574621154140517, "grad_norm": 0.6638313246314339, "learning_rate": 9.80952189903728e-06, "loss": 1.319, "step": 1643 }, { "epoch": 0.12582274605847238, "grad_norm": 0.5073999614653165, "learning_rate": 9.809219459193272e-06, "loss": 1.2207, "step": 1644 }, { "epoch": 0.12589928057553956, "grad_norm": 0.6257348914898534, "learning_rate": 9.808916785299435e-06, "loss": 1.453, "step": 1645 }, { "epoch": 0.12597581509260677, "grad_norm": 0.602281243740175, "learning_rate": 9.808613877374364e-06, "loss": 1.6337, "step": 1646 }, { "epoch": 0.12605234960967396, "grad_norm": 0.549345230711357, "learning_rate": 9.80831073543667e-06, "loss": 1.2002, "step": 1647 }, { "epoch": 0.12612888412674117, "grad_norm": 0.6135824727053996, "learning_rate": 9.80800735950498e-06, "loss": 1.3548, "step": 1648 }, { "epoch": 0.12620541864380835, "grad_norm": 0.9642974964176103, "learning_rate": 9.807703749597935e-06, "loss": 1.5941, "step": 1649 }, { "epoch": 0.12628195316087557, "grad_norm": 0.7060617155096323, "learning_rate": 9.807399905734188e-06, "loss": 1.6304, "step": 1650 }, { "epoch": 0.12635848767794275, "grad_norm": 0.7640535990079559, "learning_rate": 9.807095827932408e-06, "loss": 1.8442, "step": 1651 }, { "epoch": 0.12643502219500996, "grad_norm": 0.5207091659195356, "learning_rate": 9.80679151621128e-06, "loss": 1.1393, "step": 1652 }, { "epoch": 0.12651155671207714, "grad_norm": 0.5306091443531649, "learning_rate": 9.806486970589501e-06, "loss": 1.1857, "step": 1653 }, { "epoch": 0.12658809122914436, "grad_norm": 0.6119703787625838, "learning_rate": 9.806182191085785e-06, "loss": 1.524, "step": 1654 }, { "epoch": 0.12666462574621154, "grad_norm": 0.6028800974087345, "learning_rate": 9.805877177718855e-06, "loss": 1.3808, "step": 1655 }, { "epoch": 0.12674116026327875, "grad_norm": 0.5991543526093243, "learning_rate": 9.805571930507453e-06, "loss": 1.2163, "step": 1656 }, { "epoch": 0.12681769478034594, "grad_norm": 0.5688348496440709, "learning_rate": 9.805266449470337e-06, "loss": 1.3689, "step": 1657 }, { "epoch": 0.12689422929741315, "grad_norm": 0.6758971750697785, "learning_rate": 9.804960734626274e-06, "loss": 1.7389, "step": 1658 }, { "epoch": 0.12697076381448033, "grad_norm": 0.7381364939006634, "learning_rate": 9.80465478599405e-06, "loss": 1.3572, "step": 1659 }, { "epoch": 0.12704729833154751, "grad_norm": 0.5688559949472397, "learning_rate": 9.80434860359246e-06, "loss": 1.3155, "step": 1660 }, { "epoch": 0.12712383284861473, "grad_norm": 0.7719798822678036, "learning_rate": 9.804042187440319e-06, "loss": 1.5599, "step": 1661 }, { "epoch": 0.1272003673656819, "grad_norm": 0.5059029394305847, "learning_rate": 9.803735537556454e-06, "loss": 0.9574, "step": 1662 }, { "epoch": 0.12727690188274912, "grad_norm": 0.5883084381279785, "learning_rate": 9.803428653959708e-06, "loss": 1.2631, "step": 1663 }, { "epoch": 0.1273534363998163, "grad_norm": 0.82560493086411, "learning_rate": 9.803121536668935e-06, "loss": 1.1534, "step": 1664 }, { "epoch": 0.12742997091688352, "grad_norm": 0.6264086095885084, "learning_rate": 9.802814185703007e-06, "loss": 1.3738, "step": 1665 }, { "epoch": 0.1275065054339507, "grad_norm": 0.6239905877048162, "learning_rate": 9.802506601080805e-06, "loss": 1.4481, "step": 1666 }, { "epoch": 0.1275830399510179, "grad_norm": 0.5560682016434955, "learning_rate": 9.802198782821231e-06, "loss": 1.4688, "step": 1667 }, { "epoch": 0.1276595744680851, "grad_norm": 0.6387894441612716, "learning_rate": 9.801890730943197e-06, "loss": 1.6002, "step": 1668 }, { "epoch": 0.1277361089851523, "grad_norm": 0.6385020138420565, "learning_rate": 9.801582445465632e-06, "loss": 1.6406, "step": 1669 }, { "epoch": 0.1278126435022195, "grad_norm": 0.5673240560989374, "learning_rate": 9.801273926407476e-06, "loss": 1.29, "step": 1670 }, { "epoch": 0.1278891780192867, "grad_norm": 0.5369588168256326, "learning_rate": 9.800965173787687e-06, "loss": 1.1998, "step": 1671 }, { "epoch": 0.1279657125363539, "grad_norm": 0.6365841160781369, "learning_rate": 9.800656187625234e-06, "loss": 1.2559, "step": 1672 }, { "epoch": 0.1280422470534211, "grad_norm": 1.1356295506063596, "learning_rate": 9.800346967939104e-06, "loss": 1.5045, "step": 1673 }, { "epoch": 0.12811878157048828, "grad_norm": 0.5495718020164758, "learning_rate": 9.800037514748294e-06, "loss": 1.1772, "step": 1674 }, { "epoch": 0.1281953160875555, "grad_norm": 0.543955913886195, "learning_rate": 9.799727828071821e-06, "loss": 1.2585, "step": 1675 }, { "epoch": 0.12827185060462268, "grad_norm": 0.7103214215494571, "learning_rate": 9.79941790792871e-06, "loss": 1.513, "step": 1676 }, { "epoch": 0.1283483851216899, "grad_norm": 0.5469378471427474, "learning_rate": 9.799107754338005e-06, "loss": 1.3087, "step": 1677 }, { "epoch": 0.12842491963875707, "grad_norm": 0.6112811182967061, "learning_rate": 9.798797367318762e-06, "loss": 1.3453, "step": 1678 }, { "epoch": 0.12850145415582429, "grad_norm": 0.8777821950130905, "learning_rate": 9.798486746890054e-06, "loss": 1.5902, "step": 1679 }, { "epoch": 0.12857798867289147, "grad_norm": 0.6851535887020365, "learning_rate": 9.798175893070964e-06, "loss": 1.7729, "step": 1680 }, { "epoch": 0.12865452318995868, "grad_norm": 0.7388278811595627, "learning_rate": 9.797864805880593e-06, "loss": 1.6321, "step": 1681 }, { "epoch": 0.12873105770702586, "grad_norm": 0.5560719865718711, "learning_rate": 9.797553485338053e-06, "loss": 1.4365, "step": 1682 }, { "epoch": 0.12880759222409308, "grad_norm": 0.6489280447015969, "learning_rate": 9.797241931462476e-06, "loss": 1.6912, "step": 1683 }, { "epoch": 0.12888412674116026, "grad_norm": 0.5522552438660717, "learning_rate": 9.796930144273002e-06, "loss": 1.2742, "step": 1684 }, { "epoch": 0.12896066125822747, "grad_norm": 0.8038209139920551, "learning_rate": 9.796618123788793e-06, "loss": 1.8032, "step": 1685 }, { "epoch": 0.12903719577529466, "grad_norm": 0.6347849800103772, "learning_rate": 9.796305870029013e-06, "loss": 1.5947, "step": 1686 }, { "epoch": 0.12911373029236187, "grad_norm": 0.622308088411441, "learning_rate": 9.795993383012852e-06, "loss": 1.6567, "step": 1687 }, { "epoch": 0.12919026480942905, "grad_norm": 0.5928366201059848, "learning_rate": 9.79568066275951e-06, "loss": 1.4967, "step": 1688 }, { "epoch": 0.12926679932649626, "grad_norm": 0.6740081286199465, "learning_rate": 9.795367709288201e-06, "loss": 1.5338, "step": 1689 }, { "epoch": 0.12934333384356345, "grad_norm": 0.5864452021416139, "learning_rate": 9.795054522618154e-06, "loss": 1.3241, "step": 1690 }, { "epoch": 0.12941986836063066, "grad_norm": 0.6844382046902224, "learning_rate": 9.794741102768615e-06, "loss": 1.5344, "step": 1691 }, { "epoch": 0.12949640287769784, "grad_norm": 0.5516825717552816, "learning_rate": 9.794427449758835e-06, "loss": 1.2037, "step": 1692 }, { "epoch": 0.12957293739476503, "grad_norm": 0.6020765002270778, "learning_rate": 9.79411356360809e-06, "loss": 1.5035, "step": 1693 }, { "epoch": 0.12964947191183224, "grad_norm": 0.7824726607857193, "learning_rate": 9.793799444335666e-06, "loss": 1.3218, "step": 1694 }, { "epoch": 0.12972600642889942, "grad_norm": 0.6231385953012336, "learning_rate": 9.793485091960862e-06, "loss": 1.5124, "step": 1695 }, { "epoch": 0.12980254094596663, "grad_norm": 0.5044310072124791, "learning_rate": 9.793170506502993e-06, "loss": 1.1304, "step": 1696 }, { "epoch": 0.12987907546303382, "grad_norm": 0.6269535535892362, "learning_rate": 9.79285568798139e-06, "loss": 1.5615, "step": 1697 }, { "epoch": 0.12995560998010103, "grad_norm": 0.560451691213549, "learning_rate": 9.792540636415395e-06, "loss": 1.3569, "step": 1698 }, { "epoch": 0.1300321444971682, "grad_norm": 0.5847763188455589, "learning_rate": 9.792225351824367e-06, "loss": 1.4672, "step": 1699 }, { "epoch": 0.13010867901423542, "grad_norm": 0.6517667700549242, "learning_rate": 9.791909834227676e-06, "loss": 1.402, "step": 1700 }, { "epoch": 0.1301852135313026, "grad_norm": 0.5795808154748587, "learning_rate": 9.791594083644707e-06, "loss": 1.3352, "step": 1701 }, { "epoch": 0.13026174804836982, "grad_norm": 0.6739315342626561, "learning_rate": 9.791278100094865e-06, "loss": 1.5051, "step": 1702 }, { "epoch": 0.130338282565437, "grad_norm": 0.5032571806962084, "learning_rate": 9.790961883597562e-06, "loss": 1.0745, "step": 1703 }, { "epoch": 0.13041481708250421, "grad_norm": 0.5850173170209346, "learning_rate": 9.790645434172228e-06, "loss": 1.3581, "step": 1704 }, { "epoch": 0.1304913515995714, "grad_norm": 0.637958784361177, "learning_rate": 9.790328751838308e-06, "loss": 1.336, "step": 1705 }, { "epoch": 0.1305678861166386, "grad_norm": 1.2876977811206403, "learning_rate": 9.790011836615256e-06, "loss": 1.6171, "step": 1706 }, { "epoch": 0.1306444206337058, "grad_norm": 0.7793559766589783, "learning_rate": 9.789694688522548e-06, "loss": 1.84, "step": 1707 }, { "epoch": 0.130720955150773, "grad_norm": 0.6791749803273442, "learning_rate": 9.78937730757967e-06, "loss": 1.5305, "step": 1708 }, { "epoch": 0.1307974896678402, "grad_norm": 0.6565825431827406, "learning_rate": 9.78905969380612e-06, "loss": 1.589, "step": 1709 }, { "epoch": 0.1308740241849074, "grad_norm": 0.5907727839983097, "learning_rate": 9.788741847221418e-06, "loss": 1.2384, "step": 1710 }, { "epoch": 0.13095055870197458, "grad_norm": 0.5910586043043144, "learning_rate": 9.78842376784509e-06, "loss": 1.421, "step": 1711 }, { "epoch": 0.1310270932190418, "grad_norm": 0.6878524914408964, "learning_rate": 9.788105455696679e-06, "loss": 1.633, "step": 1712 }, { "epoch": 0.13110362773610898, "grad_norm": 0.6392612104147097, "learning_rate": 9.787786910795746e-06, "loss": 1.3165, "step": 1713 }, { "epoch": 0.1311801622531762, "grad_norm": 0.5763220659514381, "learning_rate": 9.787468133161861e-06, "loss": 1.2616, "step": 1714 }, { "epoch": 0.13125669677024338, "grad_norm": 0.48311798927310085, "learning_rate": 9.787149122814613e-06, "loss": 1.0979, "step": 1715 }, { "epoch": 0.1313332312873106, "grad_norm": 0.5888193352249984, "learning_rate": 9.7868298797736e-06, "loss": 1.521, "step": 1716 }, { "epoch": 0.13140976580437777, "grad_norm": 0.5914578628790009, "learning_rate": 9.786510404058439e-06, "loss": 1.2004, "step": 1717 }, { "epoch": 0.13148630032144498, "grad_norm": 0.8823047259712627, "learning_rate": 9.786190695688758e-06, "loss": 1.2183, "step": 1718 }, { "epoch": 0.13156283483851217, "grad_norm": 0.5981673361911163, "learning_rate": 9.785870754684205e-06, "loss": 1.1571, "step": 1719 }, { "epoch": 0.13163936935557938, "grad_norm": 0.596932441094714, "learning_rate": 9.785550581064434e-06, "loss": 1.2857, "step": 1720 }, { "epoch": 0.13171590387264656, "grad_norm": 0.6032134546519763, "learning_rate": 9.785230174849118e-06, "loss": 1.3953, "step": 1721 }, { "epoch": 0.13179243838971377, "grad_norm": 1.2990469017922672, "learning_rate": 9.784909536057945e-06, "loss": 1.6687, "step": 1722 }, { "epoch": 0.13186897290678096, "grad_norm": 0.7158777939556706, "learning_rate": 9.784588664710613e-06, "loss": 1.5154, "step": 1723 }, { "epoch": 0.13194550742384817, "grad_norm": 0.7500991289724133, "learning_rate": 9.784267560826844e-06, "loss": 1.653, "step": 1724 }, { "epoch": 0.13202204194091535, "grad_norm": 0.8138393831912456, "learning_rate": 9.78394622442636e-06, "loss": 1.4277, "step": 1725 }, { "epoch": 0.13209857645798256, "grad_norm": 0.5503634443344431, "learning_rate": 9.78362465552891e-06, "loss": 1.0564, "step": 1726 }, { "epoch": 0.13217511097504975, "grad_norm": 0.6659440487578274, "learning_rate": 9.78330285415425e-06, "loss": 1.4249, "step": 1727 }, { "epoch": 0.13225164549211693, "grad_norm": 0.5981244985458706, "learning_rate": 9.78298082032215e-06, "loss": 1.5118, "step": 1728 }, { "epoch": 0.13232818000918414, "grad_norm": 0.6468080388460846, "learning_rate": 9.782658554052401e-06, "loss": 1.4275, "step": 1729 }, { "epoch": 0.13240471452625133, "grad_norm": 0.6988152105405976, "learning_rate": 9.782336055364806e-06, "loss": 1.405, "step": 1730 }, { "epoch": 0.13248124904331854, "grad_norm": 0.6464561387541314, "learning_rate": 9.782013324279172e-06, "loss": 1.5291, "step": 1731 }, { "epoch": 0.13255778356038572, "grad_norm": 0.6466056947828834, "learning_rate": 9.781690360815335e-06, "loss": 1.6554, "step": 1732 }, { "epoch": 0.13263431807745293, "grad_norm": 0.6152559090089285, "learning_rate": 9.781367164993137e-06, "loss": 1.5054, "step": 1733 }, { "epoch": 0.13271085259452012, "grad_norm": 0.6625340233162855, "learning_rate": 9.781043736832439e-06, "loss": 1.6123, "step": 1734 }, { "epoch": 0.13278738711158733, "grad_norm": 0.5553042366540768, "learning_rate": 9.780720076353106e-06, "loss": 1.2365, "step": 1735 }, { "epoch": 0.1328639216286545, "grad_norm": 0.6466273909986294, "learning_rate": 9.780396183575031e-06, "loss": 1.6129, "step": 1736 }, { "epoch": 0.13294045614572172, "grad_norm": 0.5554073339790341, "learning_rate": 9.780072058518115e-06, "loss": 1.2269, "step": 1737 }, { "epoch": 0.1330169906627889, "grad_norm": 0.6703619487598704, "learning_rate": 9.77974770120227e-06, "loss": 1.4829, "step": 1738 }, { "epoch": 0.13309352517985612, "grad_norm": 0.7288433436602655, "learning_rate": 9.779423111647428e-06, "loss": 1.5719, "step": 1739 }, { "epoch": 0.1331700596969233, "grad_norm": 0.6028559720279622, "learning_rate": 9.77909828987353e-06, "loss": 1.3941, "step": 1740 }, { "epoch": 0.13324659421399052, "grad_norm": 0.7531369727352426, "learning_rate": 9.778773235900536e-06, "loss": 1.7037, "step": 1741 }, { "epoch": 0.1333231287310577, "grad_norm": 0.8698058191889898, "learning_rate": 9.77844794974842e-06, "loss": 1.625, "step": 1742 }, { "epoch": 0.1333996632481249, "grad_norm": 0.5165439792865304, "learning_rate": 9.778122431437164e-06, "loss": 1.2136, "step": 1743 }, { "epoch": 0.1334761977651921, "grad_norm": 0.6330525230269328, "learning_rate": 9.777796680986772e-06, "loss": 1.3929, "step": 1744 }, { "epoch": 0.1335527322822593, "grad_norm": 0.6570326456279985, "learning_rate": 9.77747069841726e-06, "loss": 1.4644, "step": 1745 }, { "epoch": 0.1336292667993265, "grad_norm": 0.7491459840891894, "learning_rate": 9.777144483748655e-06, "loss": 1.6797, "step": 1746 }, { "epoch": 0.1337058013163937, "grad_norm": 1.2450535959857287, "learning_rate": 9.776818037001e-06, "loss": 1.728, "step": 1747 }, { "epoch": 0.13378233583346089, "grad_norm": 0.5845419573441129, "learning_rate": 9.776491358194356e-06, "loss": 1.2799, "step": 1748 }, { "epoch": 0.1338588703505281, "grad_norm": 0.6620469297995308, "learning_rate": 9.776164447348792e-06, "loss": 1.2768, "step": 1749 }, { "epoch": 0.13393540486759528, "grad_norm": 0.6476086212059403, "learning_rate": 9.775837304484397e-06, "loss": 1.436, "step": 1750 }, { "epoch": 0.1340119393846625, "grad_norm": 0.5387734278148253, "learning_rate": 9.77550992962127e-06, "loss": 1.3722, "step": 1751 }, { "epoch": 0.13408847390172968, "grad_norm": 0.6455849729360013, "learning_rate": 9.775182322779526e-06, "loss": 1.5386, "step": 1752 }, { "epoch": 0.1341650084187969, "grad_norm": 2.128838368057253, "learning_rate": 9.774854483979294e-06, "loss": 1.3385, "step": 1753 }, { "epoch": 0.13424154293586407, "grad_norm": 1.173316980184116, "learning_rate": 9.77452641324072e-06, "loss": 1.4811, "step": 1754 }, { "epoch": 0.13431807745293128, "grad_norm": 0.665894379610012, "learning_rate": 9.774198110583956e-06, "loss": 1.7039, "step": 1755 }, { "epoch": 0.13439461196999847, "grad_norm": 0.6931841700583646, "learning_rate": 9.77386957602918e-06, "loss": 1.6791, "step": 1756 }, { "epoch": 0.13447114648706568, "grad_norm": 0.490353560798915, "learning_rate": 9.773540809596576e-06, "loss": 0.8894, "step": 1757 }, { "epoch": 0.13454768100413286, "grad_norm": 0.6014873277040889, "learning_rate": 9.773211811306343e-06, "loss": 1.3088, "step": 1758 }, { "epoch": 0.13462421552120007, "grad_norm": 0.5266084227069793, "learning_rate": 9.772882581178695e-06, "loss": 1.1707, "step": 1759 }, { "epoch": 0.13470075003826726, "grad_norm": 0.670536119474168, "learning_rate": 9.772553119233865e-06, "loss": 1.4163, "step": 1760 }, { "epoch": 0.13477728455533444, "grad_norm": 0.6303860068770771, "learning_rate": 9.77222342549209e-06, "loss": 1.5849, "step": 1761 }, { "epoch": 0.13485381907240165, "grad_norm": 0.8702341887934134, "learning_rate": 9.771893499973635e-06, "loss": 1.9301, "step": 1762 }, { "epoch": 0.13493035358946884, "grad_norm": 0.5882825324634757, "learning_rate": 9.771563342698765e-06, "loss": 1.4921, "step": 1763 }, { "epoch": 0.13500688810653605, "grad_norm": 0.6528464465702654, "learning_rate": 9.77123295368777e-06, "loss": 1.461, "step": 1764 }, { "epoch": 0.13508342262360323, "grad_norm": 0.6495357190990135, "learning_rate": 9.770902332960947e-06, "loss": 1.537, "step": 1765 }, { "epoch": 0.13515995714067044, "grad_norm": 0.6911026139949368, "learning_rate": 9.770571480538613e-06, "loss": 1.6073, "step": 1766 }, { "epoch": 0.13523649165773763, "grad_norm": 0.5929414538620699, "learning_rate": 9.770240396441095e-06, "loss": 1.3236, "step": 1767 }, { "epoch": 0.13531302617480484, "grad_norm": 0.5754872243458116, "learning_rate": 9.769909080688737e-06, "loss": 1.408, "step": 1768 }, { "epoch": 0.13538956069187202, "grad_norm": 0.6981691115171125, "learning_rate": 9.769577533301894e-06, "loss": 1.285, "step": 1769 }, { "epoch": 0.13546609520893924, "grad_norm": 0.7684436871922978, "learning_rate": 9.769245754300938e-06, "loss": 1.8455, "step": 1770 }, { "epoch": 0.13554262972600642, "grad_norm": 0.5695174429124604, "learning_rate": 9.768913743706257e-06, "loss": 1.0722, "step": 1771 }, { "epoch": 0.13561916424307363, "grad_norm": 0.5922481454628383, "learning_rate": 9.768581501538247e-06, "loss": 1.2741, "step": 1772 }, { "epoch": 0.13569569876014081, "grad_norm": 0.7394723762848676, "learning_rate": 9.768249027817324e-06, "loss": 1.5789, "step": 1773 }, { "epoch": 0.13577223327720803, "grad_norm": 0.7107061316117694, "learning_rate": 9.767916322563915e-06, "loss": 1.7488, "step": 1774 }, { "epoch": 0.1358487677942752, "grad_norm": 0.5719230617626844, "learning_rate": 9.767583385798465e-06, "loss": 1.4152, "step": 1775 }, { "epoch": 0.13592530231134242, "grad_norm": 0.5340191181399052, "learning_rate": 9.76725021754143e-06, "loss": 1.1001, "step": 1776 }, { "epoch": 0.1360018368284096, "grad_norm": 0.5677281364799233, "learning_rate": 9.766916817813275e-06, "loss": 1.4376, "step": 1777 }, { "epoch": 0.13607837134547682, "grad_norm": 0.6011091285400357, "learning_rate": 9.766583186634493e-06, "loss": 1.1604, "step": 1778 }, { "epoch": 0.136154905862544, "grad_norm": 0.639662078976125, "learning_rate": 9.766249324025578e-06, "loss": 1.4228, "step": 1779 }, { "epoch": 0.1362314403796112, "grad_norm": 0.7928516701552885, "learning_rate": 9.765915230007049e-06, "loss": 1.7932, "step": 1780 }, { "epoch": 0.1363079748966784, "grad_norm": 0.6874896859478743, "learning_rate": 9.765580904599428e-06, "loss": 1.6963, "step": 1781 }, { "epoch": 0.1363845094137456, "grad_norm": 0.5929444729319647, "learning_rate": 9.765246347823258e-06, "loss": 1.4225, "step": 1782 }, { "epoch": 0.1364610439308128, "grad_norm": 0.57588731258687, "learning_rate": 9.764911559699099e-06, "loss": 1.3102, "step": 1783 }, { "epoch": 0.13653757844788, "grad_norm": 0.5544717887112462, "learning_rate": 9.76457654024752e-06, "loss": 1.302, "step": 1784 }, { "epoch": 0.1366141129649472, "grad_norm": 0.631803738966022, "learning_rate": 9.7642412894891e-06, "loss": 1.5302, "step": 1785 }, { "epoch": 0.1366906474820144, "grad_norm": 0.5533153506821412, "learning_rate": 9.763905807444445e-06, "loss": 1.26, "step": 1786 }, { "epoch": 0.13676718199908158, "grad_norm": 0.5979082100514258, "learning_rate": 9.763570094134165e-06, "loss": 1.3798, "step": 1787 }, { "epoch": 0.1368437165161488, "grad_norm": 0.5755071416196664, "learning_rate": 9.763234149578888e-06, "loss": 1.3696, "step": 1788 }, { "epoch": 0.13692025103321598, "grad_norm": 0.6283559091426695, "learning_rate": 9.762897973799254e-06, "loss": 1.4708, "step": 1789 }, { "epoch": 0.1369967855502832, "grad_norm": 0.6170801099818679, "learning_rate": 9.762561566815919e-06, "loss": 1.5157, "step": 1790 }, { "epoch": 0.13707332006735037, "grad_norm": 0.6015112401700649, "learning_rate": 9.762224928649554e-06, "loss": 1.492, "step": 1791 }, { "epoch": 0.13714985458441759, "grad_norm": 0.8318697591794477, "learning_rate": 9.761888059320844e-06, "loss": 1.8514, "step": 1792 }, { "epoch": 0.13722638910148477, "grad_norm": 0.533590208265388, "learning_rate": 9.761550958850485e-06, "loss": 1.2369, "step": 1793 }, { "epoch": 0.13730292361855198, "grad_norm": 0.6016018240583153, "learning_rate": 9.761213627259188e-06, "loss": 1.2999, "step": 1794 }, { "epoch": 0.13737945813561916, "grad_norm": 0.5849889298981505, "learning_rate": 9.760876064567684e-06, "loss": 1.3235, "step": 1795 }, { "epoch": 0.13745599265268635, "grad_norm": 0.7878902481936423, "learning_rate": 9.76053827079671e-06, "loss": 1.5958, "step": 1796 }, { "epoch": 0.13753252716975356, "grad_norm": 0.6838990731185814, "learning_rate": 9.760200245967025e-06, "loss": 1.6838, "step": 1797 }, { "epoch": 0.13760906168682074, "grad_norm": 0.63068840403818, "learning_rate": 9.759861990099393e-06, "loss": 1.6761, "step": 1798 }, { "epoch": 0.13768559620388796, "grad_norm": 2.5729557386467232, "learning_rate": 9.759523503214605e-06, "loss": 1.8635, "step": 1799 }, { "epoch": 0.13776213072095514, "grad_norm": 0.7130600664092502, "learning_rate": 9.75918478533345e-06, "loss": 1.7662, "step": 1800 }, { "epoch": 0.13783866523802235, "grad_norm": 0.5724870940352995, "learning_rate": 9.758845836476747e-06, "loss": 1.4685, "step": 1801 }, { "epoch": 0.13791519975508953, "grad_norm": 0.6151513051305094, "learning_rate": 9.758506656665318e-06, "loss": 1.6431, "step": 1802 }, { "epoch": 0.13799173427215675, "grad_norm": 0.5015619108078297, "learning_rate": 9.758167245920003e-06, "loss": 1.0848, "step": 1803 }, { "epoch": 0.13806826878922393, "grad_norm": 0.7234886563144337, "learning_rate": 9.757827604261658e-06, "loss": 1.7551, "step": 1804 }, { "epoch": 0.13814480330629114, "grad_norm": 0.5500481110018156, "learning_rate": 9.757487731711153e-06, "loss": 1.4249, "step": 1805 }, { "epoch": 0.13822133782335833, "grad_norm": 0.5618819375636606, "learning_rate": 9.757147628289366e-06, "loss": 1.3618, "step": 1806 }, { "epoch": 0.13829787234042554, "grad_norm": 0.5605208983027671, "learning_rate": 9.756807294017198e-06, "loss": 1.4218, "step": 1807 }, { "epoch": 0.13837440685749272, "grad_norm": 0.5820306920062677, "learning_rate": 9.75646672891556e-06, "loss": 1.3364, "step": 1808 }, { "epoch": 0.13845094137455993, "grad_norm": 0.5468412660452853, "learning_rate": 9.756125933005377e-06, "loss": 1.2142, "step": 1809 }, { "epoch": 0.13852747589162712, "grad_norm": 1.8146038226407382, "learning_rate": 9.755784906307586e-06, "loss": 1.5419, "step": 1810 }, { "epoch": 0.13860401040869433, "grad_norm": 0.64481889699959, "learning_rate": 9.755443648843144e-06, "loss": 1.5397, "step": 1811 }, { "epoch": 0.1386805449257615, "grad_norm": 0.6537937027504256, "learning_rate": 9.755102160633018e-06, "loss": 1.6421, "step": 1812 }, { "epoch": 0.13875707944282872, "grad_norm": 0.7146351333308032, "learning_rate": 9.75476044169819e-06, "loss": 1.5342, "step": 1813 }, { "epoch": 0.1388336139598959, "grad_norm": 0.7585136886174634, "learning_rate": 9.754418492059656e-06, "loss": 1.451, "step": 1814 }, { "epoch": 0.13891014847696312, "grad_norm": 0.6570213545101555, "learning_rate": 9.754076311738424e-06, "loss": 1.6073, "step": 1815 }, { "epoch": 0.1389866829940303, "grad_norm": 0.6211604900020713, "learning_rate": 9.753733900755524e-06, "loss": 1.5513, "step": 1816 }, { "epoch": 0.13906321751109751, "grad_norm": 0.7820284375121876, "learning_rate": 9.753391259131987e-06, "loss": 1.7163, "step": 1817 }, { "epoch": 0.1391397520281647, "grad_norm": 0.5767792500292388, "learning_rate": 9.753048386888874e-06, "loss": 1.3129, "step": 1818 }, { "epoch": 0.1392162865452319, "grad_norm": 0.7812701266094855, "learning_rate": 9.752705284047248e-06, "loss": 1.3951, "step": 1819 }, { "epoch": 0.1392928210622991, "grad_norm": 0.7172484561927592, "learning_rate": 9.75236195062819e-06, "loss": 1.6209, "step": 1820 }, { "epoch": 0.1393693555793663, "grad_norm": 0.5754159059068238, "learning_rate": 9.752018386652799e-06, "loss": 1.4425, "step": 1821 }, { "epoch": 0.1394458900964335, "grad_norm": 0.46175471257467604, "learning_rate": 9.751674592142179e-06, "loss": 0.8518, "step": 1822 }, { "epoch": 0.1395224246135007, "grad_norm": 0.6761933954447672, "learning_rate": 9.751330567117459e-06, "loss": 1.3063, "step": 1823 }, { "epoch": 0.13959895913056788, "grad_norm": 0.6470728823450005, "learning_rate": 9.75098631159977e-06, "loss": 1.3903, "step": 1824 }, { "epoch": 0.1396754936476351, "grad_norm": 0.5484279290982088, "learning_rate": 9.750641825610274e-06, "loss": 1.2371, "step": 1825 }, { "epoch": 0.13975202816470228, "grad_norm": 0.6389240090238062, "learning_rate": 9.75029710917013e-06, "loss": 1.311, "step": 1826 }, { "epoch": 0.1398285626817695, "grad_norm": 0.6489258650385739, "learning_rate": 9.749952162300522e-06, "loss": 1.3027, "step": 1827 }, { "epoch": 0.13990509719883668, "grad_norm": 0.6009394490488742, "learning_rate": 9.749606985022641e-06, "loss": 1.3273, "step": 1828 }, { "epoch": 0.13998163171590386, "grad_norm": 0.6200096665858778, "learning_rate": 9.749261577357698e-06, "loss": 1.4442, "step": 1829 }, { "epoch": 0.14005816623297107, "grad_norm": 0.550218128888133, "learning_rate": 9.748915939326914e-06, "loss": 1.3424, "step": 1830 }, { "epoch": 0.14013470075003825, "grad_norm": 0.5914871446329695, "learning_rate": 9.748570070951528e-06, "loss": 1.3312, "step": 1831 }, { "epoch": 0.14021123526710547, "grad_norm": 0.5581037931436308, "learning_rate": 9.748223972252793e-06, "loss": 1.3409, "step": 1832 }, { "epoch": 0.14028776978417265, "grad_norm": 0.6086840523734962, "learning_rate": 9.747877643251971e-06, "loss": 1.4402, "step": 1833 }, { "epoch": 0.14036430430123986, "grad_norm": 0.6606517523132251, "learning_rate": 9.747531083970343e-06, "loss": 1.6112, "step": 1834 }, { "epoch": 0.14044083881830705, "grad_norm": 0.6543050257417866, "learning_rate": 9.747184294429201e-06, "loss": 1.6471, "step": 1835 }, { "epoch": 0.14051737333537426, "grad_norm": 0.5610477085974692, "learning_rate": 9.746837274649855e-06, "loss": 1.1945, "step": 1836 }, { "epoch": 0.14059390785244144, "grad_norm": 0.6101113937432895, "learning_rate": 9.746490024653626e-06, "loss": 1.6392, "step": 1837 }, { "epoch": 0.14067044236950865, "grad_norm": 0.6030808501718775, "learning_rate": 9.74614254446185e-06, "loss": 1.4195, "step": 1838 }, { "epoch": 0.14074697688657584, "grad_norm": 0.8231809132381233, "learning_rate": 9.745794834095877e-06, "loss": 1.5798, "step": 1839 }, { "epoch": 0.14082351140364305, "grad_norm": 0.5756540056325996, "learning_rate": 9.745446893577071e-06, "loss": 1.3705, "step": 1840 }, { "epoch": 0.14090004592071023, "grad_norm": 0.5453659066082354, "learning_rate": 9.745098722926812e-06, "loss": 1.3041, "step": 1841 }, { "epoch": 0.14097658043777744, "grad_norm": 0.7452359345387107, "learning_rate": 9.744750322166491e-06, "loss": 1.3198, "step": 1842 }, { "epoch": 0.14105311495484463, "grad_norm": 0.5499501307787612, "learning_rate": 9.744401691317516e-06, "loss": 1.3136, "step": 1843 }, { "epoch": 0.14112964947191184, "grad_norm": 0.5165942837239237, "learning_rate": 9.744052830401307e-06, "loss": 1.2602, "step": 1844 }, { "epoch": 0.14120618398897902, "grad_norm": 0.7053195863275332, "learning_rate": 9.7437037394393e-06, "loss": 1.8412, "step": 1845 }, { "epoch": 0.14128271850604623, "grad_norm": 0.5904171271284966, "learning_rate": 9.743354418452943e-06, "loss": 1.4671, "step": 1846 }, { "epoch": 0.14135925302311342, "grad_norm": 0.5208390359374944, "learning_rate": 9.743004867463701e-06, "loss": 1.1942, "step": 1847 }, { "epoch": 0.14143578754018063, "grad_norm": 0.5465770819622361, "learning_rate": 9.74265508649305e-06, "loss": 1.23, "step": 1848 }, { "epoch": 0.1415123220572478, "grad_norm": 0.6270783346923061, "learning_rate": 9.742305075562481e-06, "loss": 1.7252, "step": 1849 }, { "epoch": 0.14158885657431503, "grad_norm": 0.6442815314465794, "learning_rate": 9.741954834693502e-06, "loss": 1.5772, "step": 1850 }, { "epoch": 0.1416653910913822, "grad_norm": 0.6169535299401562, "learning_rate": 9.741604363907634e-06, "loss": 1.5974, "step": 1851 }, { "epoch": 0.14174192560844942, "grad_norm": 0.6705855290702991, "learning_rate": 9.741253663226404e-06, "loss": 1.2823, "step": 1852 }, { "epoch": 0.1418184601255166, "grad_norm": 0.746982720628255, "learning_rate": 9.740902732671367e-06, "loss": 1.6367, "step": 1853 }, { "epoch": 0.14189499464258382, "grad_norm": 1.1688266335659698, "learning_rate": 9.740551572264082e-06, "loss": 1.4763, "step": 1854 }, { "epoch": 0.141971529159651, "grad_norm": 1.1047198480341796, "learning_rate": 9.740200182026129e-06, "loss": 1.8504, "step": 1855 }, { "epoch": 0.1420480636767182, "grad_norm": 0.8107955838255845, "learning_rate": 9.739848561979094e-06, "loss": 1.5736, "step": 1856 }, { "epoch": 0.1421245981937854, "grad_norm": 1.0220281209421818, "learning_rate": 9.739496712144581e-06, "loss": 1.6445, "step": 1857 }, { "epoch": 0.1422011327108526, "grad_norm": 0.5811717752047626, "learning_rate": 9.739144632544215e-06, "loss": 1.4106, "step": 1858 }, { "epoch": 0.1422776672279198, "grad_norm": 0.6776842401545342, "learning_rate": 9.738792323199623e-06, "loss": 1.347, "step": 1859 }, { "epoch": 0.142354201744987, "grad_norm": 0.6420621600699944, "learning_rate": 9.738439784132455e-06, "loss": 1.5728, "step": 1860 }, { "epoch": 0.1424307362620542, "grad_norm": 0.5165869793397, "learning_rate": 9.738087015364368e-06, "loss": 1.0635, "step": 1861 }, { "epoch": 0.14250727077912137, "grad_norm": 0.5131992966368818, "learning_rate": 9.737734016917042e-06, "loss": 1.0098, "step": 1862 }, { "epoch": 0.14258380529618858, "grad_norm": 0.6048085354287801, "learning_rate": 9.737380788812163e-06, "loss": 1.3748, "step": 1863 }, { "epoch": 0.14266033981325577, "grad_norm": 0.5513315275693013, "learning_rate": 9.737027331071437e-06, "loss": 1.2885, "step": 1864 }, { "epoch": 0.14273687433032298, "grad_norm": 1.0349961051552854, "learning_rate": 9.73667364371658e-06, "loss": 1.8041, "step": 1865 }, { "epoch": 0.14281340884739016, "grad_norm": 0.5777036168646804, "learning_rate": 9.736319726769323e-06, "loss": 1.4476, "step": 1866 }, { "epoch": 0.14288994336445737, "grad_norm": 0.6882375632251065, "learning_rate": 9.735965580251413e-06, "loss": 1.6823, "step": 1867 }, { "epoch": 0.14296647788152456, "grad_norm": 0.9351287670310884, "learning_rate": 9.73561120418461e-06, "loss": 1.4838, "step": 1868 }, { "epoch": 0.14304301239859177, "grad_norm": 0.567003756363207, "learning_rate": 9.735256598590687e-06, "loss": 1.3547, "step": 1869 }, { "epoch": 0.14311954691565895, "grad_norm": 0.5854492204553223, "learning_rate": 9.734901763491433e-06, "loss": 1.447, "step": 1870 }, { "epoch": 0.14319608143272616, "grad_norm": 0.5026886729373478, "learning_rate": 9.734546698908647e-06, "loss": 1.133, "step": 1871 }, { "epoch": 0.14327261594979335, "grad_norm": 0.7867554374282179, "learning_rate": 9.73419140486415e-06, "loss": 1.6393, "step": 1872 }, { "epoch": 0.14334915046686056, "grad_norm": 0.6108868143946167, "learning_rate": 9.73383588137977e-06, "loss": 1.5027, "step": 1873 }, { "epoch": 0.14342568498392774, "grad_norm": 0.60246522659747, "learning_rate": 9.73348012847735e-06, "loss": 1.3936, "step": 1874 }, { "epoch": 0.14350221950099495, "grad_norm": 0.5557805601898373, "learning_rate": 9.733124146178751e-06, "loss": 1.326, "step": 1875 }, { "epoch": 0.14357875401806214, "grad_norm": 0.6329634568473929, "learning_rate": 9.732767934505844e-06, "loss": 1.3925, "step": 1876 }, { "epoch": 0.14365528853512935, "grad_norm": 0.6293660144510751, "learning_rate": 9.732411493480516e-06, "loss": 1.3732, "step": 1877 }, { "epoch": 0.14373182305219653, "grad_norm": 0.5375246660938389, "learning_rate": 9.732054823124669e-06, "loss": 1.1808, "step": 1878 }, { "epoch": 0.14380835756926375, "grad_norm": 0.5616051793953026, "learning_rate": 9.731697923460216e-06, "loss": 1.3278, "step": 1879 }, { "epoch": 0.14388489208633093, "grad_norm": 0.49744658922106166, "learning_rate": 9.731340794509087e-06, "loss": 1.1148, "step": 1880 }, { "epoch": 0.14396142660339814, "grad_norm": 0.5738337426034323, "learning_rate": 9.730983436293224e-06, "loss": 1.3797, "step": 1881 }, { "epoch": 0.14403796112046532, "grad_norm": 0.5876606466412088, "learning_rate": 9.730625848834588e-06, "loss": 1.6241, "step": 1882 }, { "epoch": 0.14411449563753254, "grad_norm": 0.58020968137252, "learning_rate": 9.730268032155146e-06, "loss": 1.232, "step": 1883 }, { "epoch": 0.14419103015459972, "grad_norm": 0.670174796561004, "learning_rate": 9.729909986276884e-06, "loss": 1.6152, "step": 1884 }, { "epoch": 0.14426756467166693, "grad_norm": 0.5670153694925809, "learning_rate": 9.729551711221802e-06, "loss": 1.417, "step": 1885 }, { "epoch": 0.14434409918873412, "grad_norm": 0.5426942867500876, "learning_rate": 9.729193207011914e-06, "loss": 1.2, "step": 1886 }, { "epoch": 0.14442063370580133, "grad_norm": 0.6232707726668014, "learning_rate": 9.728834473669249e-06, "loss": 1.4471, "step": 1887 }, { "epoch": 0.1444971682228685, "grad_norm": 0.5856300817746, "learning_rate": 9.728475511215844e-06, "loss": 1.3693, "step": 1888 }, { "epoch": 0.14457370273993572, "grad_norm": 0.622839247986077, "learning_rate": 9.728116319673761e-06, "loss": 1.4761, "step": 1889 }, { "epoch": 0.1446502372570029, "grad_norm": 0.5326041048713795, "learning_rate": 9.727756899065063e-06, "loss": 1.1118, "step": 1890 }, { "epoch": 0.14472677177407012, "grad_norm": 0.48678802028678364, "learning_rate": 9.72739724941184e-06, "loss": 1.0297, "step": 1891 }, { "epoch": 0.1448033062911373, "grad_norm": 0.8683791140406224, "learning_rate": 9.727037370736185e-06, "loss": 1.7433, "step": 1892 }, { "epoch": 0.1448798408082045, "grad_norm": 0.5245723126499404, "learning_rate": 9.726677263060212e-06, "loss": 1.0729, "step": 1893 }, { "epoch": 0.1449563753252717, "grad_norm": 0.7209153362224263, "learning_rate": 9.726316926406049e-06, "loss": 1.7215, "step": 1894 }, { "epoch": 0.1450329098423389, "grad_norm": 0.578239183216676, "learning_rate": 9.725956360795833e-06, "loss": 1.2038, "step": 1895 }, { "epoch": 0.1451094443594061, "grad_norm": 0.5398675461791357, "learning_rate": 9.72559556625172e-06, "loss": 1.1864, "step": 1896 }, { "epoch": 0.14518597887647328, "grad_norm": 0.6216784654342408, "learning_rate": 9.725234542795878e-06, "loss": 1.485, "step": 1897 }, { "epoch": 0.1452625133935405, "grad_norm": 0.7892785183615544, "learning_rate": 9.724873290450491e-06, "loss": 1.2868, "step": 1898 }, { "epoch": 0.14533904791060767, "grad_norm": 0.614182675078787, "learning_rate": 9.724511809237753e-06, "loss": 1.5228, "step": 1899 }, { "epoch": 0.14541558242767488, "grad_norm": 0.5512564232241616, "learning_rate": 9.724150099179875e-06, "loss": 1.28, "step": 1900 }, { "epoch": 0.14549211694474207, "grad_norm": 0.6354264310068818, "learning_rate": 9.72378816029908e-06, "loss": 1.5387, "step": 1901 }, { "epoch": 0.14556865146180928, "grad_norm": 0.9180611929083063, "learning_rate": 9.723425992617612e-06, "loss": 1.486, "step": 1902 }, { "epoch": 0.14564518597887646, "grad_norm": 0.5687228052901117, "learning_rate": 9.723063596157718e-06, "loss": 1.2003, "step": 1903 }, { "epoch": 0.14572172049594367, "grad_norm": 0.6724713679579224, "learning_rate": 9.722700970941666e-06, "loss": 1.5943, "step": 1904 }, { "epoch": 0.14579825501301086, "grad_norm": 0.6289030367947234, "learning_rate": 9.722338116991739e-06, "loss": 1.4225, "step": 1905 }, { "epoch": 0.14587478953007807, "grad_norm": 0.5528289448750581, "learning_rate": 9.72197503433023e-06, "loss": 1.2302, "step": 1906 }, { "epoch": 0.14595132404714525, "grad_norm": 0.5731750475583955, "learning_rate": 9.721611722979448e-06, "loss": 1.347, "step": 1907 }, { "epoch": 0.14602785856421246, "grad_norm": 0.5622950456431637, "learning_rate": 9.721248182961716e-06, "loss": 1.2282, "step": 1908 }, { "epoch": 0.14610439308127965, "grad_norm": 1.5259177200142238, "learning_rate": 9.720884414299373e-06, "loss": 1.6779, "step": 1909 }, { "epoch": 0.14618092759834686, "grad_norm": 0.5832054414759622, "learning_rate": 9.720520417014767e-06, "loss": 1.3215, "step": 1910 }, { "epoch": 0.14625746211541404, "grad_norm": 0.6658913528095752, "learning_rate": 9.720156191130264e-06, "loss": 1.6153, "step": 1911 }, { "epoch": 0.14633399663248126, "grad_norm": 0.8988458211174036, "learning_rate": 9.719791736668244e-06, "loss": 1.7335, "step": 1912 }, { "epoch": 0.14641053114954844, "grad_norm": 0.7888383068233573, "learning_rate": 9.7194270536511e-06, "loss": 1.5198, "step": 1913 }, { "epoch": 0.14648706566661565, "grad_norm": 0.8025975074432056, "learning_rate": 9.719062142101238e-06, "loss": 1.286, "step": 1914 }, { "epoch": 0.14656360018368284, "grad_norm": 0.5928772451973213, "learning_rate": 9.718697002041079e-06, "loss": 1.4204, "step": 1915 }, { "epoch": 0.14664013470075005, "grad_norm": 0.85917842142426, "learning_rate": 9.718331633493061e-06, "loss": 1.5732, "step": 1916 }, { "epoch": 0.14671666921781723, "grad_norm": 0.5537256419918672, "learning_rate": 9.717966036479633e-06, "loss": 1.2449, "step": 1917 }, { "epoch": 0.14679320373488444, "grad_norm": 0.6077859511641811, "learning_rate": 9.717600211023254e-06, "loss": 1.3517, "step": 1918 }, { "epoch": 0.14686973825195163, "grad_norm": 0.5642509624505971, "learning_rate": 9.717234157146406e-06, "loss": 1.3145, "step": 1919 }, { "epoch": 0.14694627276901884, "grad_norm": 0.562193000063314, "learning_rate": 9.716867874871581e-06, "loss": 1.326, "step": 1920 }, { "epoch": 0.14702280728608602, "grad_norm": 0.5652451950381666, "learning_rate": 9.716501364221281e-06, "loss": 1.3633, "step": 1921 }, { "epoch": 0.14709934180315323, "grad_norm": 0.6764855901714417, "learning_rate": 9.716134625218025e-06, "loss": 1.6519, "step": 1922 }, { "epoch": 0.14717587632022042, "grad_norm": 0.5567058029840992, "learning_rate": 9.71576765788435e-06, "loss": 1.3046, "step": 1923 }, { "epoch": 0.14725241083728763, "grad_norm": 0.6118266168503794, "learning_rate": 9.715400462242803e-06, "loss": 1.5447, "step": 1924 }, { "epoch": 0.1473289453543548, "grad_norm": 0.5924857554557917, "learning_rate": 9.715033038315943e-06, "loss": 1.2824, "step": 1925 }, { "epoch": 0.14740547987142202, "grad_norm": 0.6068009908787203, "learning_rate": 9.714665386126348e-06, "loss": 1.4662, "step": 1926 }, { "epoch": 0.1474820143884892, "grad_norm": 0.5755234616598427, "learning_rate": 9.714297505696608e-06, "loss": 1.4556, "step": 1927 }, { "epoch": 0.14755854890555642, "grad_norm": 0.9155046438783141, "learning_rate": 9.713929397049325e-06, "loss": 1.5426, "step": 1928 }, { "epoch": 0.1476350834226236, "grad_norm": 0.7759059243574667, "learning_rate": 9.713561060207117e-06, "loss": 1.732, "step": 1929 }, { "epoch": 0.1477116179396908, "grad_norm": 0.5823261469198585, "learning_rate": 9.713192495192615e-06, "loss": 1.3428, "step": 1930 }, { "epoch": 0.147788152456758, "grad_norm": 0.6212505284538713, "learning_rate": 9.712823702028466e-06, "loss": 1.4992, "step": 1931 }, { "epoch": 0.14786468697382518, "grad_norm": 0.5549202688371869, "learning_rate": 9.712454680737332e-06, "loss": 1.4122, "step": 1932 }, { "epoch": 0.1479412214908924, "grad_norm": 0.5688245322424907, "learning_rate": 9.712085431341883e-06, "loss": 1.2257, "step": 1933 }, { "epoch": 0.14801775600795958, "grad_norm": 0.5981036230668364, "learning_rate": 9.711715953864807e-06, "loss": 1.5841, "step": 1934 }, { "epoch": 0.1480942905250268, "grad_norm": 0.6276016174453802, "learning_rate": 9.711346248328809e-06, "loss": 1.4366, "step": 1935 }, { "epoch": 0.14817082504209397, "grad_norm": 0.6590626949467525, "learning_rate": 9.7109763147566e-06, "loss": 1.642, "step": 1936 }, { "epoch": 0.14824735955916118, "grad_norm": 0.6109304277203546, "learning_rate": 9.710606153170914e-06, "loss": 1.5825, "step": 1937 }, { "epoch": 0.14832389407622837, "grad_norm": 0.5986133111353931, "learning_rate": 9.710235763594493e-06, "loss": 1.6026, "step": 1938 }, { "epoch": 0.14840042859329558, "grad_norm": 0.823216982732674, "learning_rate": 9.709865146050096e-06, "loss": 1.6693, "step": 1939 }, { "epoch": 0.14847696311036276, "grad_norm": 0.6060171662794713, "learning_rate": 9.709494300560493e-06, "loss": 1.3271, "step": 1940 }, { "epoch": 0.14855349762742998, "grad_norm": 0.5378536202469429, "learning_rate": 9.70912322714847e-06, "loss": 1.1668, "step": 1941 }, { "epoch": 0.14863003214449716, "grad_norm": 0.5479937177606711, "learning_rate": 9.70875192583683e-06, "loss": 1.346, "step": 1942 }, { "epoch": 0.14870656666156437, "grad_norm": 0.5090152569870372, "learning_rate": 9.708380396648382e-06, "loss": 1.1295, "step": 1943 }, { "epoch": 0.14878310117863155, "grad_norm": 0.6016941388549318, "learning_rate": 9.708008639605959e-06, "loss": 1.3983, "step": 1944 }, { "epoch": 0.14885963569569877, "grad_norm": 0.5554999231110906, "learning_rate": 9.707636654732399e-06, "loss": 1.2331, "step": 1945 }, { "epoch": 0.14893617021276595, "grad_norm": 0.4882131474440155, "learning_rate": 9.70726444205056e-06, "loss": 1.0095, "step": 1946 }, { "epoch": 0.14901270472983316, "grad_norm": 0.6497259105228209, "learning_rate": 9.70689200158331e-06, "loss": 1.7066, "step": 1947 }, { "epoch": 0.14908923924690035, "grad_norm": 0.647100855330045, "learning_rate": 9.706519333353534e-06, "loss": 1.5876, "step": 1948 }, { "epoch": 0.14916577376396756, "grad_norm": 0.6306320738393186, "learning_rate": 9.706146437384132e-06, "loss": 1.5957, "step": 1949 }, { "epoch": 0.14924230828103474, "grad_norm": 1.2006533203346519, "learning_rate": 9.705773313698012e-06, "loss": 1.5583, "step": 1950 }, { "epoch": 0.14931884279810195, "grad_norm": 0.7017514793604008, "learning_rate": 9.705399962318101e-06, "loss": 1.4846, "step": 1951 }, { "epoch": 0.14939537731516914, "grad_norm": 0.7682404302405849, "learning_rate": 9.70502638326734e-06, "loss": 2.0002, "step": 1952 }, { "epoch": 0.14947191183223635, "grad_norm": 0.575947186414124, "learning_rate": 9.70465257656868e-06, "loss": 1.3653, "step": 1953 }, { "epoch": 0.14954844634930353, "grad_norm": 0.5774863902734333, "learning_rate": 9.704278542245095e-06, "loss": 1.3808, "step": 1954 }, { "epoch": 0.14962498086637074, "grad_norm": 0.5266052851803157, "learning_rate": 9.70390428031956e-06, "loss": 1.1031, "step": 1955 }, { "epoch": 0.14970151538343793, "grad_norm": 1.2181899990832152, "learning_rate": 9.703529790815075e-06, "loss": 1.4378, "step": 1956 }, { "epoch": 0.14977804990050514, "grad_norm": 0.54011354694433, "learning_rate": 9.703155073754646e-06, "loss": 1.2042, "step": 1957 }, { "epoch": 0.14985458441757232, "grad_norm": 0.6920097255841935, "learning_rate": 9.7027801291613e-06, "loss": 1.7324, "step": 1958 }, { "epoch": 0.14993111893463953, "grad_norm": 0.6563495164490105, "learning_rate": 9.702404957058073e-06, "loss": 1.693, "step": 1959 }, { "epoch": 0.15000765345170672, "grad_norm": 0.6011902303993664, "learning_rate": 9.70202955746802e-06, "loss": 1.5276, "step": 1960 }, { "epoch": 0.15008418796877393, "grad_norm": 0.509386107563187, "learning_rate": 9.7016539304142e-06, "loss": 0.9617, "step": 1961 }, { "epoch": 0.1501607224858411, "grad_norm": 0.5475153416927626, "learning_rate": 9.701278075919698e-06, "loss": 1.3956, "step": 1962 }, { "epoch": 0.15023725700290833, "grad_norm": 0.5435405772689917, "learning_rate": 9.700901994007607e-06, "loss": 1.3677, "step": 1963 }, { "epoch": 0.1503137915199755, "grad_norm": 0.6744962609238891, "learning_rate": 9.700525684701036e-06, "loss": 1.405, "step": 1964 }, { "epoch": 0.1503903260370427, "grad_norm": 0.669488422907934, "learning_rate": 9.700149148023102e-06, "loss": 1.3049, "step": 1965 }, { "epoch": 0.1504668605541099, "grad_norm": 0.600826193891667, "learning_rate": 9.699772383996942e-06, "loss": 1.6228, "step": 1966 }, { "epoch": 0.1505433950711771, "grad_norm": 0.587955079502142, "learning_rate": 9.699395392645708e-06, "loss": 1.2322, "step": 1967 }, { "epoch": 0.1506199295882443, "grad_norm": 0.4741381256012691, "learning_rate": 9.699018173992562e-06, "loss": 0.8997, "step": 1968 }, { "epoch": 0.15069646410531148, "grad_norm": 0.5469845372714697, "learning_rate": 9.698640728060682e-06, "loss": 1.3198, "step": 1969 }, { "epoch": 0.1507729986223787, "grad_norm": 0.6318757192336094, "learning_rate": 9.698263054873258e-06, "loss": 1.3361, "step": 1970 }, { "epoch": 0.15084953313944588, "grad_norm": 0.6541639436218893, "learning_rate": 9.697885154453496e-06, "loss": 1.4054, "step": 1971 }, { "epoch": 0.1509260676565131, "grad_norm": 0.6369995911355778, "learning_rate": 9.697507026824616e-06, "loss": 1.5027, "step": 1972 }, { "epoch": 0.15100260217358027, "grad_norm": 0.6914531252888624, "learning_rate": 9.69712867200985e-06, "loss": 1.407, "step": 1973 }, { "epoch": 0.1510791366906475, "grad_norm": 0.586855457300404, "learning_rate": 9.696750090032448e-06, "loss": 1.1924, "step": 1974 }, { "epoch": 0.15115567120771467, "grad_norm": 0.518818506876326, "learning_rate": 9.696371280915668e-06, "loss": 1.1717, "step": 1975 }, { "epoch": 0.15123220572478188, "grad_norm": 0.7824051077958654, "learning_rate": 9.695992244682787e-06, "loss": 1.4704, "step": 1976 }, { "epoch": 0.15130874024184907, "grad_norm": 0.5201991279897764, "learning_rate": 9.695612981357093e-06, "loss": 1.1501, "step": 1977 }, { "epoch": 0.15138527475891628, "grad_norm": 0.6486615514453703, "learning_rate": 9.695233490961889e-06, "loss": 1.4795, "step": 1978 }, { "epoch": 0.15146180927598346, "grad_norm": 0.5316974878936969, "learning_rate": 9.694853773520493e-06, "loss": 1.1786, "step": 1979 }, { "epoch": 0.15153834379305067, "grad_norm": 0.7793764163836366, "learning_rate": 9.694473829056236e-06, "loss": 1.6021, "step": 1980 }, { "epoch": 0.15161487831011786, "grad_norm": 0.5654113347267593, "learning_rate": 9.694093657592463e-06, "loss": 1.4928, "step": 1981 }, { "epoch": 0.15169141282718507, "grad_norm": 0.5834609548560411, "learning_rate": 9.693713259152533e-06, "loss": 1.4388, "step": 1982 }, { "epoch": 0.15176794734425225, "grad_norm": 0.634271475250113, "learning_rate": 9.693332633759816e-06, "loss": 1.5231, "step": 1983 }, { "epoch": 0.15184448186131946, "grad_norm": 0.5586894669629794, "learning_rate": 9.692951781437704e-06, "loss": 1.416, "step": 1984 }, { "epoch": 0.15192101637838665, "grad_norm": 0.7014393264216588, "learning_rate": 9.692570702209592e-06, "loss": 1.7492, "step": 1985 }, { "epoch": 0.15199755089545386, "grad_norm": 0.6737286471337692, "learning_rate": 9.692189396098898e-06, "loss": 1.6716, "step": 1986 }, { "epoch": 0.15207408541252104, "grad_norm": 0.7230702373096572, "learning_rate": 9.691807863129052e-06, "loss": 1.7703, "step": 1987 }, { "epoch": 0.15215061992958825, "grad_norm": 0.5866183931908823, "learning_rate": 9.691426103323492e-06, "loss": 1.3435, "step": 1988 }, { "epoch": 0.15222715444665544, "grad_norm": 0.556051221088535, "learning_rate": 9.69104411670568e-06, "loss": 1.3884, "step": 1989 }, { "epoch": 0.15230368896372265, "grad_norm": 0.5780645423189955, "learning_rate": 9.69066190329908e-06, "loss": 1.3585, "step": 1990 }, { "epoch": 0.15238022348078983, "grad_norm": 0.6556717380968704, "learning_rate": 9.69027946312718e-06, "loss": 1.5047, "step": 1991 }, { "epoch": 0.15245675799785705, "grad_norm": 0.5387424725968031, "learning_rate": 9.689896796213479e-06, "loss": 1.3169, "step": 1992 }, { "epoch": 0.15253329251492423, "grad_norm": 0.7459230184006411, "learning_rate": 9.689513902581488e-06, "loss": 1.2588, "step": 1993 }, { "epoch": 0.15260982703199144, "grad_norm": 0.6296495089672941, "learning_rate": 9.689130782254733e-06, "loss": 1.4929, "step": 1994 }, { "epoch": 0.15268636154905862, "grad_norm": 0.6388745293655285, "learning_rate": 9.688747435256756e-06, "loss": 1.6494, "step": 1995 }, { "epoch": 0.15276289606612584, "grad_norm": 0.5366106283153093, "learning_rate": 9.688363861611108e-06, "loss": 1.1218, "step": 1996 }, { "epoch": 0.15283943058319302, "grad_norm": 0.6619518227832588, "learning_rate": 9.687980061341358e-06, "loss": 1.699, "step": 1997 }, { "epoch": 0.1529159651002602, "grad_norm": 0.6096504346541155, "learning_rate": 9.687596034471088e-06, "loss": 1.2582, "step": 1998 }, { "epoch": 0.15299249961732742, "grad_norm": 0.6282123572515631, "learning_rate": 9.687211781023893e-06, "loss": 1.5687, "step": 1999 }, { "epoch": 0.1530690341343946, "grad_norm": 0.6660372143472223, "learning_rate": 9.686827301023386e-06, "loss": 1.5647, "step": 2000 }, { "epoch": 0.1531455686514618, "grad_norm": 0.5198931859200773, "learning_rate": 9.686442594493187e-06, "loss": 1.1659, "step": 2001 }, { "epoch": 0.153222103168529, "grad_norm": 0.5171941953248856, "learning_rate": 9.686057661456932e-06, "loss": 1.2817, "step": 2002 }, { "epoch": 0.1532986376855962, "grad_norm": 0.47294435244057653, "learning_rate": 9.68567250193828e-06, "loss": 1.069, "step": 2003 }, { "epoch": 0.1533751722026634, "grad_norm": 0.5458639746286341, "learning_rate": 9.685287115960887e-06, "loss": 1.3162, "step": 2004 }, { "epoch": 0.1534517067197306, "grad_norm": 0.7108652527298229, "learning_rate": 9.68490150354844e-06, "loss": 1.2636, "step": 2005 }, { "epoch": 0.15352824123679779, "grad_norm": 0.5562599406963092, "learning_rate": 9.684515664724626e-06, "loss": 1.2654, "step": 2006 }, { "epoch": 0.153604775753865, "grad_norm": 0.6999157119732267, "learning_rate": 9.684129599513158e-06, "loss": 1.5612, "step": 2007 }, { "epoch": 0.15368131027093218, "grad_norm": 0.6873960374029626, "learning_rate": 9.68374330793775e-06, "loss": 1.5007, "step": 2008 }, { "epoch": 0.1537578447879994, "grad_norm": 0.564910981096137, "learning_rate": 9.683356790022145e-06, "loss": 1.1685, "step": 2009 }, { "epoch": 0.15383437930506658, "grad_norm": 0.7482245160578842, "learning_rate": 9.682970045790084e-06, "loss": 1.6415, "step": 2010 }, { "epoch": 0.1539109138221338, "grad_norm": 0.554724226193815, "learning_rate": 9.682583075265335e-06, "loss": 1.2828, "step": 2011 }, { "epoch": 0.15398744833920097, "grad_norm": 0.4687518668158583, "learning_rate": 9.682195878471674e-06, "loss": 0.9866, "step": 2012 }, { "epoch": 0.15406398285626818, "grad_norm": 0.6921071072187434, "learning_rate": 9.68180845543289e-06, "loss": 1.6821, "step": 2013 }, { "epoch": 0.15414051737333537, "grad_norm": 0.7074818313470577, "learning_rate": 9.681420806172788e-06, "loss": 1.4218, "step": 2014 }, { "epoch": 0.15421705189040258, "grad_norm": 0.9096614470277059, "learning_rate": 9.681032930715186e-06, "loss": 1.7942, "step": 2015 }, { "epoch": 0.15429358640746976, "grad_norm": 0.525376735701153, "learning_rate": 9.680644829083916e-06, "loss": 1.2189, "step": 2016 }, { "epoch": 0.15437012092453697, "grad_norm": 1.0234481609508639, "learning_rate": 9.680256501302825e-06, "loss": 1.4614, "step": 2017 }, { "epoch": 0.15444665544160416, "grad_norm": 0.6006975107840784, "learning_rate": 9.679867947395774e-06, "loss": 1.4386, "step": 2018 }, { "epoch": 0.15452318995867137, "grad_norm": 0.6959686828998902, "learning_rate": 9.679479167386635e-06, "loss": 1.5257, "step": 2019 }, { "epoch": 0.15459972447573855, "grad_norm": 0.7138534121066049, "learning_rate": 9.679090161299294e-06, "loss": 1.5057, "step": 2020 }, { "epoch": 0.15467625899280577, "grad_norm": 0.7506408538740874, "learning_rate": 9.678700929157658e-06, "loss": 1.7837, "step": 2021 }, { "epoch": 0.15475279350987295, "grad_norm": 0.916354144145382, "learning_rate": 9.67831147098564e-06, "loss": 1.4538, "step": 2022 }, { "epoch": 0.15482932802694016, "grad_norm": 0.5931585095569053, "learning_rate": 9.677921786807167e-06, "loss": 1.4605, "step": 2023 }, { "epoch": 0.15490586254400734, "grad_norm": 0.8139454358697051, "learning_rate": 9.677531876646186e-06, "loss": 1.486, "step": 2024 }, { "epoch": 0.15498239706107456, "grad_norm": 0.5831957629712025, "learning_rate": 9.677141740526651e-06, "loss": 1.3854, "step": 2025 }, { "epoch": 0.15505893157814174, "grad_norm": 0.5837558009440184, "learning_rate": 9.676751378472538e-06, "loss": 1.3541, "step": 2026 }, { "epoch": 0.15513546609520895, "grad_norm": 0.6819217469262536, "learning_rate": 9.676360790507826e-06, "loss": 1.4611, "step": 2027 }, { "epoch": 0.15521200061227614, "grad_norm": 0.6173716423338441, "learning_rate": 9.675969976656519e-06, "loss": 1.3993, "step": 2028 }, { "epoch": 0.15528853512934335, "grad_norm": 0.6272036616983935, "learning_rate": 9.675578936942626e-06, "loss": 1.4794, "step": 2029 }, { "epoch": 0.15536506964641053, "grad_norm": 0.6070527937051294, "learning_rate": 9.675187671390176e-06, "loss": 1.309, "step": 2030 }, { "epoch": 0.15544160416347771, "grad_norm": 0.5339805164284801, "learning_rate": 9.674796180023207e-06, "loss": 1.1497, "step": 2031 }, { "epoch": 0.15551813868054493, "grad_norm": 0.5884429817440749, "learning_rate": 9.674404462865778e-06, "loss": 1.3591, "step": 2032 }, { "epoch": 0.1555946731976121, "grad_norm": 0.9937874270511595, "learning_rate": 9.674012519941953e-06, "loss": 1.7757, "step": 2033 }, { "epoch": 0.15567120771467932, "grad_norm": 0.5747983046723857, "learning_rate": 9.673620351275815e-06, "loss": 1.359, "step": 2034 }, { "epoch": 0.1557477422317465, "grad_norm": 0.6337487038886087, "learning_rate": 9.67322795689146e-06, "loss": 1.7619, "step": 2035 }, { "epoch": 0.15582427674881372, "grad_norm": 0.8415365185627853, "learning_rate": 9.672835336813e-06, "loss": 1.6582, "step": 2036 }, { "epoch": 0.1559008112658809, "grad_norm": 0.9504049640333834, "learning_rate": 9.672442491064558e-06, "loss": 1.6996, "step": 2037 }, { "epoch": 0.1559773457829481, "grad_norm": 0.6051505730802518, "learning_rate": 9.672049419670269e-06, "loss": 1.4429, "step": 2038 }, { "epoch": 0.1560538803000153, "grad_norm": 0.5446947720932335, "learning_rate": 9.671656122654286e-06, "loss": 1.2348, "step": 2039 }, { "epoch": 0.1561304148170825, "grad_norm": 0.5895130344989817, "learning_rate": 9.671262600040777e-06, "loss": 1.4855, "step": 2040 }, { "epoch": 0.1562069493341497, "grad_norm": 0.5833892338094573, "learning_rate": 9.670868851853917e-06, "loss": 1.1066, "step": 2041 }, { "epoch": 0.1562834838512169, "grad_norm": 0.5763831048086479, "learning_rate": 9.670474878117902e-06, "loss": 1.2394, "step": 2042 }, { "epoch": 0.1563600183682841, "grad_norm": 1.0172865085687424, "learning_rate": 9.670080678856937e-06, "loss": 1.631, "step": 2043 }, { "epoch": 0.1564365528853513, "grad_norm": 0.5507881603894568, "learning_rate": 9.669686254095244e-06, "loss": 1.3679, "step": 2044 }, { "epoch": 0.15651308740241848, "grad_norm": 0.5403830451155778, "learning_rate": 9.669291603857057e-06, "loss": 1.1543, "step": 2045 }, { "epoch": 0.1565896219194857, "grad_norm": 0.6463906657006258, "learning_rate": 9.668896728166624e-06, "loss": 1.6136, "step": 2046 }, { "epoch": 0.15666615643655288, "grad_norm": 0.5564534663881707, "learning_rate": 9.66850162704821e-06, "loss": 1.198, "step": 2047 }, { "epoch": 0.1567426909536201, "grad_norm": 0.8965765574312603, "learning_rate": 9.668106300526088e-06, "loss": 1.3868, "step": 2048 }, { "epoch": 0.15681922547068727, "grad_norm": 0.5964765952326284, "learning_rate": 9.667710748624549e-06, "loss": 1.4586, "step": 2049 }, { "epoch": 0.15689575998775449, "grad_norm": 0.5397142290741773, "learning_rate": 9.667314971367899e-06, "loss": 1.3398, "step": 2050 }, { "epoch": 0.15697229450482167, "grad_norm": 0.6006099982182643, "learning_rate": 9.666918968780454e-06, "loss": 1.5196, "step": 2051 }, { "epoch": 0.15704882902188888, "grad_norm": 0.595830482436582, "learning_rate": 9.666522740886542e-06, "loss": 1.3908, "step": 2052 }, { "epoch": 0.15712536353895606, "grad_norm": 0.6289030237713259, "learning_rate": 9.666126287710516e-06, "loss": 1.407, "step": 2053 }, { "epoch": 0.15720189805602328, "grad_norm": 0.6591686057883822, "learning_rate": 9.66572960927673e-06, "loss": 1.4844, "step": 2054 }, { "epoch": 0.15727843257309046, "grad_norm": 0.5994762305586602, "learning_rate": 9.665332705609558e-06, "loss": 1.1923, "step": 2055 }, { "epoch": 0.15735496709015767, "grad_norm": 1.4772464336539968, "learning_rate": 9.664935576733387e-06, "loss": 1.6499, "step": 2056 }, { "epoch": 0.15743150160722486, "grad_norm": 0.5330162920828931, "learning_rate": 9.66453822267262e-06, "loss": 1.1058, "step": 2057 }, { "epoch": 0.15750803612429207, "grad_norm": 0.5677788865937073, "learning_rate": 9.664140643451667e-06, "loss": 1.2294, "step": 2058 }, { "epoch": 0.15758457064135925, "grad_norm": 0.5507957450544012, "learning_rate": 9.663742839094961e-06, "loss": 1.0863, "step": 2059 }, { "epoch": 0.15766110515842646, "grad_norm": 0.5609477126916494, "learning_rate": 9.66334480962694e-06, "loss": 1.3854, "step": 2060 }, { "epoch": 0.15773763967549365, "grad_norm": 0.5702743827427746, "learning_rate": 9.662946555072065e-06, "loss": 1.2792, "step": 2061 }, { "epoch": 0.15781417419256086, "grad_norm": 0.5507400575162036, "learning_rate": 9.662548075454804e-06, "loss": 1.2126, "step": 2062 }, { "epoch": 0.15789070870962804, "grad_norm": 0.5926458723515755, "learning_rate": 9.662149370799638e-06, "loss": 1.4587, "step": 2063 }, { "epoch": 0.15796724322669525, "grad_norm": 0.5699596868535146, "learning_rate": 9.661750441131069e-06, "loss": 1.3094, "step": 2064 }, { "epoch": 0.15804377774376244, "grad_norm": 0.7431323977655722, "learning_rate": 9.661351286473607e-06, "loss": 1.6545, "step": 2065 }, { "epoch": 0.15812031226082962, "grad_norm": 0.8427611813204718, "learning_rate": 9.660951906851775e-06, "loss": 1.5002, "step": 2066 }, { "epoch": 0.15819684677789683, "grad_norm": 0.5877483583072685, "learning_rate": 9.660552302290113e-06, "loss": 1.3472, "step": 2067 }, { "epoch": 0.15827338129496402, "grad_norm": 0.5940711236647643, "learning_rate": 9.660152472813174e-06, "loss": 1.4853, "step": 2068 }, { "epoch": 0.15834991581203123, "grad_norm": 1.0103674944790721, "learning_rate": 9.659752418445527e-06, "loss": 1.5848, "step": 2069 }, { "epoch": 0.1584264503290984, "grad_norm": 0.6423419188747731, "learning_rate": 9.65935213921175e-06, "loss": 1.6593, "step": 2070 }, { "epoch": 0.15850298484616562, "grad_norm": 0.5599696398608085, "learning_rate": 9.65895163513644e-06, "loss": 1.3392, "step": 2071 }, { "epoch": 0.1585795193632328, "grad_norm": 0.48621716556910016, "learning_rate": 9.658550906244202e-06, "loss": 0.9655, "step": 2072 }, { "epoch": 0.15865605388030002, "grad_norm": 0.6805295922992928, "learning_rate": 9.658149952559658e-06, "loss": 1.3947, "step": 2073 }, { "epoch": 0.1587325883973672, "grad_norm": 0.6703342181593352, "learning_rate": 9.657748774107448e-06, "loss": 1.3256, "step": 2074 }, { "epoch": 0.15880912291443441, "grad_norm": 0.7644046973380032, "learning_rate": 9.657347370912218e-06, "loss": 1.6629, "step": 2075 }, { "epoch": 0.1588856574315016, "grad_norm": 0.6343833162599162, "learning_rate": 9.656945742998631e-06, "loss": 1.5805, "step": 2076 }, { "epoch": 0.1589621919485688, "grad_norm": 0.5268727602497083, "learning_rate": 9.656543890391365e-06, "loss": 1.1991, "step": 2077 }, { "epoch": 0.159038726465636, "grad_norm": 0.6646026868523928, "learning_rate": 9.656141813115111e-06, "loss": 1.7397, "step": 2078 }, { "epoch": 0.1591152609827032, "grad_norm": 0.7537667814341386, "learning_rate": 9.655739511194577e-06, "loss": 1.6003, "step": 2079 }, { "epoch": 0.1591917954997704, "grad_norm": 0.8604570206583371, "learning_rate": 9.655336984654477e-06, "loss": 1.7532, "step": 2080 }, { "epoch": 0.1592683300168376, "grad_norm": 0.6804313671875339, "learning_rate": 9.654934233519544e-06, "loss": 1.6779, "step": 2081 }, { "epoch": 0.15934486453390478, "grad_norm": 0.6532641360422818, "learning_rate": 9.654531257814527e-06, "loss": 1.7089, "step": 2082 }, { "epoch": 0.159421399050972, "grad_norm": 0.7392443580283058, "learning_rate": 9.654128057564186e-06, "loss": 1.5724, "step": 2083 }, { "epoch": 0.15949793356803918, "grad_norm": 0.6256922502079885, "learning_rate": 9.653724632793291e-06, "loss": 1.6634, "step": 2084 }, { "epoch": 0.1595744680851064, "grad_norm": 0.5148246449895355, "learning_rate": 9.653320983526634e-06, "loss": 1.1546, "step": 2085 }, { "epoch": 0.15965100260217358, "grad_norm": 0.5641024697414911, "learning_rate": 9.652917109789013e-06, "loss": 1.2316, "step": 2086 }, { "epoch": 0.1597275371192408, "grad_norm": 0.5449627177692359, "learning_rate": 9.652513011605246e-06, "loss": 1.2087, "step": 2087 }, { "epoch": 0.15980407163630797, "grad_norm": 0.5214162744202198, "learning_rate": 9.65210868900016e-06, "loss": 1.1579, "step": 2088 }, { "epoch": 0.15988060615337518, "grad_norm": 0.6755785384672962, "learning_rate": 9.651704141998598e-06, "loss": 1.5973, "step": 2089 }, { "epoch": 0.15995714067044237, "grad_norm": 0.5430369703530129, "learning_rate": 9.651299370625416e-06, "loss": 1.2146, "step": 2090 }, { "epoch": 0.16003367518750958, "grad_norm": 1.4488808730713443, "learning_rate": 9.650894374905486e-06, "loss": 0.9552, "step": 2091 }, { "epoch": 0.16011020970457676, "grad_norm": 0.5868900802383582, "learning_rate": 9.650489154863693e-06, "loss": 1.4446, "step": 2092 }, { "epoch": 0.16018674422164397, "grad_norm": 1.4393994900427036, "learning_rate": 9.650083710524932e-06, "loss": 1.5506, "step": 2093 }, { "epoch": 0.16026327873871116, "grad_norm": 0.8031076943998536, "learning_rate": 9.649678041914118e-06, "loss": 1.6672, "step": 2094 }, { "epoch": 0.16033981325577837, "grad_norm": 0.5597891894786511, "learning_rate": 9.649272149056172e-06, "loss": 1.4106, "step": 2095 }, { "epoch": 0.16041634777284555, "grad_norm": 0.6705506788829885, "learning_rate": 9.648866031976036e-06, "loss": 1.7205, "step": 2096 }, { "epoch": 0.16049288228991276, "grad_norm": 0.5821585823770016, "learning_rate": 9.648459690698663e-06, "loss": 1.2743, "step": 2097 }, { "epoch": 0.16056941680697995, "grad_norm": 0.5482501315458402, "learning_rate": 9.648053125249024e-06, "loss": 1.4109, "step": 2098 }, { "epoch": 0.16064595132404713, "grad_norm": 0.6413170783475747, "learning_rate": 9.64764633565209e-06, "loss": 1.4763, "step": 2099 }, { "epoch": 0.16072248584111434, "grad_norm": 0.6496341551985431, "learning_rate": 9.647239321932864e-06, "loss": 1.3389, "step": 2100 }, { "epoch": 0.16079902035818153, "grad_norm": 0.6333568534494232, "learning_rate": 9.64683208411635e-06, "loss": 1.5817, "step": 2101 }, { "epoch": 0.16087555487524874, "grad_norm": 0.6257871233352507, "learning_rate": 9.646424622227571e-06, "loss": 1.473, "step": 2102 }, { "epoch": 0.16095208939231592, "grad_norm": 0.5340746792325651, "learning_rate": 9.646016936291562e-06, "loss": 1.1357, "step": 2103 }, { "epoch": 0.16102862390938313, "grad_norm": 0.5405307690382942, "learning_rate": 9.645609026333373e-06, "loss": 1.1555, "step": 2104 }, { "epoch": 0.16110515842645032, "grad_norm": 0.6437622432242835, "learning_rate": 9.645200892378065e-06, "loss": 1.3113, "step": 2105 }, { "epoch": 0.16118169294351753, "grad_norm": 0.6090754077315456, "learning_rate": 9.644792534450721e-06, "loss": 1.3501, "step": 2106 }, { "epoch": 0.1612582274605847, "grad_norm": 0.6810183161138911, "learning_rate": 9.644383952576426e-06, "loss": 1.3584, "step": 2107 }, { "epoch": 0.16133476197765192, "grad_norm": 0.6111693566746632, "learning_rate": 9.643975146780286e-06, "loss": 1.3738, "step": 2108 }, { "epoch": 0.1614112964947191, "grad_norm": 0.8912608812314898, "learning_rate": 9.64356611708742e-06, "loss": 1.2414, "step": 2109 }, { "epoch": 0.16148783101178632, "grad_norm": 3.5279105950282146, "learning_rate": 9.64315686352296e-06, "loss": 1.7076, "step": 2110 }, { "epoch": 0.1615643655288535, "grad_norm": 0.6454326281453412, "learning_rate": 9.64274738611205e-06, "loss": 1.5133, "step": 2111 }, { "epoch": 0.16164090004592072, "grad_norm": 0.6120852597836395, "learning_rate": 9.642337684879852e-06, "loss": 1.6289, "step": 2112 }, { "epoch": 0.1617174345629879, "grad_norm": 0.5537477302606583, "learning_rate": 9.641927759851538e-06, "loss": 1.2825, "step": 2113 }, { "epoch": 0.1617939690800551, "grad_norm": 0.6880037448583786, "learning_rate": 9.641517611052296e-06, "loss": 1.7388, "step": 2114 }, { "epoch": 0.1618705035971223, "grad_norm": 1.2017283467804765, "learning_rate": 9.641107238507325e-06, "loss": 1.4705, "step": 2115 }, { "epoch": 0.1619470381141895, "grad_norm": 0.6175086252721973, "learning_rate": 9.640696642241839e-06, "loss": 1.2895, "step": 2116 }, { "epoch": 0.1620235726312567, "grad_norm": 0.6841229672908746, "learning_rate": 9.64028582228107e-06, "loss": 1.5542, "step": 2117 }, { "epoch": 0.1621001071483239, "grad_norm": 0.7525796788365215, "learning_rate": 9.639874778650257e-06, "loss": 1.3676, "step": 2118 }, { "epoch": 0.16217664166539109, "grad_norm": 0.517201843142044, "learning_rate": 9.639463511374656e-06, "loss": 1.1368, "step": 2119 }, { "epoch": 0.1622531761824583, "grad_norm": 0.6400540084757446, "learning_rate": 9.639052020479536e-06, "loss": 1.5301, "step": 2120 }, { "epoch": 0.16232971069952548, "grad_norm": 0.5503407074407136, "learning_rate": 9.638640305990184e-06, "loss": 1.1877, "step": 2121 }, { "epoch": 0.1624062452165927, "grad_norm": 0.5792225585574586, "learning_rate": 9.638228367931892e-06, "loss": 1.4327, "step": 2122 }, { "epoch": 0.16248277973365988, "grad_norm": 0.5237438718521191, "learning_rate": 9.637816206329974e-06, "loss": 1.125, "step": 2123 }, { "epoch": 0.1625593142507271, "grad_norm": 0.6407523233671756, "learning_rate": 9.637403821209752e-06, "loss": 1.1256, "step": 2124 }, { "epoch": 0.16263584876779427, "grad_norm": 0.5756111263598699, "learning_rate": 9.636991212596565e-06, "loss": 1.4245, "step": 2125 }, { "epoch": 0.16271238328486148, "grad_norm": 0.6423897125987819, "learning_rate": 9.636578380515767e-06, "loss": 1.4237, "step": 2126 }, { "epoch": 0.16278891780192867, "grad_norm": 0.6385980442361296, "learning_rate": 9.636165324992722e-06, "loss": 1.5742, "step": 2127 }, { "epoch": 0.16286545231899588, "grad_norm": 0.6940668366323878, "learning_rate": 9.63575204605281e-06, "loss": 1.801, "step": 2128 }, { "epoch": 0.16294198683606306, "grad_norm": 0.6869083216398931, "learning_rate": 9.63533854372142e-06, "loss": 1.4832, "step": 2129 }, { "epoch": 0.16301852135313027, "grad_norm": 0.6218853890858868, "learning_rate": 9.634924818023965e-06, "loss": 1.4717, "step": 2130 }, { "epoch": 0.16309505587019746, "grad_norm": 0.7217816006309244, "learning_rate": 9.634510868985862e-06, "loss": 1.7131, "step": 2131 }, { "epoch": 0.16317159038726467, "grad_norm": 0.5778899270380801, "learning_rate": 9.634096696632546e-06, "loss": 1.1254, "step": 2132 }, { "epoch": 0.16324812490433185, "grad_norm": 0.5875854671829556, "learning_rate": 9.633682300989466e-06, "loss": 1.3569, "step": 2133 }, { "epoch": 0.16332465942139904, "grad_norm": 0.6082692180818654, "learning_rate": 9.633267682082083e-06, "loss": 1.3433, "step": 2134 }, { "epoch": 0.16340119393846625, "grad_norm": 0.7027693915297355, "learning_rate": 9.632852839935871e-06, "loss": 1.764, "step": 2135 }, { "epoch": 0.16347772845553343, "grad_norm": 0.6170060123170137, "learning_rate": 9.632437774576322e-06, "loss": 1.4509, "step": 2136 }, { "epoch": 0.16355426297260064, "grad_norm": 0.6173777042837625, "learning_rate": 9.632022486028935e-06, "loss": 1.4244, "step": 2137 }, { "epoch": 0.16363079748966783, "grad_norm": 0.6473773546612295, "learning_rate": 9.63160697431923e-06, "loss": 1.4954, "step": 2138 }, { "epoch": 0.16370733200673504, "grad_norm": 0.7845334438148234, "learning_rate": 9.631191239472735e-06, "loss": 1.3936, "step": 2139 }, { "epoch": 0.16378386652380222, "grad_norm": 0.6606933113028135, "learning_rate": 9.630775281514997e-06, "loss": 1.553, "step": 2140 }, { "epoch": 0.16386040104086944, "grad_norm": 0.6415831689350497, "learning_rate": 9.63035910047157e-06, "loss": 1.3907, "step": 2141 }, { "epoch": 0.16393693555793662, "grad_norm": 0.6061879094592979, "learning_rate": 9.629942696368028e-06, "loss": 1.2941, "step": 2142 }, { "epoch": 0.16401347007500383, "grad_norm": 0.5331444270235933, "learning_rate": 9.629526069229954e-06, "loss": 1.2471, "step": 2143 }, { "epoch": 0.16409000459207101, "grad_norm": 0.6803827611907484, "learning_rate": 9.62910921908295e-06, "loss": 1.729, "step": 2144 }, { "epoch": 0.16416653910913823, "grad_norm": 0.915272756002407, "learning_rate": 9.628692145952626e-06, "loss": 1.7169, "step": 2145 }, { "epoch": 0.1642430736262054, "grad_norm": 0.5556574076366176, "learning_rate": 9.62827484986461e-06, "loss": 1.3617, "step": 2146 }, { "epoch": 0.16431960814327262, "grad_norm": 0.6415762271837604, "learning_rate": 9.627857330844537e-06, "loss": 1.3175, "step": 2147 }, { "epoch": 0.1643961426603398, "grad_norm": 0.5895817951382765, "learning_rate": 9.627439588918068e-06, "loss": 1.2698, "step": 2148 }, { "epoch": 0.16447267717740702, "grad_norm": 0.8366855457625753, "learning_rate": 9.627021624110866e-06, "loss": 1.3765, "step": 2149 }, { "epoch": 0.1645492116944742, "grad_norm": 0.5815343219661528, "learning_rate": 9.62660343644861e-06, "loss": 1.3209, "step": 2150 }, { "epoch": 0.1646257462115414, "grad_norm": 0.6437613558596333, "learning_rate": 9.626185025957e-06, "loss": 1.5148, "step": 2151 }, { "epoch": 0.1647022807286086, "grad_norm": 0.5637694598453082, "learning_rate": 9.625766392661742e-06, "loss": 1.3155, "step": 2152 }, { "epoch": 0.1647788152456758, "grad_norm": 0.5636336182005095, "learning_rate": 9.625347536588558e-06, "loss": 1.2206, "step": 2153 }, { "epoch": 0.164855349762743, "grad_norm": 0.5891724962416506, "learning_rate": 9.624928457763181e-06, "loss": 1.518, "step": 2154 }, { "epoch": 0.1649318842798102, "grad_norm": 0.6090108661378475, "learning_rate": 9.624509156211365e-06, "loss": 1.4909, "step": 2155 }, { "epoch": 0.1650084187968774, "grad_norm": 0.6074566906904172, "learning_rate": 9.624089631958871e-06, "loss": 1.562, "step": 2156 }, { "epoch": 0.1650849533139446, "grad_norm": 1.1007842420294227, "learning_rate": 9.623669885031476e-06, "loss": 1.8518, "step": 2157 }, { "epoch": 0.16516148783101178, "grad_norm": 0.6371387278734632, "learning_rate": 9.623249915454971e-06, "loss": 1.4989, "step": 2158 }, { "epoch": 0.165238022348079, "grad_norm": 0.6063698966917066, "learning_rate": 9.62282972325516e-06, "loss": 1.4602, "step": 2159 }, { "epoch": 0.16531455686514618, "grad_norm": 0.5593208165463927, "learning_rate": 9.62240930845786e-06, "loss": 1.3366, "step": 2160 }, { "epoch": 0.1653910913822134, "grad_norm": 0.6062444891201099, "learning_rate": 9.621988671088902e-06, "loss": 1.4438, "step": 2161 }, { "epoch": 0.16546762589928057, "grad_norm": 0.6089083545384453, "learning_rate": 9.621567811174135e-06, "loss": 1.487, "step": 2162 }, { "epoch": 0.16554416041634779, "grad_norm": 0.5736953041264621, "learning_rate": 9.621146728739415e-06, "loss": 1.3155, "step": 2163 }, { "epoch": 0.16562069493341497, "grad_norm": 0.5892937411577459, "learning_rate": 9.620725423810614e-06, "loss": 1.2041, "step": 2164 }, { "epoch": 0.16569722945048218, "grad_norm": 0.5742281705711054, "learning_rate": 9.620303896413616e-06, "loss": 1.4501, "step": 2165 }, { "epoch": 0.16577376396754936, "grad_norm": 0.5535654205079884, "learning_rate": 9.619882146574327e-06, "loss": 1.4011, "step": 2166 }, { "epoch": 0.16585029848461655, "grad_norm": 0.5640460169358505, "learning_rate": 9.619460174318658e-06, "loss": 1.3211, "step": 2167 }, { "epoch": 0.16592683300168376, "grad_norm": 0.6324488812284663, "learning_rate": 9.619037979672534e-06, "loss": 1.3696, "step": 2168 }, { "epoch": 0.16600336751875094, "grad_norm": 0.8765961998045241, "learning_rate": 9.618615562661899e-06, "loss": 1.9595, "step": 2169 }, { "epoch": 0.16607990203581816, "grad_norm": 0.6566410977251153, "learning_rate": 9.618192923312706e-06, "loss": 1.6573, "step": 2170 }, { "epoch": 0.16615643655288534, "grad_norm": 0.5807030696802075, "learning_rate": 9.617770061650925e-06, "loss": 1.457, "step": 2171 }, { "epoch": 0.16623297106995255, "grad_norm": 0.6017573501899296, "learning_rate": 9.617346977702532e-06, "loss": 1.3214, "step": 2172 }, { "epoch": 0.16630950558701973, "grad_norm": 0.587909070898927, "learning_rate": 9.61692367149353e-06, "loss": 1.2536, "step": 2173 }, { "epoch": 0.16638604010408695, "grad_norm": 1.093704266640602, "learning_rate": 9.616500143049924e-06, "loss": 1.649, "step": 2174 }, { "epoch": 0.16646257462115413, "grad_norm": 0.5911984553203258, "learning_rate": 9.616076392397739e-06, "loss": 1.1574, "step": 2175 }, { "epoch": 0.16653910913822134, "grad_norm": 0.6343755950421811, "learning_rate": 9.615652419563009e-06, "loss": 1.5273, "step": 2176 }, { "epoch": 0.16661564365528853, "grad_norm": 0.5623387595364507, "learning_rate": 9.615228224571788e-06, "loss": 1.3895, "step": 2177 }, { "epoch": 0.16669217817235574, "grad_norm": 0.6196064209446727, "learning_rate": 9.614803807450135e-06, "loss": 1.4669, "step": 2178 }, { "epoch": 0.16676871268942292, "grad_norm": 0.5202706327787163, "learning_rate": 9.614379168224131e-06, "loss": 1.1929, "step": 2179 }, { "epoch": 0.16684524720649013, "grad_norm": 0.5401881715928046, "learning_rate": 9.613954306919867e-06, "loss": 1.1693, "step": 2180 }, { "epoch": 0.16692178172355732, "grad_norm": 0.6360240603315916, "learning_rate": 9.613529223563443e-06, "loss": 1.5183, "step": 2181 }, { "epoch": 0.16699831624062453, "grad_norm": 0.5922224224896596, "learning_rate": 9.613103918180985e-06, "loss": 1.5088, "step": 2182 }, { "epoch": 0.1670748507576917, "grad_norm": 0.5407345885603732, "learning_rate": 9.61267839079862e-06, "loss": 1.3786, "step": 2183 }, { "epoch": 0.16715138527475892, "grad_norm": 0.5563903421428817, "learning_rate": 9.612252641442494e-06, "loss": 1.2604, "step": 2184 }, { "epoch": 0.1672279197918261, "grad_norm": 0.6531918077956829, "learning_rate": 9.611826670138767e-06, "loss": 1.6746, "step": 2185 }, { "epoch": 0.16730445430889332, "grad_norm": 0.697448874159167, "learning_rate": 9.611400476913613e-06, "loss": 1.4135, "step": 2186 }, { "epoch": 0.1673809888259605, "grad_norm": 0.5714329831587254, "learning_rate": 9.610974061793217e-06, "loss": 1.1427, "step": 2187 }, { "epoch": 0.16745752334302771, "grad_norm": 0.5880344410880886, "learning_rate": 9.610547424803782e-06, "loss": 1.2363, "step": 2188 }, { "epoch": 0.1675340578600949, "grad_norm": 0.8368574794657847, "learning_rate": 9.610120565971518e-06, "loss": 1.4767, "step": 2189 }, { "epoch": 0.1676105923771621, "grad_norm": 0.5665534532293459, "learning_rate": 9.609693485322655e-06, "loss": 1.0934, "step": 2190 }, { "epoch": 0.1676871268942293, "grad_norm": 0.5668405836790584, "learning_rate": 9.60926618288343e-06, "loss": 1.3475, "step": 2191 }, { "epoch": 0.1677636614112965, "grad_norm": 0.5428606464444007, "learning_rate": 9.608838658680104e-06, "loss": 1.1247, "step": 2192 }, { "epoch": 0.1678401959283637, "grad_norm": 0.6267396331917516, "learning_rate": 9.608410912738943e-06, "loss": 1.3989, "step": 2193 }, { "epoch": 0.1679167304454309, "grad_norm": 0.5837042935444181, "learning_rate": 9.607982945086228e-06, "loss": 1.3321, "step": 2194 }, { "epoch": 0.16799326496249808, "grad_norm": 0.5656342547339799, "learning_rate": 9.607554755748254e-06, "loss": 1.4085, "step": 2195 }, { "epoch": 0.1680697994795653, "grad_norm": 0.6173675352964092, "learning_rate": 9.607126344751332e-06, "loss": 1.5822, "step": 2196 }, { "epoch": 0.16814633399663248, "grad_norm": 0.883679631463487, "learning_rate": 9.606697712121784e-06, "loss": 1.3877, "step": 2197 }, { "epoch": 0.1682228685136997, "grad_norm": 0.6386902760203125, "learning_rate": 9.606268857885947e-06, "loss": 1.5094, "step": 2198 }, { "epoch": 0.16829940303076688, "grad_norm": 0.6841960567931732, "learning_rate": 9.60583978207017e-06, "loss": 1.5905, "step": 2199 }, { "epoch": 0.1683759375478341, "grad_norm": 0.6054124635067187, "learning_rate": 9.605410484700816e-06, "loss": 1.3576, "step": 2200 }, { "epoch": 0.16845247206490127, "grad_norm": 0.812880834160577, "learning_rate": 9.604980965804267e-06, "loss": 1.6244, "step": 2201 }, { "epoch": 0.16852900658196845, "grad_norm": 0.6311833867721007, "learning_rate": 9.604551225406908e-06, "loss": 1.575, "step": 2202 }, { "epoch": 0.16860554109903567, "grad_norm": 0.612754277792903, "learning_rate": 9.604121263535148e-06, "loss": 1.4179, "step": 2203 }, { "epoch": 0.16868207561610285, "grad_norm": 0.6039079935082383, "learning_rate": 9.603691080215402e-06, "loss": 1.4282, "step": 2204 }, { "epoch": 0.16875861013317006, "grad_norm": 1.8107459260292846, "learning_rate": 9.603260675474104e-06, "loss": 1.6215, "step": 2205 }, { "epoch": 0.16883514465023725, "grad_norm": 0.7691272269428758, "learning_rate": 9.602830049337697e-06, "loss": 1.5941, "step": 2206 }, { "epoch": 0.16891167916730446, "grad_norm": 0.6384181417836585, "learning_rate": 9.602399201832642e-06, "loss": 1.6419, "step": 2207 }, { "epoch": 0.16898821368437164, "grad_norm": 0.7304450784782088, "learning_rate": 9.60196813298541e-06, "loss": 1.8071, "step": 2208 }, { "epoch": 0.16906474820143885, "grad_norm": 0.5658206238645729, "learning_rate": 9.60153684282249e-06, "loss": 1.3403, "step": 2209 }, { "epoch": 0.16914128271850604, "grad_norm": 0.549587128019039, "learning_rate": 9.601105331370379e-06, "loss": 1.2128, "step": 2210 }, { "epoch": 0.16921781723557325, "grad_norm": 0.781867067024751, "learning_rate": 9.600673598655591e-06, "loss": 1.3264, "step": 2211 }, { "epoch": 0.16929435175264043, "grad_norm": 0.6388155232968492, "learning_rate": 9.600241644704653e-06, "loss": 1.5199, "step": 2212 }, { "epoch": 0.16937088626970764, "grad_norm": 0.5883773602477306, "learning_rate": 9.599809469544106e-06, "loss": 1.345, "step": 2213 }, { "epoch": 0.16944742078677483, "grad_norm": 0.6907915696287581, "learning_rate": 9.599377073200505e-06, "loss": 1.3376, "step": 2214 }, { "epoch": 0.16952395530384204, "grad_norm": 0.8378315070742539, "learning_rate": 9.598944455700415e-06, "loss": 1.79, "step": 2215 }, { "epoch": 0.16960048982090922, "grad_norm": 0.617008484776259, "learning_rate": 9.598511617070419e-06, "loss": 1.477, "step": 2216 }, { "epoch": 0.16967702433797643, "grad_norm": 0.5760568885056864, "learning_rate": 9.598078557337112e-06, "loss": 1.2657, "step": 2217 }, { "epoch": 0.16975355885504362, "grad_norm": 0.5813876793909836, "learning_rate": 9.597645276527102e-06, "loss": 1.3329, "step": 2218 }, { "epoch": 0.16983009337211083, "grad_norm": 0.6184375680487804, "learning_rate": 9.597211774667013e-06, "loss": 1.4152, "step": 2219 }, { "epoch": 0.169906627889178, "grad_norm": 0.7822684410568462, "learning_rate": 9.596778051783477e-06, "loss": 1.8666, "step": 2220 }, { "epoch": 0.16998316240624523, "grad_norm": 0.730218188532615, "learning_rate": 9.596344107903145e-06, "loss": 1.3298, "step": 2221 }, { "epoch": 0.1700596969233124, "grad_norm": 0.614235881000896, "learning_rate": 9.59590994305268e-06, "loss": 1.4758, "step": 2222 }, { "epoch": 0.17013623144037962, "grad_norm": 0.5574127687254921, "learning_rate": 9.59547555725876e-06, "loss": 1.3601, "step": 2223 }, { "epoch": 0.1702127659574468, "grad_norm": 0.6846783334546636, "learning_rate": 9.595040950548071e-06, "loss": 1.7049, "step": 2224 }, { "epoch": 0.17028930047451402, "grad_norm": 0.6015748321236868, "learning_rate": 9.594606122947321e-06, "loss": 1.3609, "step": 2225 }, { "epoch": 0.1703658349915812, "grad_norm": 0.48041335937903396, "learning_rate": 9.594171074483223e-06, "loss": 0.9745, "step": 2226 }, { "epoch": 0.1704423695086484, "grad_norm": 0.6029660900689182, "learning_rate": 9.593735805182511e-06, "loss": 1.4462, "step": 2227 }, { "epoch": 0.1705189040257156, "grad_norm": 0.5557945936120754, "learning_rate": 9.593300315071928e-06, "loss": 1.253, "step": 2228 }, { "epoch": 0.1705954385427828, "grad_norm": 0.6015666055951184, "learning_rate": 9.592864604178231e-06, "loss": 1.6677, "step": 2229 }, { "epoch": 0.17067197305985, "grad_norm": 0.800936322427504, "learning_rate": 9.592428672528193e-06, "loss": 1.1814, "step": 2230 }, { "epoch": 0.1707485075769172, "grad_norm": 0.5539435535382777, "learning_rate": 9.591992520148598e-06, "loss": 1.2503, "step": 2231 }, { "epoch": 0.17082504209398439, "grad_norm": 0.9219162805716423, "learning_rate": 9.591556147066244e-06, "loss": 1.7913, "step": 2232 }, { "epoch": 0.1709015766110516, "grad_norm": 0.49380856363471115, "learning_rate": 9.591119553307944e-06, "loss": 1.0149, "step": 2233 }, { "epoch": 0.17097811112811878, "grad_norm": 0.6060315669253254, "learning_rate": 9.590682738900522e-06, "loss": 1.438, "step": 2234 }, { "epoch": 0.17105464564518597, "grad_norm": 0.6837590562035123, "learning_rate": 9.59024570387082e-06, "loss": 1.3301, "step": 2235 }, { "epoch": 0.17113118016225318, "grad_norm": 0.5195796572738219, "learning_rate": 9.589808448245688e-06, "loss": 1.1644, "step": 2236 }, { "epoch": 0.17120771467932036, "grad_norm": 0.7696921734734018, "learning_rate": 9.589370972051994e-06, "loss": 1.4259, "step": 2237 }, { "epoch": 0.17128424919638757, "grad_norm": 0.6246298281560055, "learning_rate": 9.588933275316617e-06, "loss": 1.385, "step": 2238 }, { "epoch": 0.17136078371345476, "grad_norm": 0.5922538335525316, "learning_rate": 9.588495358066452e-06, "loss": 1.3886, "step": 2239 }, { "epoch": 0.17143731823052197, "grad_norm": 0.5836200070862517, "learning_rate": 9.588057220328402e-06, "loss": 1.2249, "step": 2240 }, { "epoch": 0.17151385274758915, "grad_norm": 0.6128808619719696, "learning_rate": 9.587618862129392e-06, "loss": 1.541, "step": 2241 }, { "epoch": 0.17159038726465636, "grad_norm": 0.5480829007946616, "learning_rate": 9.587180283496355e-06, "loss": 1.2826, "step": 2242 }, { "epoch": 0.17166692178172355, "grad_norm": 0.5420790869550607, "learning_rate": 9.586741484456235e-06, "loss": 1.3153, "step": 2243 }, { "epoch": 0.17174345629879076, "grad_norm": 0.4746757550317301, "learning_rate": 9.586302465036e-06, "loss": 0.9674, "step": 2244 }, { "epoch": 0.17181999081585794, "grad_norm": 0.6341923024753223, "learning_rate": 9.585863225262617e-06, "loss": 1.544, "step": 2245 }, { "epoch": 0.17189652533292515, "grad_norm": 0.5678958981715243, "learning_rate": 9.585423765163079e-06, "loss": 1.2496, "step": 2246 }, { "epoch": 0.17197305984999234, "grad_norm": 0.5306081735771707, "learning_rate": 9.584984084764388e-06, "loss": 1.2834, "step": 2247 }, { "epoch": 0.17204959436705955, "grad_norm": 0.565098784431439, "learning_rate": 9.584544184093554e-06, "loss": 1.4217, "step": 2248 }, { "epoch": 0.17212612888412673, "grad_norm": 0.6867480987990571, "learning_rate": 9.584104063177612e-06, "loss": 1.3781, "step": 2249 }, { "epoch": 0.17220266340119394, "grad_norm": 0.6342504282553671, "learning_rate": 9.583663722043602e-06, "loss": 1.4867, "step": 2250 }, { "epoch": 0.17227919791826113, "grad_norm": 0.5829569539232233, "learning_rate": 9.583223160718579e-06, "loss": 1.316, "step": 2251 }, { "epoch": 0.17235573243532834, "grad_norm": 0.5788614969891752, "learning_rate": 9.582782379229613e-06, "loss": 1.2751, "step": 2252 }, { "epoch": 0.17243226695239552, "grad_norm": 0.6451083138983534, "learning_rate": 9.582341377603788e-06, "loss": 1.3953, "step": 2253 }, { "epoch": 0.17250880146946274, "grad_norm": 0.6031981229339078, "learning_rate": 9.5819001558682e-06, "loss": 1.4219, "step": 2254 }, { "epoch": 0.17258533598652992, "grad_norm": 0.5439441973136191, "learning_rate": 9.581458714049958e-06, "loss": 1.2805, "step": 2255 }, { "epoch": 0.17266187050359713, "grad_norm": 0.6603117376561083, "learning_rate": 9.581017052176186e-06, "loss": 1.5514, "step": 2256 }, { "epoch": 0.17273840502066432, "grad_norm": 0.6247815255524726, "learning_rate": 9.580575170274022e-06, "loss": 1.5607, "step": 2257 }, { "epoch": 0.17281493953773153, "grad_norm": 0.9033587410076378, "learning_rate": 9.580133068370614e-06, "loss": 1.4738, "step": 2258 }, { "epoch": 0.1728914740547987, "grad_norm": 0.8642666131035112, "learning_rate": 9.579690746493128e-06, "loss": 1.6018, "step": 2259 }, { "epoch": 0.17296800857186592, "grad_norm": 0.6736973648648935, "learning_rate": 9.579248204668743e-06, "loss": 1.5241, "step": 2260 }, { "epoch": 0.1730445430889331, "grad_norm": 0.6866511454618097, "learning_rate": 9.578805442924645e-06, "loss": 1.5917, "step": 2261 }, { "epoch": 0.17312107760600032, "grad_norm": 0.6249230006527684, "learning_rate": 9.578362461288045e-06, "loss": 1.5429, "step": 2262 }, { "epoch": 0.1731976121230675, "grad_norm": 1.005520178391912, "learning_rate": 9.577919259786157e-06, "loss": 1.0428, "step": 2263 }, { "epoch": 0.1732741466401347, "grad_norm": 0.6191973801678918, "learning_rate": 9.577475838446214e-06, "loss": 1.4208, "step": 2264 }, { "epoch": 0.1733506811572019, "grad_norm": 0.5751812978469297, "learning_rate": 9.577032197295462e-06, "loss": 1.3477, "step": 2265 }, { "epoch": 0.1734272156742691, "grad_norm": 0.5747643161459071, "learning_rate": 9.576588336361157e-06, "loss": 1.2872, "step": 2266 }, { "epoch": 0.1735037501913363, "grad_norm": 0.6389443092986297, "learning_rate": 9.576144255670575e-06, "loss": 1.3905, "step": 2267 }, { "epoch": 0.17358028470840348, "grad_norm": 0.5679356367017567, "learning_rate": 9.575699955250996e-06, "loss": 1.3301, "step": 2268 }, { "epoch": 0.1736568192254707, "grad_norm": 0.6653162306042332, "learning_rate": 9.575255435129724e-06, "loss": 1.6326, "step": 2269 }, { "epoch": 0.17373335374253787, "grad_norm": 0.7608542712714269, "learning_rate": 9.574810695334069e-06, "loss": 1.6124, "step": 2270 }, { "epoch": 0.17380988825960508, "grad_norm": 0.5203465121003682, "learning_rate": 9.57436573589136e-06, "loss": 1.0702, "step": 2271 }, { "epoch": 0.17388642277667227, "grad_norm": 0.5200431220493401, "learning_rate": 9.573920556828934e-06, "loss": 1.2698, "step": 2272 }, { "epoch": 0.17396295729373948, "grad_norm": 0.5746674735762327, "learning_rate": 9.573475158174146e-06, "loss": 1.2689, "step": 2273 }, { "epoch": 0.17403949181080666, "grad_norm": 0.6082133260943107, "learning_rate": 9.57302953995436e-06, "loss": 1.3851, "step": 2274 }, { "epoch": 0.17411602632787387, "grad_norm": 0.6278943365853425, "learning_rate": 9.572583702196958e-06, "loss": 1.5068, "step": 2275 }, { "epoch": 0.17419256084494106, "grad_norm": 0.6681033001306238, "learning_rate": 9.572137644929333e-06, "loss": 1.6231, "step": 2276 }, { "epoch": 0.17426909536200827, "grad_norm": 1.3704272882519237, "learning_rate": 9.571691368178894e-06, "loss": 1.5712, "step": 2277 }, { "epoch": 0.17434562987907545, "grad_norm": 0.6407520724834911, "learning_rate": 9.57124487197306e-06, "loss": 1.4987, "step": 2278 }, { "epoch": 0.17442216439614266, "grad_norm": 0.6044974800511179, "learning_rate": 9.570798156339263e-06, "loss": 1.5033, "step": 2279 }, { "epoch": 0.17449869891320985, "grad_norm": 0.6105131416773367, "learning_rate": 9.570351221304954e-06, "loss": 1.4393, "step": 2280 }, { "epoch": 0.17457523343027706, "grad_norm": 0.6580116289936488, "learning_rate": 9.569904066897592e-06, "loss": 1.3874, "step": 2281 }, { "epoch": 0.17465176794734424, "grad_norm": 0.6725998787296967, "learning_rate": 9.569456693144654e-06, "loss": 1.6401, "step": 2282 }, { "epoch": 0.17472830246441146, "grad_norm": 0.5488743553732436, "learning_rate": 9.569009100073625e-06, "loss": 1.3667, "step": 2283 }, { "epoch": 0.17480483698147864, "grad_norm": 0.6339334264237917, "learning_rate": 9.568561287712007e-06, "loss": 1.3976, "step": 2284 }, { "epoch": 0.17488137149854585, "grad_norm": 0.5415317309658754, "learning_rate": 9.568113256087316e-06, "loss": 1.2774, "step": 2285 }, { "epoch": 0.17495790601561304, "grad_norm": 0.5423121703829704, "learning_rate": 9.56766500522708e-06, "loss": 1.2145, "step": 2286 }, { "epoch": 0.17503444053268025, "grad_norm": 0.6519195136207047, "learning_rate": 9.567216535158839e-06, "loss": 1.5206, "step": 2287 }, { "epoch": 0.17511097504974743, "grad_norm": 0.6697054243048381, "learning_rate": 9.566767845910152e-06, "loss": 1.5832, "step": 2288 }, { "epoch": 0.17518750956681464, "grad_norm": 0.5722538037198271, "learning_rate": 9.566318937508586e-06, "loss": 1.2246, "step": 2289 }, { "epoch": 0.17526404408388183, "grad_norm": 0.9296362584583991, "learning_rate": 9.565869809981723e-06, "loss": 1.5751, "step": 2290 }, { "epoch": 0.17534057860094904, "grad_norm": 0.651033971673011, "learning_rate": 9.56542046335716e-06, "loss": 1.4599, "step": 2291 }, { "epoch": 0.17541711311801622, "grad_norm": 0.719920591635547, "learning_rate": 9.564970897662504e-06, "loss": 1.2463, "step": 2292 }, { "epoch": 0.17549364763508343, "grad_norm": 0.5881475470014438, "learning_rate": 9.564521112925378e-06, "loss": 1.4062, "step": 2293 }, { "epoch": 0.17557018215215062, "grad_norm": 0.6663912968999121, "learning_rate": 9.564071109173419e-06, "loss": 1.6471, "step": 2294 }, { "epoch": 0.17564671666921783, "grad_norm": 0.7787491712428583, "learning_rate": 9.563620886434278e-06, "loss": 1.3431, "step": 2295 }, { "epoch": 0.175723251186285, "grad_norm": 0.7336131399272143, "learning_rate": 9.563170444735616e-06, "loss": 1.315, "step": 2296 }, { "epoch": 0.17579978570335222, "grad_norm": 0.5180718900628111, "learning_rate": 9.56271978410511e-06, "loss": 1.1112, "step": 2297 }, { "epoch": 0.1758763202204194, "grad_norm": 0.8564598118567298, "learning_rate": 9.56226890457045e-06, "loss": 1.6082, "step": 2298 }, { "epoch": 0.17595285473748662, "grad_norm": 0.5357807450988352, "learning_rate": 9.561817806159339e-06, "loss": 1.2712, "step": 2299 }, { "epoch": 0.1760293892545538, "grad_norm": 0.534617011014926, "learning_rate": 9.561366488899494e-06, "loss": 1.2142, "step": 2300 }, { "epoch": 0.17610592377162101, "grad_norm": 0.9010336545193942, "learning_rate": 9.560914952818645e-06, "loss": 1.4963, "step": 2301 }, { "epoch": 0.1761824582886882, "grad_norm": 1.1380759735858126, "learning_rate": 9.560463197944537e-06, "loss": 1.6838, "step": 2302 }, { "epoch": 0.17625899280575538, "grad_norm": 0.5249887333920763, "learning_rate": 9.560011224304928e-06, "loss": 1.1379, "step": 2303 }, { "epoch": 0.1763355273228226, "grad_norm": 0.9954632809527824, "learning_rate": 9.559559031927582e-06, "loss": 1.5815, "step": 2304 }, { "epoch": 0.17641206183988978, "grad_norm": 0.5356344287253242, "learning_rate": 9.559106620840294e-06, "loss": 0.981, "step": 2305 }, { "epoch": 0.176488596356957, "grad_norm": 0.5388165841250941, "learning_rate": 9.558653991070853e-06, "loss": 1.2728, "step": 2306 }, { "epoch": 0.17656513087402417, "grad_norm": 1.3267696111516458, "learning_rate": 9.558201142647071e-06, "loss": 1.5512, "step": 2307 }, { "epoch": 0.17664166539109138, "grad_norm": 0.5468972800917714, "learning_rate": 9.557748075596775e-06, "loss": 1.1921, "step": 2308 }, { "epoch": 0.17671819990815857, "grad_norm": 0.597627489721506, "learning_rate": 9.557294789947801e-06, "loss": 1.1343, "step": 2309 }, { "epoch": 0.17679473442522578, "grad_norm": 0.5660987041042057, "learning_rate": 9.556841285728003e-06, "loss": 1.2727, "step": 2310 }, { "epoch": 0.17687126894229296, "grad_norm": 0.5880820314118721, "learning_rate": 9.55638756296524e-06, "loss": 1.3671, "step": 2311 }, { "epoch": 0.17694780345936018, "grad_norm": 0.7061719362569757, "learning_rate": 9.555933621687397e-06, "loss": 1.4438, "step": 2312 }, { "epoch": 0.17702433797642736, "grad_norm": 0.6226730052953722, "learning_rate": 9.555479461922361e-06, "loss": 1.4497, "step": 2313 }, { "epoch": 0.17710087249349457, "grad_norm": 0.9092615925378945, "learning_rate": 9.555025083698037e-06, "loss": 1.3423, "step": 2314 }, { "epoch": 0.17717740701056175, "grad_norm": 0.6435742775256178, "learning_rate": 9.554570487042345e-06, "loss": 1.556, "step": 2315 }, { "epoch": 0.17725394152762897, "grad_norm": 0.6412742698091456, "learning_rate": 9.554115671983216e-06, "loss": 1.6225, "step": 2316 }, { "epoch": 0.17733047604469615, "grad_norm": 1.0845641715931624, "learning_rate": 9.553660638548595e-06, "loss": 1.741, "step": 2317 }, { "epoch": 0.17740701056176336, "grad_norm": 1.1056724048550277, "learning_rate": 9.553205386766442e-06, "loss": 1.5463, "step": 2318 }, { "epoch": 0.17748354507883055, "grad_norm": 0.6267004448505119, "learning_rate": 9.552749916664727e-06, "loss": 1.4356, "step": 2319 }, { "epoch": 0.17756007959589776, "grad_norm": 0.7659082608332283, "learning_rate": 9.552294228271436e-06, "loss": 1.5583, "step": 2320 }, { "epoch": 0.17763661411296494, "grad_norm": 0.5220075419901191, "learning_rate": 9.55183832161457e-06, "loss": 1.1074, "step": 2321 }, { "epoch": 0.17771314863003215, "grad_norm": 0.5474940642973203, "learning_rate": 9.551382196722139e-06, "loss": 1.4146, "step": 2322 }, { "epoch": 0.17778968314709934, "grad_norm": 0.7368404077646498, "learning_rate": 9.550925853622168e-06, "loss": 1.7987, "step": 2323 }, { "epoch": 0.17786621766416655, "grad_norm": 0.6162277536790685, "learning_rate": 9.550469292342698e-06, "loss": 1.4232, "step": 2324 }, { "epoch": 0.17794275218123373, "grad_norm": 0.5862236885889057, "learning_rate": 9.55001251291178e-06, "loss": 1.4264, "step": 2325 }, { "epoch": 0.17801928669830094, "grad_norm": 0.6478861264263345, "learning_rate": 9.549555515357482e-06, "loss": 1.6151, "step": 2326 }, { "epoch": 0.17809582121536813, "grad_norm": 0.7575061888978906, "learning_rate": 9.549098299707881e-06, "loss": 1.393, "step": 2327 }, { "epoch": 0.17817235573243534, "grad_norm": 0.9181648268121179, "learning_rate": 9.548640865991072e-06, "loss": 1.6153, "step": 2328 }, { "epoch": 0.17824889024950252, "grad_norm": 0.5875079272541563, "learning_rate": 9.548183214235159e-06, "loss": 1.5475, "step": 2329 }, { "epoch": 0.17832542476656973, "grad_norm": 0.6488389837786853, "learning_rate": 9.547725344468262e-06, "loss": 1.5138, "step": 2330 }, { "epoch": 0.17840195928363692, "grad_norm": 1.0262593230645087, "learning_rate": 9.547267256718516e-06, "loss": 1.4165, "step": 2331 }, { "epoch": 0.17847849380070413, "grad_norm": 0.5907194250141442, "learning_rate": 9.546808951014063e-06, "loss": 1.3352, "step": 2332 }, { "epoch": 0.1785550283177713, "grad_norm": 0.5880932690128596, "learning_rate": 9.546350427383066e-06, "loss": 1.2647, "step": 2333 }, { "epoch": 0.17863156283483853, "grad_norm": 0.6608365145933051, "learning_rate": 9.545891685853697e-06, "loss": 1.5142, "step": 2334 }, { "epoch": 0.1787080973519057, "grad_norm": 0.5993027093327249, "learning_rate": 9.545432726454142e-06, "loss": 1.3073, "step": 2335 }, { "epoch": 0.1787846318689729, "grad_norm": 0.5461049064056324, "learning_rate": 9.544973549212602e-06, "loss": 1.4476, "step": 2336 }, { "epoch": 0.1788611663860401, "grad_norm": 0.5985995098774358, "learning_rate": 9.54451415415729e-06, "loss": 1.3351, "step": 2337 }, { "epoch": 0.1789377009031073, "grad_norm": 1.108276614325986, "learning_rate": 9.544054541316434e-06, "loss": 1.704, "step": 2338 }, { "epoch": 0.1790142354201745, "grad_norm": 0.6121338173952869, "learning_rate": 9.54359471071827e-06, "loss": 1.5954, "step": 2339 }, { "epoch": 0.17909076993724168, "grad_norm": 1.696445038418981, "learning_rate": 9.543134662391054e-06, "loss": 1.9209, "step": 2340 }, { "epoch": 0.1791673044543089, "grad_norm": 0.6690620382072475, "learning_rate": 9.542674396363052e-06, "loss": 1.6518, "step": 2341 }, { "epoch": 0.17924383897137608, "grad_norm": 0.5914768103766508, "learning_rate": 9.542213912662544e-06, "loss": 1.4514, "step": 2342 }, { "epoch": 0.1793203734884433, "grad_norm": 0.5584929278725109, "learning_rate": 9.541753211317826e-06, "loss": 1.1873, "step": 2343 }, { "epoch": 0.17939690800551047, "grad_norm": 0.597606176868557, "learning_rate": 9.541292292357202e-06, "loss": 1.4406, "step": 2344 }, { "epoch": 0.1794734425225777, "grad_norm": 0.6594734185021864, "learning_rate": 9.540831155808993e-06, "loss": 1.644, "step": 2345 }, { "epoch": 0.17954997703964487, "grad_norm": 0.610247580377697, "learning_rate": 9.540369801701532e-06, "loss": 1.48, "step": 2346 }, { "epoch": 0.17962651155671208, "grad_norm": 0.6805948006543439, "learning_rate": 9.539908230063168e-06, "loss": 1.5723, "step": 2347 }, { "epoch": 0.17970304607377927, "grad_norm": 0.6691857452595145, "learning_rate": 9.53944644092226e-06, "loss": 1.373, "step": 2348 }, { "epoch": 0.17977958059084648, "grad_norm": 0.6133538210852416, "learning_rate": 9.53898443430718e-06, "loss": 1.1734, "step": 2349 }, { "epoch": 0.17985611510791366, "grad_norm": 0.582199483629905, "learning_rate": 9.538522210246319e-06, "loss": 1.3445, "step": 2350 }, { "epoch": 0.17993264962498087, "grad_norm": 0.5733663167789067, "learning_rate": 9.538059768768072e-06, "loss": 1.3761, "step": 2351 }, { "epoch": 0.18000918414204806, "grad_norm": 0.6842577608740041, "learning_rate": 9.53759710990086e-06, "loss": 1.4798, "step": 2352 }, { "epoch": 0.18008571865911527, "grad_norm": 0.7967453446075092, "learning_rate": 9.537134233673104e-06, "loss": 1.3545, "step": 2353 }, { "epoch": 0.18016225317618245, "grad_norm": 0.728953104998555, "learning_rate": 9.536671140113248e-06, "loss": 1.6714, "step": 2354 }, { "epoch": 0.18023878769324966, "grad_norm": 0.6195790224837153, "learning_rate": 9.536207829249742e-06, "loss": 1.4583, "step": 2355 }, { "epoch": 0.18031532221031685, "grad_norm": 0.6142936145123259, "learning_rate": 9.53574430111106e-06, "loss": 1.4951, "step": 2356 }, { "epoch": 0.18039185672738406, "grad_norm": 0.9647537711417203, "learning_rate": 9.535280555725674e-06, "loss": 1.6919, "step": 2357 }, { "epoch": 0.18046839124445124, "grad_norm": 0.7513144677290576, "learning_rate": 9.534816593122083e-06, "loss": 1.4368, "step": 2358 }, { "epoch": 0.18054492576151845, "grad_norm": 0.6557537547113953, "learning_rate": 9.534352413328795e-06, "loss": 1.4447, "step": 2359 }, { "epoch": 0.18062146027858564, "grad_norm": 0.979728766879998, "learning_rate": 9.533888016374327e-06, "loss": 1.6207, "step": 2360 }, { "epoch": 0.18069799479565285, "grad_norm": 0.5482964475142991, "learning_rate": 9.533423402287216e-06, "loss": 1.2495, "step": 2361 }, { "epoch": 0.18077452931272003, "grad_norm": 1.0423806548718693, "learning_rate": 9.53295857109601e-06, "loss": 1.6742, "step": 2362 }, { "epoch": 0.18085106382978725, "grad_norm": 0.6610958764250465, "learning_rate": 9.532493522829265e-06, "loss": 1.465, "step": 2363 }, { "epoch": 0.18092759834685443, "grad_norm": 0.6394831224241366, "learning_rate": 9.53202825751556e-06, "loss": 1.6026, "step": 2364 }, { "epoch": 0.18100413286392164, "grad_norm": 0.9558279319871996, "learning_rate": 9.531562775183479e-06, "loss": 1.4974, "step": 2365 }, { "epoch": 0.18108066738098882, "grad_norm": 0.5785721281475003, "learning_rate": 9.531097075861624e-06, "loss": 1.4907, "step": 2366 }, { "epoch": 0.18115720189805604, "grad_norm": 0.6098981342942772, "learning_rate": 9.530631159578608e-06, "loss": 1.5742, "step": 2367 }, { "epoch": 0.18123373641512322, "grad_norm": 0.5497778839084487, "learning_rate": 9.53016502636306e-06, "loss": 1.2926, "step": 2368 }, { "epoch": 0.18131027093219043, "grad_norm": 0.5674323436311707, "learning_rate": 9.52969867624362e-06, "loss": 1.2911, "step": 2369 }, { "epoch": 0.18138680544925762, "grad_norm": 0.8057163629634397, "learning_rate": 9.52923210924894e-06, "loss": 1.7724, "step": 2370 }, { "epoch": 0.1814633399663248, "grad_norm": 0.6350966918084169, "learning_rate": 9.52876532540769e-06, "loss": 1.6565, "step": 2371 }, { "epoch": 0.181539874483392, "grad_norm": 0.723545418672838, "learning_rate": 9.528298324748549e-06, "loss": 1.6687, "step": 2372 }, { "epoch": 0.1816164090004592, "grad_norm": 0.5281459844392881, "learning_rate": 9.527831107300211e-06, "loss": 0.9774, "step": 2373 }, { "epoch": 0.1816929435175264, "grad_norm": 0.6908098425125048, "learning_rate": 9.527363673091382e-06, "loss": 1.7873, "step": 2374 }, { "epoch": 0.1817694780345936, "grad_norm": 0.48008905427826987, "learning_rate": 9.526896022150788e-06, "loss": 1.0395, "step": 2375 }, { "epoch": 0.1818460125516608, "grad_norm": 0.577722206934639, "learning_rate": 9.526428154507157e-06, "loss": 1.3504, "step": 2376 }, { "epoch": 0.18192254706872799, "grad_norm": 0.6437370228663707, "learning_rate": 9.525960070189237e-06, "loss": 1.4002, "step": 2377 }, { "epoch": 0.1819990815857952, "grad_norm": 0.612851631032631, "learning_rate": 9.52549176922579e-06, "loss": 1.4968, "step": 2378 }, { "epoch": 0.18207561610286238, "grad_norm": 0.5917431199419625, "learning_rate": 9.525023251645592e-06, "loss": 1.348, "step": 2379 }, { "epoch": 0.1821521506199296, "grad_norm": 0.658817958758677, "learning_rate": 9.524554517477426e-06, "loss": 1.5121, "step": 2380 }, { "epoch": 0.18222868513699678, "grad_norm": 0.6015859062988862, "learning_rate": 9.524085566750094e-06, "loss": 1.3956, "step": 2381 }, { "epoch": 0.182305219654064, "grad_norm": 0.6029346469225427, "learning_rate": 9.523616399492409e-06, "loss": 1.5184, "step": 2382 }, { "epoch": 0.18238175417113117, "grad_norm": 0.5499337407356418, "learning_rate": 9.523147015733199e-06, "loss": 1.2488, "step": 2383 }, { "epoch": 0.18245828868819838, "grad_norm": 0.5986167836681713, "learning_rate": 9.522677415501304e-06, "loss": 1.2769, "step": 2384 }, { "epoch": 0.18253482320526557, "grad_norm": 0.596188343380193, "learning_rate": 9.52220759882558e-06, "loss": 1.4065, "step": 2385 }, { "epoch": 0.18261135772233278, "grad_norm": 1.163983017158942, "learning_rate": 9.521737565734888e-06, "loss": 1.609, "step": 2386 }, { "epoch": 0.18268789223939996, "grad_norm": 0.6985797470518043, "learning_rate": 9.521267316258116e-06, "loss": 1.5973, "step": 2387 }, { "epoch": 0.18276442675646717, "grad_norm": 0.7588428689356486, "learning_rate": 9.520796850424152e-06, "loss": 1.6652, "step": 2388 }, { "epoch": 0.18284096127353436, "grad_norm": 0.5530912234686998, "learning_rate": 9.520326168261904e-06, "loss": 1.2815, "step": 2389 }, { "epoch": 0.18291749579060157, "grad_norm": 0.6385504216998373, "learning_rate": 9.51985526980029e-06, "loss": 1.54, "step": 2390 }, { "epoch": 0.18299403030766875, "grad_norm": 1.309048739056793, "learning_rate": 9.51938415506825e-06, "loss": 1.6059, "step": 2391 }, { "epoch": 0.18307056482473597, "grad_norm": 0.6616397582334896, "learning_rate": 9.518912824094723e-06, "loss": 1.502, "step": 2392 }, { "epoch": 0.18314709934180315, "grad_norm": 0.6562214322530809, "learning_rate": 9.518441276908676e-06, "loss": 1.679, "step": 2393 }, { "epoch": 0.18322363385887036, "grad_norm": 0.6283609203446269, "learning_rate": 9.517969513539077e-06, "loss": 1.3457, "step": 2394 }, { "epoch": 0.18330016837593754, "grad_norm": 0.5561611812043589, "learning_rate": 9.517497534014915e-06, "loss": 1.383, "step": 2395 }, { "epoch": 0.18337670289300476, "grad_norm": 0.5147981787441801, "learning_rate": 9.517025338365187e-06, "loss": 1.087, "step": 2396 }, { "epoch": 0.18345323741007194, "grad_norm": 0.6586568995418777, "learning_rate": 9.516552926618911e-06, "loss": 1.6192, "step": 2397 }, { "epoch": 0.18352977192713915, "grad_norm": 0.5722582397771262, "learning_rate": 9.51608029880511e-06, "loss": 1.2675, "step": 2398 }, { "epoch": 0.18360630644420634, "grad_norm": 0.6470870846180744, "learning_rate": 9.515607454952824e-06, "loss": 1.441, "step": 2399 }, { "epoch": 0.18368284096127355, "grad_norm": 0.6345087433139583, "learning_rate": 9.515134395091106e-06, "loss": 1.4983, "step": 2400 }, { "epoch": 0.18375937547834073, "grad_norm": 0.6708888008498411, "learning_rate": 9.514661119249024e-06, "loss": 1.7371, "step": 2401 }, { "epoch": 0.18383590999540794, "grad_norm": 0.5786689909376587, "learning_rate": 9.514187627455655e-06, "loss": 1.2667, "step": 2402 }, { "epoch": 0.18391244451247513, "grad_norm": 0.5986277461285956, "learning_rate": 9.513713919740091e-06, "loss": 1.5372, "step": 2403 }, { "epoch": 0.1839889790295423, "grad_norm": 0.6448819753068258, "learning_rate": 9.513239996131443e-06, "loss": 1.6152, "step": 2404 }, { "epoch": 0.18406551354660952, "grad_norm": 0.778041815755431, "learning_rate": 9.512765856658824e-06, "loss": 1.6417, "step": 2405 }, { "epoch": 0.1841420480636767, "grad_norm": 0.6548296202449087, "learning_rate": 9.512291501351372e-06, "loss": 1.6211, "step": 2406 }, { "epoch": 0.18421858258074392, "grad_norm": 0.5545961935038427, "learning_rate": 9.511816930238228e-06, "loss": 1.3243, "step": 2407 }, { "epoch": 0.1842951170978111, "grad_norm": 0.6510491210698165, "learning_rate": 9.511342143348556e-06, "loss": 1.6174, "step": 2408 }, { "epoch": 0.1843716516148783, "grad_norm": 0.710361256472586, "learning_rate": 9.510867140711524e-06, "loss": 1.6623, "step": 2409 }, { "epoch": 0.1844481861319455, "grad_norm": 1.6961240459288918, "learning_rate": 9.51039192235632e-06, "loss": 1.3556, "step": 2410 }, { "epoch": 0.1845247206490127, "grad_norm": 0.8098554050427347, "learning_rate": 9.509916488312143e-06, "loss": 1.5116, "step": 2411 }, { "epoch": 0.1846012551660799, "grad_norm": 0.5389920826094513, "learning_rate": 9.509440838608204e-06, "loss": 1.2583, "step": 2412 }, { "epoch": 0.1846777896831471, "grad_norm": 0.6717820230122615, "learning_rate": 9.508964973273728e-06, "loss": 1.4855, "step": 2413 }, { "epoch": 0.1847543242002143, "grad_norm": 0.6659215776118045, "learning_rate": 9.508488892337956e-06, "loss": 1.5305, "step": 2414 }, { "epoch": 0.1848308587172815, "grad_norm": 1.556117768556538, "learning_rate": 9.508012595830135e-06, "loss": 1.8647, "step": 2415 }, { "epoch": 0.18490739323434868, "grad_norm": 0.6490089478907636, "learning_rate": 9.507536083779534e-06, "loss": 1.6064, "step": 2416 }, { "epoch": 0.1849839277514159, "grad_norm": 0.5657504404042114, "learning_rate": 9.507059356215432e-06, "loss": 1.4171, "step": 2417 }, { "epoch": 0.18506046226848308, "grad_norm": 0.6370577802200875, "learning_rate": 9.506582413167119e-06, "loss": 1.6157, "step": 2418 }, { "epoch": 0.1851369967855503, "grad_norm": 0.4797721543785514, "learning_rate": 9.506105254663899e-06, "loss": 0.9859, "step": 2419 }, { "epoch": 0.18521353130261747, "grad_norm": 0.6130664519491787, "learning_rate": 9.50562788073509e-06, "loss": 1.4406, "step": 2420 }, { "epoch": 0.18529006581968469, "grad_norm": 0.5255220517404666, "learning_rate": 9.505150291410025e-06, "loss": 1.1556, "step": 2421 }, { "epoch": 0.18536660033675187, "grad_norm": 0.6363802391534515, "learning_rate": 9.504672486718048e-06, "loss": 1.2298, "step": 2422 }, { "epoch": 0.18544313485381908, "grad_norm": 0.5592212481272689, "learning_rate": 9.504194466688516e-06, "loss": 1.2148, "step": 2423 }, { "epoch": 0.18551966937088626, "grad_norm": 1.0730210318383886, "learning_rate": 9.5037162313508e-06, "loss": 1.7572, "step": 2424 }, { "epoch": 0.18559620388795348, "grad_norm": 0.9880060565544898, "learning_rate": 9.503237780734286e-06, "loss": 1.5957, "step": 2425 }, { "epoch": 0.18567273840502066, "grad_norm": 0.6038688244474786, "learning_rate": 9.50275911486837e-06, "loss": 1.5946, "step": 2426 }, { "epoch": 0.18574927292208787, "grad_norm": 0.5217413966937489, "learning_rate": 9.502280233782463e-06, "loss": 1.2063, "step": 2427 }, { "epoch": 0.18582580743915506, "grad_norm": 0.5724580292606068, "learning_rate": 9.501801137505986e-06, "loss": 1.3817, "step": 2428 }, { "epoch": 0.18590234195622227, "grad_norm": 0.6493020909927663, "learning_rate": 9.501321826068381e-06, "loss": 1.5087, "step": 2429 }, { "epoch": 0.18597887647328945, "grad_norm": 0.6110785625642519, "learning_rate": 9.500842299499095e-06, "loss": 1.5099, "step": 2430 }, { "epoch": 0.18605541099035666, "grad_norm": 0.6092718931006792, "learning_rate": 9.500362557827594e-06, "loss": 1.4512, "step": 2431 }, { "epoch": 0.18613194550742385, "grad_norm": 0.7062612954848063, "learning_rate": 9.499882601083352e-06, "loss": 1.366, "step": 2432 }, { "epoch": 0.18620848002449106, "grad_norm": 0.5981068707391418, "learning_rate": 9.499402429295862e-06, "loss": 1.386, "step": 2433 }, { "epoch": 0.18628501454155824, "grad_norm": 0.6038547112975389, "learning_rate": 9.498922042494623e-06, "loss": 1.4165, "step": 2434 }, { "epoch": 0.18636154905862545, "grad_norm": 0.6098110665391062, "learning_rate": 9.498441440709156e-06, "loss": 1.4774, "step": 2435 }, { "epoch": 0.18643808357569264, "grad_norm": 0.6286359829170229, "learning_rate": 9.497960623968987e-06, "loss": 1.4633, "step": 2436 }, { "epoch": 0.18651461809275985, "grad_norm": 0.9183781531740246, "learning_rate": 9.497479592303661e-06, "loss": 1.5342, "step": 2437 }, { "epoch": 0.18659115260982703, "grad_norm": 0.6294370275374304, "learning_rate": 9.496998345742734e-06, "loss": 1.4566, "step": 2438 }, { "epoch": 0.18666768712689422, "grad_norm": 0.5707825719672719, "learning_rate": 9.496516884315775e-06, "loss": 1.3671, "step": 2439 }, { "epoch": 0.18674422164396143, "grad_norm": 0.6394362750317522, "learning_rate": 9.496035208052364e-06, "loss": 1.424, "step": 2440 }, { "epoch": 0.1868207561610286, "grad_norm": 0.4968402405927793, "learning_rate": 9.4955533169821e-06, "loss": 1.1009, "step": 2441 }, { "epoch": 0.18689729067809582, "grad_norm": 0.8469301389904159, "learning_rate": 9.49507121113459e-06, "loss": 1.7998, "step": 2442 }, { "epoch": 0.186973825195163, "grad_norm": 0.6595105410200158, "learning_rate": 9.494588890539457e-06, "loss": 1.5378, "step": 2443 }, { "epoch": 0.18705035971223022, "grad_norm": 0.6433921064217485, "learning_rate": 9.494106355226335e-06, "loss": 1.3114, "step": 2444 }, { "epoch": 0.1871268942292974, "grad_norm": 0.578088543228235, "learning_rate": 9.493623605224875e-06, "loss": 1.3055, "step": 2445 }, { "epoch": 0.18720342874636461, "grad_norm": 0.6168957379301864, "learning_rate": 9.493140640564734e-06, "loss": 1.4106, "step": 2446 }, { "epoch": 0.1872799632634318, "grad_norm": 0.6214098208389461, "learning_rate": 9.492657461275592e-06, "loss": 1.2675, "step": 2447 }, { "epoch": 0.187356497780499, "grad_norm": 0.5644918679036085, "learning_rate": 9.492174067387133e-06, "loss": 1.2853, "step": 2448 }, { "epoch": 0.1874330322975662, "grad_norm": 0.8083268309267833, "learning_rate": 9.491690458929061e-06, "loss": 1.3676, "step": 2449 }, { "epoch": 0.1875095668146334, "grad_norm": 0.6030038491948156, "learning_rate": 9.491206635931088e-06, "loss": 1.3179, "step": 2450 }, { "epoch": 0.1875861013317006, "grad_norm": 0.8855953653031812, "learning_rate": 9.490722598422942e-06, "loss": 1.7944, "step": 2451 }, { "epoch": 0.1876626358487678, "grad_norm": 0.6493713698959928, "learning_rate": 9.490238346434366e-06, "loss": 1.5245, "step": 2452 }, { "epoch": 0.18773917036583498, "grad_norm": 0.5548451263688101, "learning_rate": 9.48975387999511e-06, "loss": 1.2271, "step": 2453 }, { "epoch": 0.1878157048829022, "grad_norm": 0.6095071647355932, "learning_rate": 9.489269199134945e-06, "loss": 1.5932, "step": 2454 }, { "epoch": 0.18789223939996938, "grad_norm": 0.8389091346635167, "learning_rate": 9.488784303883649e-06, "loss": 1.5132, "step": 2455 }, { "epoch": 0.1879687739170366, "grad_norm": 0.5980147861522557, "learning_rate": 9.488299194271015e-06, "loss": 1.3459, "step": 2456 }, { "epoch": 0.18804530843410378, "grad_norm": 0.6927847432637013, "learning_rate": 9.487813870326851e-06, "loss": 1.4196, "step": 2457 }, { "epoch": 0.188121842951171, "grad_norm": 0.5549214514920853, "learning_rate": 9.487328332080978e-06, "loss": 1.3942, "step": 2458 }, { "epoch": 0.18819837746823817, "grad_norm": 0.556097492161671, "learning_rate": 9.486842579563226e-06, "loss": 1.3339, "step": 2459 }, { "epoch": 0.18827491198530538, "grad_norm": 0.5286692322725117, "learning_rate": 9.48635661280344e-06, "loss": 1.1234, "step": 2460 }, { "epoch": 0.18835144650237257, "grad_norm": 0.6652058660024308, "learning_rate": 9.485870431831486e-06, "loss": 1.7973, "step": 2461 }, { "epoch": 0.18842798101943978, "grad_norm": 0.5906037191962538, "learning_rate": 9.485384036677228e-06, "loss": 1.4287, "step": 2462 }, { "epoch": 0.18850451553650696, "grad_norm": 0.6044134353047816, "learning_rate": 9.484897427370556e-06, "loss": 1.3621, "step": 2463 }, { "epoch": 0.18858105005357417, "grad_norm": 0.9433450356268368, "learning_rate": 9.48441060394137e-06, "loss": 1.4759, "step": 2464 }, { "epoch": 0.18865758457064136, "grad_norm": 0.5694928057065611, "learning_rate": 9.483923566419579e-06, "loss": 1.3756, "step": 2465 }, { "epoch": 0.18873411908770857, "grad_norm": 0.4805229305659048, "learning_rate": 9.48343631483511e-06, "loss": 1.146, "step": 2466 }, { "epoch": 0.18881065360477575, "grad_norm": 1.167714949643102, "learning_rate": 9.4829488492179e-06, "loss": 1.845, "step": 2467 }, { "epoch": 0.18888718812184296, "grad_norm": 0.6563585654936648, "learning_rate": 9.482461169597898e-06, "loss": 1.777, "step": 2468 }, { "epoch": 0.18896372263891015, "grad_norm": 0.610781802970201, "learning_rate": 9.481973276005074e-06, "loss": 1.3752, "step": 2469 }, { "epoch": 0.18904025715597736, "grad_norm": 0.6999425960460794, "learning_rate": 9.481485168469401e-06, "loss": 1.4101, "step": 2470 }, { "epoch": 0.18911679167304454, "grad_norm": 0.5921137327381221, "learning_rate": 9.480996847020874e-06, "loss": 1.4089, "step": 2471 }, { "epoch": 0.18919332619011173, "grad_norm": 0.6890959783088183, "learning_rate": 9.480508311689494e-06, "loss": 1.8513, "step": 2472 }, { "epoch": 0.18926986070717894, "grad_norm": 0.5284425307880051, "learning_rate": 9.480019562505277e-06, "loss": 1.2871, "step": 2473 }, { "epoch": 0.18934639522424612, "grad_norm": 0.6717781407951473, "learning_rate": 9.479530599498256e-06, "loss": 1.5164, "step": 2474 }, { "epoch": 0.18942292974131333, "grad_norm": 0.5859569539570985, "learning_rate": 9.479041422698473e-06, "loss": 1.3368, "step": 2475 }, { "epoch": 0.18949946425838052, "grad_norm": 0.5249542702777088, "learning_rate": 9.478552032135984e-06, "loss": 1.2592, "step": 2476 }, { "epoch": 0.18957599877544773, "grad_norm": 0.6529469241084533, "learning_rate": 9.478062427840859e-06, "loss": 1.6818, "step": 2477 }, { "epoch": 0.1896525332925149, "grad_norm": 0.5512235722353415, "learning_rate": 9.477572609843182e-06, "loss": 1.3656, "step": 2478 }, { "epoch": 0.18972906780958212, "grad_norm": 0.5947523224377826, "learning_rate": 9.477082578173047e-06, "loss": 1.6199, "step": 2479 }, { "epoch": 0.1898056023266493, "grad_norm": 0.6164129575341002, "learning_rate": 9.476592332860565e-06, "loss": 1.4913, "step": 2480 }, { "epoch": 0.18988213684371652, "grad_norm": 0.6529864083213912, "learning_rate": 9.476101873935857e-06, "loss": 1.6054, "step": 2481 }, { "epoch": 0.1899586713607837, "grad_norm": 0.6769730900520372, "learning_rate": 9.475611201429058e-06, "loss": 1.6765, "step": 2482 }, { "epoch": 0.19003520587785092, "grad_norm": 0.7209224664186418, "learning_rate": 9.475120315370317e-06, "loss": 1.8516, "step": 2483 }, { "epoch": 0.1901117403949181, "grad_norm": 0.5726479365984373, "learning_rate": 9.474629215789795e-06, "loss": 1.2677, "step": 2484 }, { "epoch": 0.1901882749119853, "grad_norm": 0.6613337223578758, "learning_rate": 9.474137902717667e-06, "loss": 1.5596, "step": 2485 }, { "epoch": 0.1902648094290525, "grad_norm": 0.6153649691764422, "learning_rate": 9.473646376184118e-06, "loss": 1.5056, "step": 2486 }, { "epoch": 0.1903413439461197, "grad_norm": 0.5909610422578743, "learning_rate": 9.473154636219356e-06, "loss": 1.4442, "step": 2487 }, { "epoch": 0.1904178784631869, "grad_norm": 0.5252129187377571, "learning_rate": 9.472662682853585e-06, "loss": 1.2037, "step": 2488 }, { "epoch": 0.1904944129802541, "grad_norm": 0.5426587540027046, "learning_rate": 9.472170516117041e-06, "loss": 1.2693, "step": 2489 }, { "epoch": 0.19057094749732129, "grad_norm": 0.658473171167478, "learning_rate": 9.47167813603996e-06, "loss": 1.5747, "step": 2490 }, { "epoch": 0.1906474820143885, "grad_norm": 0.6076399217346934, "learning_rate": 9.471185542652595e-06, "loss": 1.6005, "step": 2491 }, { "epoch": 0.19072401653145568, "grad_norm": 0.7893590691470047, "learning_rate": 9.470692735985216e-06, "loss": 1.2719, "step": 2492 }, { "epoch": 0.1908005510485229, "grad_norm": 0.9101622566175587, "learning_rate": 9.470199716068098e-06, "loss": 1.7875, "step": 2493 }, { "epoch": 0.19087708556559008, "grad_norm": 0.6097482001389499, "learning_rate": 9.469706482931535e-06, "loss": 1.3731, "step": 2494 }, { "epoch": 0.1909536200826573, "grad_norm": 0.6027840905602218, "learning_rate": 9.469213036605835e-06, "loss": 1.3983, "step": 2495 }, { "epoch": 0.19103015459972447, "grad_norm": 0.5833355196597042, "learning_rate": 9.468719377121315e-06, "loss": 1.4348, "step": 2496 }, { "epoch": 0.19110668911679168, "grad_norm": 0.6360304975614391, "learning_rate": 9.468225504508304e-06, "loss": 1.1481, "step": 2497 }, { "epoch": 0.19118322363385887, "grad_norm": 0.6334782400377394, "learning_rate": 9.467731418797152e-06, "loss": 1.4139, "step": 2498 }, { "epoch": 0.19125975815092608, "grad_norm": 0.797973811013684, "learning_rate": 9.467237120018217e-06, "loss": 1.6425, "step": 2499 }, { "epoch": 0.19133629266799326, "grad_norm": 0.6851880086888095, "learning_rate": 9.466742608201867e-06, "loss": 1.4286, "step": 2500 }, { "epoch": 0.19141282718506047, "grad_norm": 0.5719828273771068, "learning_rate": 9.466247883378485e-06, "loss": 1.4805, "step": 2501 }, { "epoch": 0.19148936170212766, "grad_norm": 0.6162622079937263, "learning_rate": 9.465752945578473e-06, "loss": 1.4833, "step": 2502 }, { "epoch": 0.19156589621919487, "grad_norm": 0.6262131166534103, "learning_rate": 9.46525779483224e-06, "loss": 1.5212, "step": 2503 }, { "epoch": 0.19164243073626205, "grad_norm": 0.614272779719569, "learning_rate": 9.464762431170208e-06, "loss": 1.2764, "step": 2504 }, { "epoch": 0.19171896525332924, "grad_norm": 0.6190016474204563, "learning_rate": 9.464266854622815e-06, "loss": 1.4133, "step": 2505 }, { "epoch": 0.19179549977039645, "grad_norm": 0.6667257019722221, "learning_rate": 9.46377106522051e-06, "loss": 1.7757, "step": 2506 }, { "epoch": 0.19187203428746363, "grad_norm": 0.5693497371820471, "learning_rate": 9.463275062993757e-06, "loss": 1.4701, "step": 2507 }, { "epoch": 0.19194856880453084, "grad_norm": 0.5974638552761726, "learning_rate": 9.46277884797303e-06, "loss": 1.2913, "step": 2508 }, { "epoch": 0.19202510332159803, "grad_norm": 0.6112813819239395, "learning_rate": 9.462282420188818e-06, "loss": 1.5397, "step": 2509 }, { "epoch": 0.19210163783866524, "grad_norm": 0.7446488030245166, "learning_rate": 9.461785779671624e-06, "loss": 1.5476, "step": 2510 }, { "epoch": 0.19217817235573242, "grad_norm": 0.5821708494467462, "learning_rate": 9.461288926451965e-06, "loss": 1.2848, "step": 2511 }, { "epoch": 0.19225470687279964, "grad_norm": 0.6160834916478276, "learning_rate": 9.460791860560365e-06, "loss": 1.3646, "step": 2512 }, { "epoch": 0.19233124138986682, "grad_norm": 0.5917697671851946, "learning_rate": 9.460294582027368e-06, "loss": 1.53, "step": 2513 }, { "epoch": 0.19240777590693403, "grad_norm": 0.5631853868229375, "learning_rate": 9.459797090883529e-06, "loss": 1.3936, "step": 2514 }, { "epoch": 0.19248431042400121, "grad_norm": 0.6845209974623552, "learning_rate": 9.459299387159412e-06, "loss": 1.6966, "step": 2515 }, { "epoch": 0.19256084494106843, "grad_norm": 0.719619042495395, "learning_rate": 9.4588014708856e-06, "loss": 1.8275, "step": 2516 }, { "epoch": 0.1926373794581356, "grad_norm": 0.5408050041870515, "learning_rate": 9.458303342092684e-06, "loss": 1.184, "step": 2517 }, { "epoch": 0.19271391397520282, "grad_norm": 0.7076857530180466, "learning_rate": 9.457805000811276e-06, "loss": 1.5659, "step": 2518 }, { "epoch": 0.19279044849227, "grad_norm": 0.6187344211256055, "learning_rate": 9.457306447071989e-06, "loss": 1.4668, "step": 2519 }, { "epoch": 0.19286698300933722, "grad_norm": 0.5586383899144524, "learning_rate": 9.45680768090546e-06, "loss": 1.2872, "step": 2520 }, { "epoch": 0.1929435175264044, "grad_norm": 0.6037997721213386, "learning_rate": 9.456308702342333e-06, "loss": 1.2344, "step": 2521 }, { "epoch": 0.1930200520434716, "grad_norm": 0.507994755571701, "learning_rate": 9.455809511413265e-06, "loss": 1.186, "step": 2522 }, { "epoch": 0.1930965865605388, "grad_norm": 0.7466951700144941, "learning_rate": 9.455310108148931e-06, "loss": 1.3478, "step": 2523 }, { "epoch": 0.193173121077606, "grad_norm": 0.5914962003901416, "learning_rate": 9.454810492580015e-06, "loss": 1.4167, "step": 2524 }, { "epoch": 0.1932496555946732, "grad_norm": 0.6652380686931048, "learning_rate": 9.454310664737216e-06, "loss": 1.4545, "step": 2525 }, { "epoch": 0.1933261901117404, "grad_norm": 0.5200810156030573, "learning_rate": 9.45381062465124e-06, "loss": 1.0305, "step": 2526 }, { "epoch": 0.1934027246288076, "grad_norm": 0.6644339849003862, "learning_rate": 9.453310372352816e-06, "loss": 1.6123, "step": 2527 }, { "epoch": 0.1934792591458748, "grad_norm": 0.679127401239907, "learning_rate": 9.45280990787268e-06, "loss": 1.6949, "step": 2528 }, { "epoch": 0.19355579366294198, "grad_norm": 0.6614964320983711, "learning_rate": 9.452309231241579e-06, "loss": 1.554, "step": 2529 }, { "epoch": 0.1936323281800092, "grad_norm": 0.6189541428452106, "learning_rate": 9.451808342490281e-06, "loss": 1.4388, "step": 2530 }, { "epoch": 0.19370886269707638, "grad_norm": 0.560521796503492, "learning_rate": 9.451307241649558e-06, "loss": 1.2986, "step": 2531 }, { "epoch": 0.1937853972141436, "grad_norm": 0.5991785384545596, "learning_rate": 9.4508059287502e-06, "loss": 1.6427, "step": 2532 }, { "epoch": 0.19386193173121077, "grad_norm": 0.7631339283593587, "learning_rate": 9.450304403823011e-06, "loss": 1.4445, "step": 2533 }, { "epoch": 0.19393846624827799, "grad_norm": 0.590673389969192, "learning_rate": 9.449802666898803e-06, "loss": 1.4282, "step": 2534 }, { "epoch": 0.19401500076534517, "grad_norm": 0.6585176068725829, "learning_rate": 9.449300718008409e-06, "loss": 1.7989, "step": 2535 }, { "epoch": 0.19409153528241238, "grad_norm": 0.6018086637197837, "learning_rate": 9.448798557182667e-06, "loss": 1.4286, "step": 2536 }, { "epoch": 0.19416806979947956, "grad_norm": 0.5810717688810491, "learning_rate": 9.448296184452432e-06, "loss": 1.2542, "step": 2537 }, { "epoch": 0.19424460431654678, "grad_norm": 0.5304819127447731, "learning_rate": 9.44779359984857e-06, "loss": 1.1685, "step": 2538 }, { "epoch": 0.19432113883361396, "grad_norm": 1.2630132944237782, "learning_rate": 9.447290803401961e-06, "loss": 1.2956, "step": 2539 }, { "epoch": 0.19439767335068114, "grad_norm": 0.5805272185502862, "learning_rate": 9.4467877951435e-06, "loss": 1.3812, "step": 2540 }, { "epoch": 0.19447420786774836, "grad_norm": 0.5952210377211671, "learning_rate": 9.446284575104096e-06, "loss": 1.4455, "step": 2541 }, { "epoch": 0.19455074238481554, "grad_norm": 0.6485938644394366, "learning_rate": 9.445781143314663e-06, "loss": 1.6678, "step": 2542 }, { "epoch": 0.19462727690188275, "grad_norm": 0.6011757658536289, "learning_rate": 9.445277499806134e-06, "loss": 1.5546, "step": 2543 }, { "epoch": 0.19470381141894993, "grad_norm": 0.6341240200832878, "learning_rate": 9.444773644609457e-06, "loss": 1.6944, "step": 2544 }, { "epoch": 0.19478034593601715, "grad_norm": 0.5842999312160249, "learning_rate": 9.444269577755592e-06, "loss": 1.542, "step": 2545 }, { "epoch": 0.19485688045308433, "grad_norm": 0.5470196902541187, "learning_rate": 9.443765299275506e-06, "loss": 1.3784, "step": 2546 }, { "epoch": 0.19493341497015154, "grad_norm": 0.5979090981218105, "learning_rate": 9.443260809200184e-06, "loss": 1.4485, "step": 2547 }, { "epoch": 0.19500994948721873, "grad_norm": 0.5100614984751525, "learning_rate": 9.442756107560626e-06, "loss": 1.0741, "step": 2548 }, { "epoch": 0.19508648400428594, "grad_norm": 0.5668486468632276, "learning_rate": 9.442251194387841e-06, "loss": 1.5106, "step": 2549 }, { "epoch": 0.19516301852135312, "grad_norm": 0.6339298009604946, "learning_rate": 9.44174606971285e-06, "loss": 1.3843, "step": 2550 }, { "epoch": 0.19523955303842033, "grad_norm": 0.5933499922594547, "learning_rate": 9.441240733566694e-06, "loss": 1.4344, "step": 2551 }, { "epoch": 0.19531608755548752, "grad_norm": 0.6539722563043769, "learning_rate": 9.440735185980419e-06, "loss": 1.6603, "step": 2552 }, { "epoch": 0.19539262207255473, "grad_norm": 0.649708305685599, "learning_rate": 9.440229426985088e-06, "loss": 1.3824, "step": 2553 }, { "epoch": 0.1954691565896219, "grad_norm": 0.6596014338994824, "learning_rate": 9.439723456611777e-06, "loss": 1.5853, "step": 2554 }, { "epoch": 0.19554569110668912, "grad_norm": 0.5649696399835824, "learning_rate": 9.439217274891574e-06, "loss": 1.2388, "step": 2555 }, { "epoch": 0.1956222256237563, "grad_norm": 0.5463507138028975, "learning_rate": 9.43871088185558e-06, "loss": 1.252, "step": 2556 }, { "epoch": 0.19569876014082352, "grad_norm": 0.5916154689513652, "learning_rate": 9.43820427753491e-06, "loss": 1.3398, "step": 2557 }, { "epoch": 0.1957752946578907, "grad_norm": 1.0039884661961065, "learning_rate": 9.437697461960692e-06, "loss": 1.5932, "step": 2558 }, { "epoch": 0.19585182917495791, "grad_norm": 0.5501249053374825, "learning_rate": 9.437190435164065e-06, "loss": 1.2958, "step": 2559 }, { "epoch": 0.1959283636920251, "grad_norm": 0.7409209268558027, "learning_rate": 9.436683197176181e-06, "loss": 1.3304, "step": 2560 }, { "epoch": 0.1960048982090923, "grad_norm": 0.6037286091159368, "learning_rate": 9.436175748028207e-06, "loss": 1.5155, "step": 2561 }, { "epoch": 0.1960814327261595, "grad_norm": 0.5284063193949501, "learning_rate": 9.435668087751325e-06, "loss": 1.1593, "step": 2562 }, { "epoch": 0.1961579672432267, "grad_norm": 0.6172583951030409, "learning_rate": 9.435160216376724e-06, "loss": 1.4719, "step": 2563 }, { "epoch": 0.1962345017602939, "grad_norm": 0.4993146988327529, "learning_rate": 9.43465213393561e-06, "loss": 1.0943, "step": 2564 }, { "epoch": 0.1963110362773611, "grad_norm": 0.6243946721096696, "learning_rate": 9.434143840459202e-06, "loss": 1.2389, "step": 2565 }, { "epoch": 0.19638757079442828, "grad_norm": 0.6265205791412705, "learning_rate": 9.43363533597873e-06, "loss": 1.639, "step": 2566 }, { "epoch": 0.1964641053114955, "grad_norm": 0.5834165610020107, "learning_rate": 9.433126620525438e-06, "loss": 1.4183, "step": 2567 }, { "epoch": 0.19654063982856268, "grad_norm": 0.6111036041968181, "learning_rate": 9.432617694130583e-06, "loss": 1.4915, "step": 2568 }, { "epoch": 0.1966171743456299, "grad_norm": 0.7036605041262463, "learning_rate": 9.432108556825435e-06, "loss": 1.6471, "step": 2569 }, { "epoch": 0.19669370886269708, "grad_norm": 0.6015832666249928, "learning_rate": 9.43159920864128e-06, "loss": 1.3635, "step": 2570 }, { "epoch": 0.1967702433797643, "grad_norm": 0.576020355907279, "learning_rate": 9.431089649609408e-06, "loss": 1.3709, "step": 2571 }, { "epoch": 0.19684677789683147, "grad_norm": 0.586707738570242, "learning_rate": 9.430579879761133e-06, "loss": 1.5873, "step": 2572 }, { "epoch": 0.19692331241389865, "grad_norm": 0.6091404058672745, "learning_rate": 9.430069899127772e-06, "loss": 1.3057, "step": 2573 }, { "epoch": 0.19699984693096587, "grad_norm": 0.6272127265509118, "learning_rate": 9.429559707740664e-06, "loss": 1.2369, "step": 2574 }, { "epoch": 0.19707638144803305, "grad_norm": 0.5491965481406451, "learning_rate": 9.429049305631156e-06, "loss": 1.4191, "step": 2575 }, { "epoch": 0.19715291596510026, "grad_norm": 0.5306045142789302, "learning_rate": 9.428538692830608e-06, "loss": 1.2209, "step": 2576 }, { "epoch": 0.19722945048216745, "grad_norm": 0.6634093389398336, "learning_rate": 9.428027869370393e-06, "loss": 1.6614, "step": 2577 }, { "epoch": 0.19730598499923466, "grad_norm": 0.6381127267704699, "learning_rate": 9.427516835281896e-06, "loss": 1.5523, "step": 2578 }, { "epoch": 0.19738251951630184, "grad_norm": 0.48370309639403924, "learning_rate": 9.42700559059652e-06, "loss": 0.9404, "step": 2579 }, { "epoch": 0.19745905403336905, "grad_norm": 0.724554688710746, "learning_rate": 9.426494135345674e-06, "loss": 1.6743, "step": 2580 }, { "epoch": 0.19753558855043624, "grad_norm": 0.6061924389667218, "learning_rate": 9.425982469560787e-06, "loss": 1.3869, "step": 2581 }, { "epoch": 0.19761212306750345, "grad_norm": 0.7409107872563641, "learning_rate": 9.425470593273294e-06, "loss": 1.4652, "step": 2582 }, { "epoch": 0.19768865758457063, "grad_norm": 0.5560619794922308, "learning_rate": 9.424958506514647e-06, "loss": 1.3615, "step": 2583 }, { "epoch": 0.19776519210163784, "grad_norm": 0.6051864737734823, "learning_rate": 9.424446209316314e-06, "loss": 1.2429, "step": 2584 }, { "epoch": 0.19784172661870503, "grad_norm": 0.5829149479264629, "learning_rate": 9.423933701709766e-06, "loss": 1.4667, "step": 2585 }, { "epoch": 0.19791826113577224, "grad_norm": 0.5909917850527889, "learning_rate": 9.423420983726496e-06, "loss": 1.3115, "step": 2586 }, { "epoch": 0.19799479565283942, "grad_norm": 0.5958214371139319, "learning_rate": 9.422908055398007e-06, "loss": 1.3287, "step": 2587 }, { "epoch": 0.19807133016990663, "grad_norm": 0.5658747345892534, "learning_rate": 9.422394916755814e-06, "loss": 1.1652, "step": 2588 }, { "epoch": 0.19814786468697382, "grad_norm": 0.5331087401113809, "learning_rate": 9.421881567831448e-06, "loss": 1.2889, "step": 2589 }, { "epoch": 0.19822439920404103, "grad_norm": 0.614446931590165, "learning_rate": 9.421368008656448e-06, "loss": 1.474, "step": 2590 }, { "epoch": 0.1983009337211082, "grad_norm": 0.8689990786981426, "learning_rate": 9.42085423926237e-06, "loss": 1.7047, "step": 2591 }, { "epoch": 0.19837746823817543, "grad_norm": 0.5423420433876586, "learning_rate": 9.420340259680779e-06, "loss": 1.3152, "step": 2592 }, { "epoch": 0.1984540027552426, "grad_norm": 0.6314557964498965, "learning_rate": 9.41982606994326e-06, "loss": 1.5135, "step": 2593 }, { "epoch": 0.19853053727230982, "grad_norm": 0.5796910755966207, "learning_rate": 9.419311670081403e-06, "loss": 1.3654, "step": 2594 }, { "epoch": 0.198607071789377, "grad_norm": 0.5749661363128951, "learning_rate": 9.418797060126814e-06, "loss": 1.0576, "step": 2595 }, { "epoch": 0.19868360630644422, "grad_norm": 0.6734383096654731, "learning_rate": 9.418282240111115e-06, "loss": 1.3406, "step": 2596 }, { "epoch": 0.1987601408235114, "grad_norm": 0.5394542075090508, "learning_rate": 9.417767210065937e-06, "loss": 1.3714, "step": 2597 }, { "epoch": 0.1988366753405786, "grad_norm": 0.6085218083406126, "learning_rate": 9.417251970022924e-06, "loss": 1.4268, "step": 2598 }, { "epoch": 0.1989132098576458, "grad_norm": 0.6184321971628269, "learning_rate": 9.416736520013732e-06, "loss": 1.4957, "step": 2599 }, { "epoch": 0.198989744374713, "grad_norm": 0.5769982101626413, "learning_rate": 9.416220860070037e-06, "loss": 1.2824, "step": 2600 }, { "epoch": 0.1990662788917802, "grad_norm": 0.5977740053030296, "learning_rate": 9.415704990223518e-06, "loss": 1.5879, "step": 2601 }, { "epoch": 0.1991428134088474, "grad_norm": 0.5545856382687947, "learning_rate": 9.415188910505874e-06, "loss": 1.2986, "step": 2602 }, { "epoch": 0.19921934792591459, "grad_norm": 0.5608861145962232, "learning_rate": 9.414672620948815e-06, "loss": 1.2264, "step": 2603 }, { "epoch": 0.1992958824429818, "grad_norm": 0.6123415133415223, "learning_rate": 9.414156121584061e-06, "loss": 1.4856, "step": 2604 }, { "epoch": 0.19937241696004898, "grad_norm": 0.665594185061563, "learning_rate": 9.41363941244335e-06, "loss": 1.5904, "step": 2605 }, { "epoch": 0.1994489514771162, "grad_norm": 0.6232708602415248, "learning_rate": 9.413122493558426e-06, "loss": 1.5975, "step": 2606 }, { "epoch": 0.19952548599418338, "grad_norm": 0.6480569769938251, "learning_rate": 9.412605364961054e-06, "loss": 1.6973, "step": 2607 }, { "epoch": 0.19960202051125056, "grad_norm": 0.6351219199620362, "learning_rate": 9.412088026683008e-06, "loss": 1.3685, "step": 2608 }, { "epoch": 0.19967855502831777, "grad_norm": 0.6093444580104479, "learning_rate": 9.41157047875607e-06, "loss": 1.5156, "step": 2609 }, { "epoch": 0.19975508954538496, "grad_norm": 0.853396311697249, "learning_rate": 9.411052721212047e-06, "loss": 1.481, "step": 2610 }, { "epoch": 0.19983162406245217, "grad_norm": 0.5483991851407093, "learning_rate": 9.410534754082745e-06, "loss": 1.295, "step": 2611 }, { "epoch": 0.19990815857951935, "grad_norm": 0.9209709385921744, "learning_rate": 9.410016577399994e-06, "loss": 1.9543, "step": 2612 }, { "epoch": 0.19998469309658656, "grad_norm": 0.8250662982890737, "learning_rate": 9.40949819119563e-06, "loss": 1.6944, "step": 2613 }, { "epoch": 0.20006122761365375, "grad_norm": 0.47960600991260316, "learning_rate": 9.408979595501504e-06, "loss": 0.9028, "step": 2614 }, { "epoch": 0.20013776213072096, "grad_norm": 0.5857509496314587, "learning_rate": 9.40846079034948e-06, "loss": 1.4942, "step": 2615 }, { "epoch": 0.20021429664778814, "grad_norm": 0.6128545327723488, "learning_rate": 9.40794177577144e-06, "loss": 1.5034, "step": 2616 }, { "epoch": 0.20029083116485535, "grad_norm": 0.6880577473952008, "learning_rate": 9.407422551799264e-06, "loss": 1.67, "step": 2617 }, { "epoch": 0.20036736568192254, "grad_norm": 0.6209852242677998, "learning_rate": 9.406903118464862e-06, "loss": 1.4525, "step": 2618 }, { "epoch": 0.20044390019898975, "grad_norm": 0.6597284416548216, "learning_rate": 9.406383475800149e-06, "loss": 1.6352, "step": 2619 }, { "epoch": 0.20052043471605693, "grad_norm": 0.5739948811600588, "learning_rate": 9.40586362383705e-06, "loss": 1.4113, "step": 2620 }, { "epoch": 0.20059696923312414, "grad_norm": 0.6358682227426086, "learning_rate": 9.405343562607509e-06, "loss": 1.4857, "step": 2621 }, { "epoch": 0.20067350375019133, "grad_norm": 0.5044317342679558, "learning_rate": 9.404823292143481e-06, "loss": 1.1241, "step": 2622 }, { "epoch": 0.20075003826725854, "grad_norm": 0.5391139334661291, "learning_rate": 9.404302812476932e-06, "loss": 1.3185, "step": 2623 }, { "epoch": 0.20082657278432572, "grad_norm": 0.5580095500622325, "learning_rate": 9.403782123639838e-06, "loss": 1.1984, "step": 2624 }, { "epoch": 0.20090310730139294, "grad_norm": 0.6100279363625968, "learning_rate": 9.403261225664197e-06, "loss": 1.434, "step": 2625 }, { "epoch": 0.20097964181846012, "grad_norm": 0.656543951231507, "learning_rate": 9.402740118582012e-06, "loss": 1.5613, "step": 2626 }, { "epoch": 0.20105617633552733, "grad_norm": 0.5799526342720278, "learning_rate": 9.402218802425303e-06, "loss": 1.2385, "step": 2627 }, { "epoch": 0.20113271085259452, "grad_norm": 0.6295641035064069, "learning_rate": 9.401697277226098e-06, "loss": 1.6274, "step": 2628 }, { "epoch": 0.20120924536966173, "grad_norm": 0.6632456822236157, "learning_rate": 9.401175543016445e-06, "loss": 1.6046, "step": 2629 }, { "epoch": 0.2012857798867289, "grad_norm": 0.6522632311364002, "learning_rate": 9.400653599828397e-06, "loss": 1.4834, "step": 2630 }, { "epoch": 0.20136231440379612, "grad_norm": 0.5670509049943099, "learning_rate": 9.40013144769403e-06, "loss": 1.3342, "step": 2631 }, { "epoch": 0.2014388489208633, "grad_norm": 0.5990435702225954, "learning_rate": 9.399609086645418e-06, "loss": 1.5992, "step": 2632 }, { "epoch": 0.20151538343793052, "grad_norm": 0.4952595036313013, "learning_rate": 9.399086516714661e-06, "loss": 1.2721, "step": 2633 }, { "epoch": 0.2015919179549977, "grad_norm": 0.6395302069698245, "learning_rate": 9.398563737933871e-06, "loss": 1.5971, "step": 2634 }, { "epoch": 0.2016684524720649, "grad_norm": 0.8227433197354895, "learning_rate": 9.398040750335161e-06, "loss": 1.8469, "step": 2635 }, { "epoch": 0.2017449869891321, "grad_norm": 0.5793072049324473, "learning_rate": 9.39751755395067e-06, "loss": 1.2915, "step": 2636 }, { "epoch": 0.2018215215061993, "grad_norm": 0.634449520603955, "learning_rate": 9.396994148812546e-06, "loss": 1.628, "step": 2637 }, { "epoch": 0.2018980560232665, "grad_norm": 0.662140215106565, "learning_rate": 9.396470534952943e-06, "loss": 1.452, "step": 2638 }, { "epoch": 0.2019745905403337, "grad_norm": 0.7474043299354597, "learning_rate": 9.39594671240404e-06, "loss": 1.5771, "step": 2639 }, { "epoch": 0.2020511250574009, "grad_norm": 0.766104217183742, "learning_rate": 9.395422681198018e-06, "loss": 1.8997, "step": 2640 }, { "epoch": 0.20212765957446807, "grad_norm": 1.1051874672487156, "learning_rate": 9.394898441367076e-06, "loss": 1.55, "step": 2641 }, { "epoch": 0.20220419409153528, "grad_norm": 0.619769428124402, "learning_rate": 9.394373992943424e-06, "loss": 1.51, "step": 2642 }, { "epoch": 0.20228072860860247, "grad_norm": 0.6140293189192166, "learning_rate": 9.393849335959288e-06, "loss": 1.5587, "step": 2643 }, { "epoch": 0.20235726312566968, "grad_norm": 0.5871662439508523, "learning_rate": 9.393324470446903e-06, "loss": 1.3566, "step": 2644 }, { "epoch": 0.20243379764273686, "grad_norm": 0.7169776928869014, "learning_rate": 9.392799396438519e-06, "loss": 1.5059, "step": 2645 }, { "epoch": 0.20251033215980407, "grad_norm": 0.6308787003793694, "learning_rate": 9.392274113966395e-06, "loss": 1.3989, "step": 2646 }, { "epoch": 0.20258686667687126, "grad_norm": 0.5966321725825243, "learning_rate": 9.391748623062811e-06, "loss": 1.4051, "step": 2647 }, { "epoch": 0.20266340119393847, "grad_norm": 1.1501851832532464, "learning_rate": 9.39122292376005e-06, "loss": 1.341, "step": 2648 }, { "epoch": 0.20273993571100565, "grad_norm": 0.5932856623521126, "learning_rate": 9.390697016090417e-06, "loss": 1.2941, "step": 2649 }, { "epoch": 0.20281647022807286, "grad_norm": 0.7594032129085511, "learning_rate": 9.390170900086222e-06, "loss": 1.4872, "step": 2650 }, { "epoch": 0.20289300474514005, "grad_norm": 0.6091354592475839, "learning_rate": 9.389644575779792e-06, "loss": 1.6188, "step": 2651 }, { "epoch": 0.20296953926220726, "grad_norm": 0.6811443818358515, "learning_rate": 9.389118043203467e-06, "loss": 1.764, "step": 2652 }, { "epoch": 0.20304607377927444, "grad_norm": 0.6860100145734757, "learning_rate": 9.388591302389597e-06, "loss": 1.456, "step": 2653 }, { "epoch": 0.20312260829634166, "grad_norm": 0.6946697747597638, "learning_rate": 9.388064353370547e-06, "loss": 1.4101, "step": 2654 }, { "epoch": 0.20319914281340884, "grad_norm": 0.5625778003614326, "learning_rate": 9.387537196178696e-06, "loss": 1.3596, "step": 2655 }, { "epoch": 0.20327567733047605, "grad_norm": 0.6047251190747334, "learning_rate": 9.387009830846432e-06, "loss": 1.6144, "step": 2656 }, { "epoch": 0.20335221184754323, "grad_norm": 0.6490608556737826, "learning_rate": 9.38648225740616e-06, "loss": 1.6549, "step": 2657 }, { "epoch": 0.20342874636461045, "grad_norm": 0.6677053648545713, "learning_rate": 9.385954475890293e-06, "loss": 1.5518, "step": 2658 }, { "epoch": 0.20350528088167763, "grad_norm": 0.5925552028076989, "learning_rate": 9.385426486331262e-06, "loss": 1.3722, "step": 2659 }, { "epoch": 0.20358181539874484, "grad_norm": 0.7795071500145592, "learning_rate": 9.384898288761506e-06, "loss": 1.4662, "step": 2660 }, { "epoch": 0.20365834991581203, "grad_norm": 0.5794855946308987, "learning_rate": 9.384369883213482e-06, "loss": 1.2939, "step": 2661 }, { "epoch": 0.20373488443287924, "grad_norm": 0.48585702729185876, "learning_rate": 9.383841269719653e-06, "loss": 1.0022, "step": 2662 }, { "epoch": 0.20381141894994642, "grad_norm": 0.9483934127385584, "learning_rate": 9.3833124483125e-06, "loss": 1.2821, "step": 2663 }, { "epoch": 0.20388795346701363, "grad_norm": 0.5759849132779227, "learning_rate": 9.382783419024519e-06, "loss": 1.4595, "step": 2664 }, { "epoch": 0.20396448798408082, "grad_norm": 0.5544969191485193, "learning_rate": 9.38225418188821e-06, "loss": 1.4206, "step": 2665 }, { "epoch": 0.20404102250114803, "grad_norm": 0.547167428097613, "learning_rate": 9.381724736936094e-06, "loss": 1.2389, "step": 2666 }, { "epoch": 0.2041175570182152, "grad_norm": 0.5376617607069567, "learning_rate": 9.3811950842007e-06, "loss": 1.2007, "step": 2667 }, { "epoch": 0.20419409153528242, "grad_norm": 0.5760717128995627, "learning_rate": 9.380665223714573e-06, "loss": 1.343, "step": 2668 }, { "epoch": 0.2042706260523496, "grad_norm": 0.6158665877304118, "learning_rate": 9.380135155510268e-06, "loss": 1.4257, "step": 2669 }, { "epoch": 0.20434716056941682, "grad_norm": 0.7103448273757544, "learning_rate": 9.379604879620355e-06, "loss": 1.4068, "step": 2670 }, { "epoch": 0.204423695086484, "grad_norm": 0.7381324505553071, "learning_rate": 9.379074396077414e-06, "loss": 1.3446, "step": 2671 }, { "epoch": 0.20450022960355121, "grad_norm": 0.6887493343184486, "learning_rate": 9.378543704914042e-06, "loss": 1.5109, "step": 2672 }, { "epoch": 0.2045767641206184, "grad_norm": 0.5769607677195707, "learning_rate": 9.378012806162842e-06, "loss": 1.5135, "step": 2673 }, { "epoch": 0.2046532986376856, "grad_norm": 1.1994277975311098, "learning_rate": 9.37748169985644e-06, "loss": 1.6739, "step": 2674 }, { "epoch": 0.2047298331547528, "grad_norm": 0.6045950926306641, "learning_rate": 9.376950386027463e-06, "loss": 1.4538, "step": 2675 }, { "epoch": 0.20480636767181998, "grad_norm": 0.6246961155603147, "learning_rate": 9.376418864708561e-06, "loss": 1.4316, "step": 2676 }, { "epoch": 0.2048829021888872, "grad_norm": 0.5956152635758671, "learning_rate": 9.37588713593239e-06, "loss": 1.5097, "step": 2677 }, { "epoch": 0.20495943670595437, "grad_norm": 0.5842372888327255, "learning_rate": 9.375355199731621e-06, "loss": 1.3958, "step": 2678 }, { "epoch": 0.20503597122302158, "grad_norm": 0.7034244072968353, "learning_rate": 9.37482305613894e-06, "loss": 1.4923, "step": 2679 }, { "epoch": 0.20511250574008877, "grad_norm": 0.5506407435831533, "learning_rate": 9.37429070518704e-06, "loss": 1.387, "step": 2680 }, { "epoch": 0.20518904025715598, "grad_norm": 0.5447984606019161, "learning_rate": 9.373758146908632e-06, "loss": 1.3187, "step": 2681 }, { "epoch": 0.20526557477422316, "grad_norm": 0.5560387302447816, "learning_rate": 9.373225381336437e-06, "loss": 1.5231, "step": 2682 }, { "epoch": 0.20534210929129038, "grad_norm": 0.5439487850938355, "learning_rate": 9.372692408503193e-06, "loss": 1.315, "step": 2683 }, { "epoch": 0.20541864380835756, "grad_norm": 0.6865721281890346, "learning_rate": 9.372159228441641e-06, "loss": 1.6538, "step": 2684 }, { "epoch": 0.20549517832542477, "grad_norm": 0.5329594284414373, "learning_rate": 9.37162584118455e-06, "loss": 1.2101, "step": 2685 }, { "epoch": 0.20557171284249195, "grad_norm": 0.5693540675632066, "learning_rate": 9.371092246764684e-06, "loss": 1.3539, "step": 2686 }, { "epoch": 0.20564824735955917, "grad_norm": 0.9246290702805386, "learning_rate": 9.370558445214833e-06, "loss": 1.7253, "step": 2687 }, { "epoch": 0.20572478187662635, "grad_norm": 0.6607754876445555, "learning_rate": 9.370024436567797e-06, "loss": 1.6033, "step": 2688 }, { "epoch": 0.20580131639369356, "grad_norm": 0.5196718835210944, "learning_rate": 9.369490220856385e-06, "loss": 1.031, "step": 2689 }, { "epoch": 0.20587785091076075, "grad_norm": 0.6037184118771141, "learning_rate": 9.36895579811342e-06, "loss": 1.4042, "step": 2690 }, { "epoch": 0.20595438542782796, "grad_norm": 0.557922805212899, "learning_rate": 9.36842116837174e-06, "loss": 1.0817, "step": 2691 }, { "epoch": 0.20603091994489514, "grad_norm": 0.5166379702706597, "learning_rate": 9.367886331664194e-06, "loss": 1.0759, "step": 2692 }, { "epoch": 0.20610745446196235, "grad_norm": 0.6474922081579766, "learning_rate": 9.367351288023642e-06, "loss": 1.5678, "step": 2693 }, { "epoch": 0.20618398897902954, "grad_norm": 0.5276110662609398, "learning_rate": 9.366816037482963e-06, "loss": 1.1951, "step": 2694 }, { "epoch": 0.20626052349609675, "grad_norm": 0.49507384915538843, "learning_rate": 9.366280580075041e-06, "loss": 1.0587, "step": 2695 }, { "epoch": 0.20633705801316393, "grad_norm": 0.6870157048658061, "learning_rate": 9.365744915832778e-06, "loss": 1.7505, "step": 2696 }, { "epoch": 0.20641359253023114, "grad_norm": 0.6054376652021195, "learning_rate": 9.365209044789084e-06, "loss": 1.4804, "step": 2697 }, { "epoch": 0.20649012704729833, "grad_norm": 0.6446697076073142, "learning_rate": 9.364672966976884e-06, "loss": 1.5027, "step": 2698 }, { "epoch": 0.20656666156436554, "grad_norm": 0.6979778384712484, "learning_rate": 9.364136682429122e-06, "loss": 1.6665, "step": 2699 }, { "epoch": 0.20664319608143272, "grad_norm": 0.5628925200036841, "learning_rate": 9.363600191178744e-06, "loss": 1.2548, "step": 2700 }, { "epoch": 0.20671973059849993, "grad_norm": 0.6158606817194467, "learning_rate": 9.363063493258716e-06, "loss": 1.5901, "step": 2701 }, { "epoch": 0.20679626511556712, "grad_norm": 0.7878565707415445, "learning_rate": 9.362526588702012e-06, "loss": 1.4169, "step": 2702 }, { "epoch": 0.20687279963263433, "grad_norm": 0.599705471120039, "learning_rate": 9.361989477541622e-06, "loss": 1.2427, "step": 2703 }, { "epoch": 0.2069493341497015, "grad_norm": 0.6438485889352076, "learning_rate": 9.36145215981055e-06, "loss": 1.7727, "step": 2704 }, { "epoch": 0.20702586866676873, "grad_norm": 0.5297546895554759, "learning_rate": 9.360914635541807e-06, "loss": 1.1412, "step": 2705 }, { "epoch": 0.2071024031838359, "grad_norm": 0.6480606801527328, "learning_rate": 9.360376904768423e-06, "loss": 1.7038, "step": 2706 }, { "epoch": 0.20717893770090312, "grad_norm": 0.5769242891815185, "learning_rate": 9.359838967523435e-06, "loss": 1.3275, "step": 2707 }, { "epoch": 0.2072554722179703, "grad_norm": 0.7367168274981446, "learning_rate": 9.359300823839898e-06, "loss": 1.6281, "step": 2708 }, { "epoch": 0.2073320067350375, "grad_norm": 0.9045424266647077, "learning_rate": 9.358762473750874e-06, "loss": 1.5672, "step": 2709 }, { "epoch": 0.2074085412521047, "grad_norm": 0.7860338070065883, "learning_rate": 9.358223917289445e-06, "loss": 1.2339, "step": 2710 }, { "epoch": 0.20748507576917188, "grad_norm": 0.5484115172903924, "learning_rate": 9.357685154488699e-06, "loss": 1.3109, "step": 2711 }, { "epoch": 0.2075616102862391, "grad_norm": 0.738952444105095, "learning_rate": 9.357146185381737e-06, "loss": 1.7122, "step": 2712 }, { "epoch": 0.20763814480330628, "grad_norm": 0.6469939115875805, "learning_rate": 9.35660701000168e-06, "loss": 1.4415, "step": 2713 }, { "epoch": 0.2077146793203735, "grad_norm": 0.5894533125739977, "learning_rate": 9.356067628381653e-06, "loss": 1.3105, "step": 2714 }, { "epoch": 0.20779121383744067, "grad_norm": 0.6263351144123938, "learning_rate": 9.3555280405548e-06, "loss": 1.4211, "step": 2715 }, { "epoch": 0.2078677483545079, "grad_norm": 0.689857520219765, "learning_rate": 9.35498824655427e-06, "loss": 1.5486, "step": 2716 }, { "epoch": 0.20794428287157507, "grad_norm": 0.5703877712417181, "learning_rate": 9.354448246413234e-06, "loss": 1.2954, "step": 2717 }, { "epoch": 0.20802081738864228, "grad_norm": 0.8771209666210846, "learning_rate": 9.353908040164871e-06, "loss": 1.4934, "step": 2718 }, { "epoch": 0.20809735190570947, "grad_norm": 0.6819108175647396, "learning_rate": 9.353367627842371e-06, "loss": 1.2645, "step": 2719 }, { "epoch": 0.20817388642277668, "grad_norm": 0.6246480514983861, "learning_rate": 9.35282700947894e-06, "loss": 1.5122, "step": 2720 }, { "epoch": 0.20825042093984386, "grad_norm": 0.5426786066238635, "learning_rate": 9.352286185107792e-06, "loss": 1.1685, "step": 2721 }, { "epoch": 0.20832695545691107, "grad_norm": 0.7667008106046078, "learning_rate": 9.351745154762163e-06, "loss": 1.3421, "step": 2722 }, { "epoch": 0.20840348997397826, "grad_norm": 0.556158653432086, "learning_rate": 9.351203918475288e-06, "loss": 1.1791, "step": 2723 }, { "epoch": 0.20848002449104547, "grad_norm": 0.7798367541403498, "learning_rate": 9.350662476280427e-06, "loss": 1.2111, "step": 2724 }, { "epoch": 0.20855655900811265, "grad_norm": 0.624095255742722, "learning_rate": 9.350120828210849e-06, "loss": 1.4071, "step": 2725 }, { "epoch": 0.20863309352517986, "grad_norm": 0.5839628678263277, "learning_rate": 9.349578974299831e-06, "loss": 1.5247, "step": 2726 }, { "epoch": 0.20870962804224705, "grad_norm": 0.5523642501131576, "learning_rate": 9.349036914580669e-06, "loss": 1.2706, "step": 2727 }, { "epoch": 0.20878616255931426, "grad_norm": 0.7915283669875962, "learning_rate": 9.348494649086664e-06, "loss": 1.6884, "step": 2728 }, { "epoch": 0.20886269707638144, "grad_norm": 0.5963480626253419, "learning_rate": 9.34795217785114e-06, "loss": 1.5454, "step": 2729 }, { "epoch": 0.20893923159344865, "grad_norm": 0.6814630170490333, "learning_rate": 9.347409500907425e-06, "loss": 1.5673, "step": 2730 }, { "epoch": 0.20901576611051584, "grad_norm": 0.5667449479412554, "learning_rate": 9.346866618288863e-06, "loss": 1.2465, "step": 2731 }, { "epoch": 0.20909230062758305, "grad_norm": 0.5606189252974992, "learning_rate": 9.346323530028812e-06, "loss": 1.1651, "step": 2732 }, { "epoch": 0.20916883514465023, "grad_norm": 0.500069437920667, "learning_rate": 9.34578023616064e-06, "loss": 1.0285, "step": 2733 }, { "epoch": 0.20924536966171745, "grad_norm": 0.6311571460583165, "learning_rate": 9.345236736717728e-06, "loss": 1.5904, "step": 2734 }, { "epoch": 0.20932190417878463, "grad_norm": 0.5401183544608981, "learning_rate": 9.344693031733469e-06, "loss": 1.4139, "step": 2735 }, { "epoch": 0.20939843869585184, "grad_norm": 0.5337606615096294, "learning_rate": 9.344149121241272e-06, "loss": 1.109, "step": 2736 }, { "epoch": 0.20947497321291902, "grad_norm": 0.5822737108975495, "learning_rate": 9.343605005274556e-06, "loss": 1.3864, "step": 2737 }, { "epoch": 0.20955150772998624, "grad_norm": 0.7365562669569407, "learning_rate": 9.343060683866754e-06, "loss": 1.5998, "step": 2738 }, { "epoch": 0.20962804224705342, "grad_norm": 1.3440394177390047, "learning_rate": 9.342516157051309e-06, "loss": 1.7573, "step": 2739 }, { "epoch": 0.20970457676412063, "grad_norm": 0.5905106119669935, "learning_rate": 9.341971424861677e-06, "loss": 1.4398, "step": 2740 }, { "epoch": 0.20978111128118782, "grad_norm": 1.1680673861113158, "learning_rate": 9.341426487331334e-06, "loss": 1.6059, "step": 2741 }, { "epoch": 0.209857645798255, "grad_norm": 0.6686688596885831, "learning_rate": 9.340881344493757e-06, "loss": 1.3938, "step": 2742 }, { "epoch": 0.2099341803153222, "grad_norm": 0.5706024029354232, "learning_rate": 9.340335996382442e-06, "loss": 1.2615, "step": 2743 }, { "epoch": 0.2100107148323894, "grad_norm": 0.6195516449641979, "learning_rate": 9.339790443030899e-06, "loss": 1.4627, "step": 2744 }, { "epoch": 0.2100872493494566, "grad_norm": 0.5328258295436562, "learning_rate": 9.339244684472646e-06, "loss": 1.1143, "step": 2745 }, { "epoch": 0.2101637838665238, "grad_norm": 0.5809266220715745, "learning_rate": 9.338698720741218e-06, "loss": 1.2952, "step": 2746 }, { "epoch": 0.210240318383591, "grad_norm": 1.2448596254454818, "learning_rate": 9.338152551870158e-06, "loss": 1.4162, "step": 2747 }, { "epoch": 0.21031685290065819, "grad_norm": 0.6308853295508423, "learning_rate": 9.337606177893028e-06, "loss": 1.5775, "step": 2748 }, { "epoch": 0.2103933874177254, "grad_norm": 0.9564023967658202, "learning_rate": 9.337059598843397e-06, "loss": 1.4439, "step": 2749 }, { "epoch": 0.21046992193479258, "grad_norm": 0.6917962013817698, "learning_rate": 9.336512814754848e-06, "loss": 1.6726, "step": 2750 }, { "epoch": 0.2105464564518598, "grad_norm": 0.5939010272099806, "learning_rate": 9.335965825660977e-06, "loss": 1.3792, "step": 2751 }, { "epoch": 0.21062299096892698, "grad_norm": 0.6646328578685905, "learning_rate": 9.335418631595392e-06, "loss": 1.4099, "step": 2752 }, { "epoch": 0.2106995254859942, "grad_norm": 0.6583178347487464, "learning_rate": 9.334871232591717e-06, "loss": 1.6156, "step": 2753 }, { "epoch": 0.21077606000306137, "grad_norm": 0.5327741238680976, "learning_rate": 9.334323628683582e-06, "loss": 1.276, "step": 2754 }, { "epoch": 0.21085259452012858, "grad_norm": 0.7627746343122614, "learning_rate": 9.333775819904639e-06, "loss": 1.7956, "step": 2755 }, { "epoch": 0.21092912903719577, "grad_norm": 0.574415639383386, "learning_rate": 9.333227806288541e-06, "loss": 1.3665, "step": 2756 }, { "epoch": 0.21100566355426298, "grad_norm": 0.5627217955323014, "learning_rate": 9.332679587868962e-06, "loss": 1.371, "step": 2757 }, { "epoch": 0.21108219807133016, "grad_norm": 0.6457250912653899, "learning_rate": 9.332131164679586e-06, "loss": 1.5231, "step": 2758 }, { "epoch": 0.21115873258839737, "grad_norm": 0.5693123744960792, "learning_rate": 9.33158253675411e-06, "loss": 1.4867, "step": 2759 }, { "epoch": 0.21123526710546456, "grad_norm": 0.6784942057666957, "learning_rate": 9.331033704126246e-06, "loss": 1.3908, "step": 2760 }, { "epoch": 0.21131180162253177, "grad_norm": 0.8335938786996877, "learning_rate": 9.330484666829713e-06, "loss": 1.402, "step": 2761 }, { "epoch": 0.21138833613959895, "grad_norm": 0.5795622819271616, "learning_rate": 9.329935424898243e-06, "loss": 1.3374, "step": 2762 }, { "epoch": 0.21146487065666617, "grad_norm": 0.5805104156375123, "learning_rate": 9.329385978365588e-06, "loss": 1.3085, "step": 2763 }, { "epoch": 0.21154140517373335, "grad_norm": 0.7703035940664277, "learning_rate": 9.328836327265504e-06, "loss": 1.6089, "step": 2764 }, { "epoch": 0.21161793969080056, "grad_norm": 0.6459456155602202, "learning_rate": 9.328286471631765e-06, "loss": 1.4325, "step": 2765 }, { "epoch": 0.21169447420786774, "grad_norm": 0.5470657781582562, "learning_rate": 9.327736411498158e-06, "loss": 1.264, "step": 2766 }, { "epoch": 0.21177100872493496, "grad_norm": 0.6359003370129879, "learning_rate": 9.327186146898476e-06, "loss": 1.4773, "step": 2767 }, { "epoch": 0.21184754324200214, "grad_norm": 0.5351741598050407, "learning_rate": 9.32663567786653e-06, "loss": 1.2727, "step": 2768 }, { "epoch": 0.21192407775906935, "grad_norm": 0.6814194969365414, "learning_rate": 9.326085004436145e-06, "loss": 1.4171, "step": 2769 }, { "epoch": 0.21200061227613654, "grad_norm": 0.559762524196686, "learning_rate": 9.325534126641153e-06, "loss": 1.3307, "step": 2770 }, { "epoch": 0.21207714679320375, "grad_norm": 0.689283941017871, "learning_rate": 9.324983044515401e-06, "loss": 1.3989, "step": 2771 }, { "epoch": 0.21215368131027093, "grad_norm": 1.0048202277172316, "learning_rate": 9.324431758092755e-06, "loss": 1.1573, "step": 2772 }, { "epoch": 0.21223021582733814, "grad_norm": 0.612714325761956, "learning_rate": 9.323880267407081e-06, "loss": 1.327, "step": 2773 }, { "epoch": 0.21230675034440533, "grad_norm": 0.5543675222529076, "learning_rate": 9.323328572492267e-06, "loss": 1.3342, "step": 2774 }, { "epoch": 0.21238328486147254, "grad_norm": 0.5611333826376702, "learning_rate": 9.32277667338221e-06, "loss": 1.2534, "step": 2775 }, { "epoch": 0.21245981937853972, "grad_norm": 0.5884837354508896, "learning_rate": 9.322224570110824e-06, "loss": 1.4306, "step": 2776 }, { "epoch": 0.2125363538956069, "grad_norm": 0.5472570102252493, "learning_rate": 9.321672262712026e-06, "loss": 1.3719, "step": 2777 }, { "epoch": 0.21261288841267412, "grad_norm": 0.6953139597433576, "learning_rate": 9.321119751219755e-06, "loss": 1.655, "step": 2778 }, { "epoch": 0.2126894229297413, "grad_norm": 0.7016519891347187, "learning_rate": 9.32056703566796e-06, "loss": 1.2352, "step": 2779 }, { "epoch": 0.2127659574468085, "grad_norm": 0.5046680337063736, "learning_rate": 9.320014116090596e-06, "loss": 1.1849, "step": 2780 }, { "epoch": 0.2128424919638757, "grad_norm": 0.5421827942300137, "learning_rate": 9.319460992521642e-06, "loss": 1.2444, "step": 2781 }, { "epoch": 0.2129190264809429, "grad_norm": 0.6018069443399656, "learning_rate": 9.318907664995081e-06, "loss": 1.348, "step": 2782 }, { "epoch": 0.2129955609980101, "grad_norm": 0.5618823536405796, "learning_rate": 9.318354133544913e-06, "loss": 1.2271, "step": 2783 }, { "epoch": 0.2130720955150773, "grad_norm": 0.5226345532142079, "learning_rate": 9.317800398205145e-06, "loss": 1.2931, "step": 2784 }, { "epoch": 0.2131486300321445, "grad_norm": 0.5745698009016007, "learning_rate": 9.317246459009802e-06, "loss": 1.3902, "step": 2785 }, { "epoch": 0.2132251645492117, "grad_norm": 0.7056537549453724, "learning_rate": 9.316692315992922e-06, "loss": 1.3931, "step": 2786 }, { "epoch": 0.21330169906627888, "grad_norm": 1.066971480376684, "learning_rate": 9.316137969188547e-06, "loss": 1.7503, "step": 2787 }, { "epoch": 0.2133782335833461, "grad_norm": 0.7508712607460337, "learning_rate": 9.315583418630746e-06, "loss": 1.421, "step": 2788 }, { "epoch": 0.21345476810041328, "grad_norm": 0.5464247105361424, "learning_rate": 9.315028664353587e-06, "loss": 1.1084, "step": 2789 }, { "epoch": 0.2135313026174805, "grad_norm": 0.5774925015893329, "learning_rate": 9.314473706391158e-06, "loss": 1.2888, "step": 2790 }, { "epoch": 0.21360783713454767, "grad_norm": 0.6136970930468671, "learning_rate": 9.313918544777553e-06, "loss": 1.2912, "step": 2791 }, { "epoch": 0.21368437165161489, "grad_norm": 0.6008124066076338, "learning_rate": 9.313363179546887e-06, "loss": 1.3613, "step": 2792 }, { "epoch": 0.21376090616868207, "grad_norm": 0.596403834012247, "learning_rate": 9.312807610733283e-06, "loss": 1.2916, "step": 2793 }, { "epoch": 0.21383744068574928, "grad_norm": 0.6328716322398094, "learning_rate": 9.312251838370877e-06, "loss": 1.51, "step": 2794 }, { "epoch": 0.21391397520281646, "grad_norm": 0.6313131671562399, "learning_rate": 9.311695862493815e-06, "loss": 1.4492, "step": 2795 }, { "epoch": 0.21399050971988368, "grad_norm": 0.7054111208912057, "learning_rate": 9.31113968313626e-06, "loss": 1.5602, "step": 2796 }, { "epoch": 0.21406704423695086, "grad_norm": 0.6050148620826352, "learning_rate": 9.310583300332381e-06, "loss": 1.4758, "step": 2797 }, { "epoch": 0.21414357875401807, "grad_norm": 0.5704891834154963, "learning_rate": 9.31002671411637e-06, "loss": 1.2811, "step": 2798 }, { "epoch": 0.21422011327108526, "grad_norm": 0.6560821566627671, "learning_rate": 9.309469924522421e-06, "loss": 1.5371, "step": 2799 }, { "epoch": 0.21429664778815247, "grad_norm": 0.8538769713126643, "learning_rate": 9.30891293158475e-06, "loss": 1.6453, "step": 2800 }, { "epoch": 0.21437318230521965, "grad_norm": 0.6824027176455026, "learning_rate": 9.308355735337573e-06, "loss": 1.7411, "step": 2801 }, { "epoch": 0.21444971682228686, "grad_norm": 1.6956112251330226, "learning_rate": 9.307798335815132e-06, "loss": 1.565, "step": 2802 }, { "epoch": 0.21452625133935405, "grad_norm": 0.7884198836563591, "learning_rate": 9.307240733051671e-06, "loss": 1.6295, "step": 2803 }, { "epoch": 0.21460278585642126, "grad_norm": 0.6498214094416475, "learning_rate": 9.306682927081455e-06, "loss": 1.4809, "step": 2804 }, { "epoch": 0.21467932037348844, "grad_norm": 0.6742139232153898, "learning_rate": 9.306124917938752e-06, "loss": 1.447, "step": 2805 }, { "epoch": 0.21475585489055565, "grad_norm": 0.6110637089982589, "learning_rate": 9.30556670565785e-06, "loss": 1.5342, "step": 2806 }, { "epoch": 0.21483238940762284, "grad_norm": 0.584778921226734, "learning_rate": 9.305008290273051e-06, "loss": 1.4779, "step": 2807 }, { "epoch": 0.21490892392469005, "grad_norm": 0.8348670109498618, "learning_rate": 9.304449671818662e-06, "loss": 1.637, "step": 2808 }, { "epoch": 0.21498545844175723, "grad_norm": 0.4993708675934409, "learning_rate": 9.303890850329005e-06, "loss": 1.1721, "step": 2809 }, { "epoch": 0.21506199295882442, "grad_norm": 0.5219405651835235, "learning_rate": 9.303331825838419e-06, "loss": 1.1783, "step": 2810 }, { "epoch": 0.21513852747589163, "grad_norm": 0.5863410606366516, "learning_rate": 9.302772598381249e-06, "loss": 1.2831, "step": 2811 }, { "epoch": 0.2152150619929588, "grad_norm": 0.8636667534381454, "learning_rate": 9.302213167991857e-06, "loss": 1.5022, "step": 2812 }, { "epoch": 0.21529159651002602, "grad_norm": 0.5318876264688238, "learning_rate": 9.30165353470462e-06, "loss": 1.292, "step": 2813 }, { "epoch": 0.2153681310270932, "grad_norm": 0.594792910274833, "learning_rate": 9.301093698553915e-06, "loss": 1.3005, "step": 2814 }, { "epoch": 0.21544466554416042, "grad_norm": 0.5637652952964821, "learning_rate": 9.300533659574148e-06, "loss": 1.3376, "step": 2815 }, { "epoch": 0.2155212000612276, "grad_norm": 1.0733477491312096, "learning_rate": 9.299973417799723e-06, "loss": 1.5524, "step": 2816 }, { "epoch": 0.21559773457829481, "grad_norm": 0.5651800532976236, "learning_rate": 9.29941297326507e-06, "loss": 1.3071, "step": 2817 }, { "epoch": 0.215674269095362, "grad_norm": 0.5741593688847305, "learning_rate": 9.29885232600462e-06, "loss": 1.254, "step": 2818 }, { "epoch": 0.2157508036124292, "grad_norm": 0.6107045451677965, "learning_rate": 9.298291476052819e-06, "loss": 1.5271, "step": 2819 }, { "epoch": 0.2158273381294964, "grad_norm": 0.5630043504442792, "learning_rate": 9.297730423444131e-06, "loss": 1.3957, "step": 2820 }, { "epoch": 0.2159038726465636, "grad_norm": 0.5916384664484057, "learning_rate": 9.297169168213027e-06, "loss": 1.4926, "step": 2821 }, { "epoch": 0.2159804071636308, "grad_norm": 0.8693518574525331, "learning_rate": 9.296607710393993e-06, "loss": 1.4052, "step": 2822 }, { "epoch": 0.216056941680698, "grad_norm": 0.7995456573826673, "learning_rate": 9.296046050021525e-06, "loss": 1.5759, "step": 2823 }, { "epoch": 0.21613347619776518, "grad_norm": 0.602010306203561, "learning_rate": 9.295484187130136e-06, "loss": 1.5035, "step": 2824 }, { "epoch": 0.2162100107148324, "grad_norm": 0.5790205469089801, "learning_rate": 9.294922121754347e-06, "loss": 1.3941, "step": 2825 }, { "epoch": 0.21628654523189958, "grad_norm": 0.6914275181340932, "learning_rate": 9.294359853928691e-06, "loss": 1.3461, "step": 2826 }, { "epoch": 0.2163630797489668, "grad_norm": 0.693579281375449, "learning_rate": 9.29379738368772e-06, "loss": 1.6266, "step": 2827 }, { "epoch": 0.21643961426603398, "grad_norm": 0.571003757903321, "learning_rate": 9.293234711065991e-06, "loss": 1.4382, "step": 2828 }, { "epoch": 0.2165161487831012, "grad_norm": 0.5186750736604039, "learning_rate": 9.292671836098074e-06, "loss": 1.0825, "step": 2829 }, { "epoch": 0.21659268330016837, "grad_norm": 0.932582794273477, "learning_rate": 9.292108758818558e-06, "loss": 1.3006, "step": 2830 }, { "epoch": 0.21666921781723558, "grad_norm": 0.6551351291503885, "learning_rate": 9.291545479262038e-06, "loss": 1.6904, "step": 2831 }, { "epoch": 0.21674575233430277, "grad_norm": 0.6242756059596396, "learning_rate": 9.290981997463125e-06, "loss": 1.6992, "step": 2832 }, { "epoch": 0.21682228685136998, "grad_norm": 0.5144876026439665, "learning_rate": 9.290418313456437e-06, "loss": 1.1853, "step": 2833 }, { "epoch": 0.21689882136843716, "grad_norm": 0.5939966461728328, "learning_rate": 9.289854427276614e-06, "loss": 1.4853, "step": 2834 }, { "epoch": 0.21697535588550437, "grad_norm": 0.5964311000524088, "learning_rate": 9.2892903389583e-06, "loss": 1.2384, "step": 2835 }, { "epoch": 0.21705189040257156, "grad_norm": 0.6221864639079425, "learning_rate": 9.288726048536153e-06, "loss": 1.4496, "step": 2836 }, { "epoch": 0.21712842491963877, "grad_norm": 0.5704318728712859, "learning_rate": 9.288161556044846e-06, "loss": 1.3241, "step": 2837 }, { "epoch": 0.21720495943670595, "grad_norm": 0.6052206797232202, "learning_rate": 9.287596861519065e-06, "loss": 1.5128, "step": 2838 }, { "epoch": 0.21728149395377316, "grad_norm": 0.5149819124553041, "learning_rate": 9.287031964993504e-06, "loss": 1.1399, "step": 2839 }, { "epoch": 0.21735802847084035, "grad_norm": 0.5550814341024605, "learning_rate": 9.286466866502872e-06, "loss": 1.3452, "step": 2840 }, { "epoch": 0.21743456298790756, "grad_norm": 0.6365153526759377, "learning_rate": 9.285901566081891e-06, "loss": 1.3006, "step": 2841 }, { "epoch": 0.21751109750497474, "grad_norm": 0.5779842037920513, "learning_rate": 9.285336063765297e-06, "loss": 1.3382, "step": 2842 }, { "epoch": 0.21758763202204195, "grad_norm": 0.6292965389579579, "learning_rate": 9.284770359587832e-06, "loss": 1.5975, "step": 2843 }, { "epoch": 0.21766416653910914, "grad_norm": 0.6475770979485284, "learning_rate": 9.284204453584255e-06, "loss": 1.4831, "step": 2844 }, { "epoch": 0.21774070105617632, "grad_norm": 0.5885786328034638, "learning_rate": 9.283638345789343e-06, "loss": 1.394, "step": 2845 }, { "epoch": 0.21781723557324353, "grad_norm": 0.5517649851734195, "learning_rate": 9.283072036237871e-06, "loss": 1.2509, "step": 2846 }, { "epoch": 0.21789377009031072, "grad_norm": 0.5672016463158185, "learning_rate": 9.28250552496464e-06, "loss": 1.4201, "step": 2847 }, { "epoch": 0.21797030460737793, "grad_norm": 0.4825374544006927, "learning_rate": 9.281938812004456e-06, "loss": 1.0469, "step": 2848 }, { "epoch": 0.2180468391244451, "grad_norm": 0.5916772948034099, "learning_rate": 9.281371897392142e-06, "loss": 1.5247, "step": 2849 }, { "epoch": 0.21812337364151232, "grad_norm": 0.6100834809944339, "learning_rate": 9.280804781162527e-06, "loss": 1.3319, "step": 2850 }, { "epoch": 0.2181999081585795, "grad_norm": 0.45998110664294567, "learning_rate": 9.28023746335046e-06, "loss": 0.8674, "step": 2851 }, { "epoch": 0.21827644267564672, "grad_norm": 0.6027809440665889, "learning_rate": 9.279669943990797e-06, "loss": 1.4233, "step": 2852 }, { "epoch": 0.2183529771927139, "grad_norm": 0.5955760278024089, "learning_rate": 9.279102223118408e-06, "loss": 1.4466, "step": 2853 }, { "epoch": 0.21842951170978112, "grad_norm": 0.7099970560024442, "learning_rate": 9.278534300768176e-06, "loss": 1.3832, "step": 2854 }, { "epoch": 0.2185060462268483, "grad_norm": 0.6189404884950968, "learning_rate": 9.277966176974995e-06, "loss": 1.5797, "step": 2855 }, { "epoch": 0.2185825807439155, "grad_norm": 0.6715485847415436, "learning_rate": 9.277397851773772e-06, "loss": 1.4806, "step": 2856 }, { "epoch": 0.2186591152609827, "grad_norm": 0.6528022742690381, "learning_rate": 9.27682932519943e-06, "loss": 1.5972, "step": 2857 }, { "epoch": 0.2187356497780499, "grad_norm": 0.7328058944131682, "learning_rate": 9.276260597286897e-06, "loss": 1.6443, "step": 2858 }, { "epoch": 0.2188121842951171, "grad_norm": 0.559351897216595, "learning_rate": 9.27569166807112e-06, "loss": 1.2213, "step": 2859 }, { "epoch": 0.2188887188121843, "grad_norm": 0.7211752719550308, "learning_rate": 9.275122537587052e-06, "loss": 1.3724, "step": 2860 }, { "epoch": 0.21896525332925149, "grad_norm": 0.8907000062086001, "learning_rate": 9.274553205869666e-06, "loss": 1.7745, "step": 2861 }, { "epoch": 0.2190417878463187, "grad_norm": 0.5470374559545568, "learning_rate": 9.273983672953941e-06, "loss": 1.3012, "step": 2862 }, { "epoch": 0.21911832236338588, "grad_norm": 0.5255863198844926, "learning_rate": 9.273413938874874e-06, "loss": 0.9731, "step": 2863 }, { "epoch": 0.2191948568804531, "grad_norm": 0.6622033377115337, "learning_rate": 9.272844003667465e-06, "loss": 1.132, "step": 2864 }, { "epoch": 0.21927139139752028, "grad_norm": 0.5409610870004938, "learning_rate": 9.27227386736674e-06, "loss": 1.288, "step": 2865 }, { "epoch": 0.2193479259145875, "grad_norm": 0.5455687378526449, "learning_rate": 9.271703530007727e-06, "loss": 1.2104, "step": 2866 }, { "epoch": 0.21942446043165467, "grad_norm": 0.7194044285978218, "learning_rate": 9.271132991625465e-06, "loss": 1.7167, "step": 2867 }, { "epoch": 0.21950099494872188, "grad_norm": 0.5052744419610284, "learning_rate": 9.270562252255016e-06, "loss": 1.0131, "step": 2868 }, { "epoch": 0.21957752946578907, "grad_norm": 0.5994397180265835, "learning_rate": 9.269991311931445e-06, "loss": 1.6035, "step": 2869 }, { "epoch": 0.21965406398285628, "grad_norm": 0.5440246653382251, "learning_rate": 9.269420170689832e-06, "loss": 1.0656, "step": 2870 }, { "epoch": 0.21973059849992346, "grad_norm": 0.5244115793084162, "learning_rate": 9.268848828565267e-06, "loss": 1.1118, "step": 2871 }, { "epoch": 0.21980713301699067, "grad_norm": 0.5750127034735362, "learning_rate": 9.268277285592862e-06, "loss": 1.49, "step": 2872 }, { "epoch": 0.21988366753405786, "grad_norm": 0.6091052772753003, "learning_rate": 9.26770554180773e-06, "loss": 1.529, "step": 2873 }, { "epoch": 0.21996020205112507, "grad_norm": 0.6183007074551639, "learning_rate": 9.267133597244998e-06, "loss": 1.4026, "step": 2874 }, { "epoch": 0.22003673656819225, "grad_norm": 0.6012103031164208, "learning_rate": 9.266561451939815e-06, "loss": 1.2766, "step": 2875 }, { "epoch": 0.22011327108525947, "grad_norm": 0.5246321094359278, "learning_rate": 9.265989105927327e-06, "loss": 0.9869, "step": 2876 }, { "epoch": 0.22018980560232665, "grad_norm": 4.16168513107314, "learning_rate": 9.265416559242705e-06, "loss": 1.6415, "step": 2877 }, { "epoch": 0.22026634011939383, "grad_norm": 0.5840882286560013, "learning_rate": 9.264843811921132e-06, "loss": 1.3981, "step": 2878 }, { "epoch": 0.22034287463646104, "grad_norm": 0.5392843971926485, "learning_rate": 9.264270863997792e-06, "loss": 1.0503, "step": 2879 }, { "epoch": 0.22041940915352823, "grad_norm": 0.5076767948158262, "learning_rate": 9.263697715507893e-06, "loss": 1.1729, "step": 2880 }, { "epoch": 0.22049594367059544, "grad_norm": 1.6228837745801277, "learning_rate": 9.263124366486648e-06, "loss": 1.654, "step": 2881 }, { "epoch": 0.22057247818766262, "grad_norm": 0.5958524021412172, "learning_rate": 9.26255081696929e-06, "loss": 1.487, "step": 2882 }, { "epoch": 0.22064901270472984, "grad_norm": 0.5738963963789587, "learning_rate": 9.261977066991054e-06, "loss": 1.4288, "step": 2883 }, { "epoch": 0.22072554722179702, "grad_norm": 0.5597517787102739, "learning_rate": 9.261403116587198e-06, "loss": 1.3208, "step": 2884 }, { "epoch": 0.22080208173886423, "grad_norm": 0.5794743472866665, "learning_rate": 9.260828965792983e-06, "loss": 1.374, "step": 2885 }, { "epoch": 0.22087861625593141, "grad_norm": 0.5011587046283182, "learning_rate": 9.260254614643689e-06, "loss": 1.1957, "step": 2886 }, { "epoch": 0.22095515077299863, "grad_norm": 0.5253554789231112, "learning_rate": 9.259680063174605e-06, "loss": 1.1828, "step": 2887 }, { "epoch": 0.2210316852900658, "grad_norm": 0.612647083027659, "learning_rate": 9.259105311421032e-06, "loss": 1.3878, "step": 2888 }, { "epoch": 0.22110821980713302, "grad_norm": 0.636123153949528, "learning_rate": 9.258530359418289e-06, "loss": 1.6236, "step": 2889 }, { "epoch": 0.2211847543242002, "grad_norm": 0.5839959415807672, "learning_rate": 9.257955207201695e-06, "loss": 1.3119, "step": 2890 }, { "epoch": 0.22126128884126742, "grad_norm": 0.6534904053728939, "learning_rate": 9.257379854806599e-06, "loss": 1.6024, "step": 2891 }, { "epoch": 0.2213378233583346, "grad_norm": 0.5570800764285114, "learning_rate": 9.256804302268342e-06, "loss": 1.229, "step": 2892 }, { "epoch": 0.2214143578754018, "grad_norm": 0.6109750823214578, "learning_rate": 9.256228549622294e-06, "loss": 1.5536, "step": 2893 }, { "epoch": 0.221490892392469, "grad_norm": 0.49413244104732124, "learning_rate": 9.255652596903832e-06, "loss": 1.1248, "step": 2894 }, { "epoch": 0.2215674269095362, "grad_norm": 0.6423351284773917, "learning_rate": 9.255076444148341e-06, "loss": 1.5807, "step": 2895 }, { "epoch": 0.2216439614266034, "grad_norm": 0.6912282000880661, "learning_rate": 9.25450009139122e-06, "loss": 1.6435, "step": 2896 }, { "epoch": 0.2217204959436706, "grad_norm": 0.6373307893173104, "learning_rate": 9.253923538667888e-06, "loss": 1.6187, "step": 2897 }, { "epoch": 0.2217970304607378, "grad_norm": 1.3257078590155587, "learning_rate": 9.253346786013764e-06, "loss": 1.7783, "step": 2898 }, { "epoch": 0.221873564977805, "grad_norm": 0.6469042577662137, "learning_rate": 9.252769833464288e-06, "loss": 1.7069, "step": 2899 }, { "epoch": 0.22195009949487218, "grad_norm": 0.8832947663604163, "learning_rate": 9.25219268105491e-06, "loss": 1.6804, "step": 2900 }, { "epoch": 0.2220266340119394, "grad_norm": 0.6334187153974109, "learning_rate": 9.251615328821091e-06, "loss": 1.5324, "step": 2901 }, { "epoch": 0.22210316852900658, "grad_norm": 0.6380847100796085, "learning_rate": 9.251037776798306e-06, "loss": 1.7229, "step": 2902 }, { "epoch": 0.2221797030460738, "grad_norm": 0.6430369945017076, "learning_rate": 9.250460025022042e-06, "loss": 1.4212, "step": 2903 }, { "epoch": 0.22225623756314097, "grad_norm": 0.5493355195314975, "learning_rate": 9.249882073527796e-06, "loss": 1.3365, "step": 2904 }, { "epoch": 0.22233277208020819, "grad_norm": 0.6119748395654839, "learning_rate": 9.249303922351081e-06, "loss": 1.2347, "step": 2905 }, { "epoch": 0.22240930659727537, "grad_norm": 0.647872312245449, "learning_rate": 9.248725571527418e-06, "loss": 1.6024, "step": 2906 }, { "epoch": 0.22248584111434258, "grad_norm": 0.5433644982490519, "learning_rate": 9.248147021092346e-06, "loss": 1.3173, "step": 2907 }, { "epoch": 0.22256237563140976, "grad_norm": 0.6665718477310408, "learning_rate": 9.24756827108141e-06, "loss": 1.5314, "step": 2908 }, { "epoch": 0.22263891014847698, "grad_norm": 0.620413686427423, "learning_rate": 9.246989321530171e-06, "loss": 1.3819, "step": 2909 }, { "epoch": 0.22271544466554416, "grad_norm": 0.6186620410744662, "learning_rate": 9.246410172474201e-06, "loss": 1.568, "step": 2910 }, { "epoch": 0.22279197918261137, "grad_norm": 0.5699157388344374, "learning_rate": 9.245830823949086e-06, "loss": 1.3254, "step": 2911 }, { "epoch": 0.22286851369967856, "grad_norm": 0.5283532267037016, "learning_rate": 9.24525127599042e-06, "loss": 1.1905, "step": 2912 }, { "epoch": 0.22294504821674574, "grad_norm": 0.6320121082166719, "learning_rate": 9.244671528633817e-06, "loss": 1.441, "step": 2913 }, { "epoch": 0.22302158273381295, "grad_norm": 0.4974684533544483, "learning_rate": 9.244091581914893e-06, "loss": 1.0203, "step": 2914 }, { "epoch": 0.22309811725088013, "grad_norm": 0.551567461885857, "learning_rate": 9.243511435869286e-06, "loss": 1.1883, "step": 2915 }, { "epoch": 0.22317465176794735, "grad_norm": 0.6457814099643409, "learning_rate": 9.242931090532639e-06, "loss": 1.487, "step": 2916 }, { "epoch": 0.22325118628501453, "grad_norm": 0.7528323586549964, "learning_rate": 9.242350545940612e-06, "loss": 1.8077, "step": 2917 }, { "epoch": 0.22332772080208174, "grad_norm": 0.5318376136177945, "learning_rate": 9.241769802128873e-06, "loss": 1.3536, "step": 2918 }, { "epoch": 0.22340425531914893, "grad_norm": 0.7307495984234036, "learning_rate": 9.24118885913311e-06, "loss": 1.5871, "step": 2919 }, { "epoch": 0.22348078983621614, "grad_norm": 0.5436561276509078, "learning_rate": 9.24060771698901e-06, "loss": 1.374, "step": 2920 }, { "epoch": 0.22355732435328332, "grad_norm": 0.6188728863152806, "learning_rate": 9.240026375732288e-06, "loss": 1.6442, "step": 2921 }, { "epoch": 0.22363385887035053, "grad_norm": 1.043439276291369, "learning_rate": 9.239444835398657e-06, "loss": 1.395, "step": 2922 }, { "epoch": 0.22371039338741772, "grad_norm": 0.6267780517963348, "learning_rate": 9.238863096023852e-06, "loss": 1.3873, "step": 2923 }, { "epoch": 0.22378692790448493, "grad_norm": 0.550512556294395, "learning_rate": 9.238281157643617e-06, "loss": 1.2332, "step": 2924 }, { "epoch": 0.2238634624215521, "grad_norm": 0.5803177936112421, "learning_rate": 9.237699020293705e-06, "loss": 1.4673, "step": 2925 }, { "epoch": 0.22393999693861932, "grad_norm": 0.7830753426647806, "learning_rate": 9.237116684009888e-06, "loss": 1.9925, "step": 2926 }, { "epoch": 0.2240165314556865, "grad_norm": 0.6396657574269928, "learning_rate": 9.236534148827944e-06, "loss": 1.5233, "step": 2927 }, { "epoch": 0.22409306597275372, "grad_norm": 0.7195670441213655, "learning_rate": 9.235951414783668e-06, "loss": 1.7283, "step": 2928 }, { "epoch": 0.2241696004898209, "grad_norm": 0.5802073748846114, "learning_rate": 9.23536848191286e-06, "loss": 1.4396, "step": 2929 }, { "epoch": 0.22424613500688811, "grad_norm": 1.179114069885837, "learning_rate": 9.234785350251345e-06, "loss": 1.4325, "step": 2930 }, { "epoch": 0.2243226695239553, "grad_norm": 0.6153785291223238, "learning_rate": 9.234202019834944e-06, "loss": 1.4963, "step": 2931 }, { "epoch": 0.2243992040410225, "grad_norm": 0.5780035301002545, "learning_rate": 9.233618490699505e-06, "loss": 1.3899, "step": 2932 }, { "epoch": 0.2244757385580897, "grad_norm": 0.5931899369070678, "learning_rate": 9.233034762880877e-06, "loss": 1.3948, "step": 2933 }, { "epoch": 0.2245522730751569, "grad_norm": 0.6699912982788547, "learning_rate": 9.232450836414932e-06, "loss": 1.5449, "step": 2934 }, { "epoch": 0.2246288075922241, "grad_norm": 0.8213634863468958, "learning_rate": 9.23186671133754e-06, "loss": 1.5802, "step": 2935 }, { "epoch": 0.2247053421092913, "grad_norm": 0.5693760528549173, "learning_rate": 9.231282387684598e-06, "loss": 1.3692, "step": 2936 }, { "epoch": 0.22478187662635848, "grad_norm": 0.5632112829683696, "learning_rate": 9.230697865492007e-06, "loss": 1.3083, "step": 2937 }, { "epoch": 0.2248584111434257, "grad_norm": 0.5689101674200683, "learning_rate": 9.23011314479568e-06, "loss": 1.2151, "step": 2938 }, { "epoch": 0.22493494566049288, "grad_norm": 0.5150367370917385, "learning_rate": 9.229528225631546e-06, "loss": 1.0682, "step": 2939 }, { "epoch": 0.2250114801775601, "grad_norm": 0.6372483420109796, "learning_rate": 9.228943108035542e-06, "loss": 1.6213, "step": 2940 }, { "epoch": 0.22508801469462728, "grad_norm": 0.5472581499874474, "learning_rate": 9.228357792043623e-06, "loss": 1.2123, "step": 2941 }, { "epoch": 0.2251645492116945, "grad_norm": 0.5302934354972165, "learning_rate": 9.227772277691749e-06, "loss": 1.1836, "step": 2942 }, { "epoch": 0.22524108372876167, "grad_norm": 0.6168221705936863, "learning_rate": 9.227186565015896e-06, "loss": 1.6149, "step": 2943 }, { "epoch": 0.22531761824582888, "grad_norm": 0.5609870391460354, "learning_rate": 9.226600654052056e-06, "loss": 1.3709, "step": 2944 }, { "epoch": 0.22539415276289607, "grad_norm": 0.5279049443119868, "learning_rate": 9.226014544836223e-06, "loss": 1.2338, "step": 2945 }, { "epoch": 0.22547068727996325, "grad_norm": 0.5779706732183274, "learning_rate": 9.225428237404414e-06, "loss": 1.4007, "step": 2946 }, { "epoch": 0.22554722179703046, "grad_norm": 0.709807193389685, "learning_rate": 9.224841731792652e-06, "loss": 1.4764, "step": 2947 }, { "epoch": 0.22562375631409765, "grad_norm": 1.296819202347162, "learning_rate": 9.224255028036973e-06, "loss": 1.7847, "step": 2948 }, { "epoch": 0.22570029083116486, "grad_norm": 0.6290878181579858, "learning_rate": 9.223668126173427e-06, "loss": 1.5657, "step": 2949 }, { "epoch": 0.22577682534823204, "grad_norm": 0.6248650281884182, "learning_rate": 9.223081026238072e-06, "loss": 1.3352, "step": 2950 }, { "epoch": 0.22585335986529925, "grad_norm": 0.5995786641033957, "learning_rate": 9.222493728266986e-06, "loss": 1.3744, "step": 2951 }, { "epoch": 0.22592989438236644, "grad_norm": 0.5992787393595228, "learning_rate": 9.221906232296252e-06, "loss": 1.2978, "step": 2952 }, { "epoch": 0.22600642889943365, "grad_norm": 0.6696899977075135, "learning_rate": 9.221318538361965e-06, "loss": 1.7232, "step": 2953 }, { "epoch": 0.22608296341650083, "grad_norm": 0.5165395086077925, "learning_rate": 9.22073064650024e-06, "loss": 1.1948, "step": 2954 }, { "epoch": 0.22615949793356804, "grad_norm": 0.7099991975661846, "learning_rate": 9.220142556747196e-06, "loss": 1.7991, "step": 2955 }, { "epoch": 0.22623603245063523, "grad_norm": 0.6235839673658639, "learning_rate": 9.219554269138966e-06, "loss": 1.4608, "step": 2956 }, { "epoch": 0.22631256696770244, "grad_norm": 0.678201467704131, "learning_rate": 9.218965783711695e-06, "loss": 1.799, "step": 2957 }, { "epoch": 0.22638910148476962, "grad_norm": 0.7163352397739156, "learning_rate": 9.218377100501545e-06, "loss": 1.6849, "step": 2958 }, { "epoch": 0.22646563600183683, "grad_norm": 0.6469273394443323, "learning_rate": 9.217788219544686e-06, "loss": 1.5903, "step": 2959 }, { "epoch": 0.22654217051890402, "grad_norm": 0.5582205936351797, "learning_rate": 9.217199140877298e-06, "loss": 1.2635, "step": 2960 }, { "epoch": 0.22661870503597123, "grad_norm": 0.4931820799005221, "learning_rate": 9.216609864535577e-06, "loss": 1.138, "step": 2961 }, { "epoch": 0.2266952395530384, "grad_norm": 0.568448636512861, "learning_rate": 9.21602039055573e-06, "loss": 1.3016, "step": 2962 }, { "epoch": 0.22677177407010563, "grad_norm": 1.006321170638451, "learning_rate": 9.215430718973978e-06, "loss": 1.4933, "step": 2963 }, { "epoch": 0.2268483085871728, "grad_norm": 0.6329996231794616, "learning_rate": 9.21484084982655e-06, "loss": 1.531, "step": 2964 }, { "epoch": 0.22692484310424002, "grad_norm": 0.5956038898103504, "learning_rate": 9.214250783149688e-06, "loss": 1.5289, "step": 2965 }, { "epoch": 0.2270013776213072, "grad_norm": 0.7397427058707947, "learning_rate": 9.21366051897965e-06, "loss": 1.2681, "step": 2966 }, { "epoch": 0.22707791213837442, "grad_norm": 0.7576692672598782, "learning_rate": 9.213070057352704e-06, "loss": 1.9186, "step": 2967 }, { "epoch": 0.2271544466554416, "grad_norm": 0.5142394942750421, "learning_rate": 9.212479398305125e-06, "loss": 1.0552, "step": 2968 }, { "epoch": 0.2272309811725088, "grad_norm": 0.6843188664205342, "learning_rate": 9.21188854187321e-06, "loss": 1.5313, "step": 2969 }, { "epoch": 0.227307515689576, "grad_norm": 0.6154619044057958, "learning_rate": 9.211297488093262e-06, "loss": 1.3597, "step": 2970 }, { "epoch": 0.2273840502066432, "grad_norm": 0.5956588456749338, "learning_rate": 9.210706237001597e-06, "loss": 1.5703, "step": 2971 }, { "epoch": 0.2274605847237104, "grad_norm": 0.7470578531259487, "learning_rate": 9.21011478863454e-06, "loss": 1.919, "step": 2972 }, { "epoch": 0.2275371192407776, "grad_norm": 0.489978275926886, "learning_rate": 9.209523143028437e-06, "loss": 0.9809, "step": 2973 }, { "epoch": 0.22761365375784479, "grad_norm": 0.7748706758707827, "learning_rate": 9.208931300219636e-06, "loss": 1.5659, "step": 2974 }, { "epoch": 0.227690188274912, "grad_norm": 0.7198978865408664, "learning_rate": 9.208339260244503e-06, "loss": 1.7613, "step": 2975 }, { "epoch": 0.22776672279197918, "grad_norm": 0.5368157633010269, "learning_rate": 9.207747023139414e-06, "loss": 1.3252, "step": 2976 }, { "epoch": 0.2278432573090464, "grad_norm": 0.5680314714222899, "learning_rate": 9.20715458894076e-06, "loss": 1.2665, "step": 2977 }, { "epoch": 0.22791979182611358, "grad_norm": 0.6668086144763179, "learning_rate": 9.206561957684938e-06, "loss": 1.479, "step": 2978 }, { "epoch": 0.22799632634318076, "grad_norm": 0.7007010002526963, "learning_rate": 9.205969129408367e-06, "loss": 1.6699, "step": 2979 }, { "epoch": 0.22807286086024797, "grad_norm": 0.5434907522106786, "learning_rate": 9.205376104147467e-06, "loss": 1.3645, "step": 2980 }, { "epoch": 0.22814939537731516, "grad_norm": 0.5793648700052092, "learning_rate": 9.204782881938678e-06, "loss": 1.371, "step": 2981 }, { "epoch": 0.22822592989438237, "grad_norm": 0.9286286693908924, "learning_rate": 9.204189462818444e-06, "loss": 1.5484, "step": 2982 }, { "epoch": 0.22830246441144955, "grad_norm": 0.6267352436803406, "learning_rate": 9.203595846823236e-06, "loss": 1.4277, "step": 2983 }, { "epoch": 0.22837899892851676, "grad_norm": 0.7235144196035652, "learning_rate": 9.203002033989518e-06, "loss": 1.7147, "step": 2984 }, { "epoch": 0.22845553344558395, "grad_norm": 0.8446554430904315, "learning_rate": 9.202408024353783e-06, "loss": 1.6148, "step": 2985 }, { "epoch": 0.22853206796265116, "grad_norm": 0.7904492079997368, "learning_rate": 9.201813817952525e-06, "loss": 1.5191, "step": 2986 }, { "epoch": 0.22860860247971834, "grad_norm": 0.7773412656237939, "learning_rate": 9.201219414822253e-06, "loss": 1.4628, "step": 2987 }, { "epoch": 0.22868513699678555, "grad_norm": 0.5876474037267786, "learning_rate": 9.200624814999493e-06, "loss": 1.3866, "step": 2988 }, { "epoch": 0.22876167151385274, "grad_norm": 0.7958378021089709, "learning_rate": 9.200030018520774e-06, "loss": 1.5535, "step": 2989 }, { "epoch": 0.22883820603091995, "grad_norm": 0.5516564867128011, "learning_rate": 9.199435025422645e-06, "loss": 1.2607, "step": 2990 }, { "epoch": 0.22891474054798713, "grad_norm": 0.640977065581737, "learning_rate": 9.198839835741663e-06, "loss": 1.4879, "step": 2991 }, { "epoch": 0.22899127506505434, "grad_norm": 0.5903262601314082, "learning_rate": 9.198244449514401e-06, "loss": 1.3556, "step": 2992 }, { "epoch": 0.22906780958212153, "grad_norm": 0.9146901813123693, "learning_rate": 9.197648866777437e-06, "loss": 1.491, "step": 2993 }, { "epoch": 0.22914434409918874, "grad_norm": 0.6492776819285517, "learning_rate": 9.197053087567367e-06, "loss": 1.7098, "step": 2994 }, { "epoch": 0.22922087861625592, "grad_norm": 0.5399394342199009, "learning_rate": 9.1964571119208e-06, "loss": 1.1569, "step": 2995 }, { "epoch": 0.22929741313332314, "grad_norm": 0.7333146265795789, "learning_rate": 9.195860939874352e-06, "loss": 1.1083, "step": 2996 }, { "epoch": 0.22937394765039032, "grad_norm": 0.5839542830839729, "learning_rate": 9.195264571464652e-06, "loss": 1.4591, "step": 2997 }, { "epoch": 0.22945048216745753, "grad_norm": 0.5858576192638062, "learning_rate": 9.194668006728346e-06, "loss": 1.3982, "step": 2998 }, { "epoch": 0.22952701668452472, "grad_norm": 0.7209835431092433, "learning_rate": 9.194071245702087e-06, "loss": 1.5108, "step": 2999 }, { "epoch": 0.22960355120159193, "grad_norm": 0.5823574853030947, "learning_rate": 9.193474288422543e-06, "loss": 1.4372, "step": 3000 }, { "epoch": 0.2296800857186591, "grad_norm": 0.5520788443587171, "learning_rate": 9.192877134926392e-06, "loss": 1.0651, "step": 3001 }, { "epoch": 0.22975662023572632, "grad_norm": 0.5556137650642591, "learning_rate": 9.192279785250326e-06, "loss": 1.233, "step": 3002 }, { "epoch": 0.2298331547527935, "grad_norm": 0.627580350999761, "learning_rate": 9.191682239431045e-06, "loss": 1.6218, "step": 3003 }, { "epoch": 0.22990968926986072, "grad_norm": 0.669735611166172, "learning_rate": 9.191084497505266e-06, "loss": 1.4189, "step": 3004 }, { "epoch": 0.2299862237869279, "grad_norm": 0.5820586575496974, "learning_rate": 9.190486559509717e-06, "loss": 1.3462, "step": 3005 }, { "epoch": 0.2300627583039951, "grad_norm": 0.5147951523728151, "learning_rate": 9.189888425481135e-06, "loss": 1.133, "step": 3006 }, { "epoch": 0.2301392928210623, "grad_norm": 0.6346476565423921, "learning_rate": 9.189290095456273e-06, "loss": 1.5794, "step": 3007 }, { "epoch": 0.2302158273381295, "grad_norm": 0.9100231343724255, "learning_rate": 9.188691569471893e-06, "loss": 1.4892, "step": 3008 }, { "epoch": 0.2302923618551967, "grad_norm": 0.8647595930363854, "learning_rate": 9.18809284756477e-06, "loss": 1.8303, "step": 3009 }, { "epoch": 0.2303688963722639, "grad_norm": 0.5625963044794842, "learning_rate": 9.187493929771694e-06, "loss": 1.4649, "step": 3010 }, { "epoch": 0.2304454308893311, "grad_norm": 0.5705837188937557, "learning_rate": 9.18689481612946e-06, "loss": 1.5925, "step": 3011 }, { "epoch": 0.2305219654063983, "grad_norm": 0.6098676991529609, "learning_rate": 9.186295506674882e-06, "loss": 1.5408, "step": 3012 }, { "epoch": 0.23059849992346548, "grad_norm": 0.617127821758131, "learning_rate": 9.185696001444782e-06, "loss": 1.5664, "step": 3013 }, { "epoch": 0.23067503444053267, "grad_norm": 0.5972376581374066, "learning_rate": 9.185096300475999e-06, "loss": 1.6265, "step": 3014 }, { "epoch": 0.23075156895759988, "grad_norm": 0.8674184306502188, "learning_rate": 9.184496403805376e-06, "loss": 1.7261, "step": 3015 }, { "epoch": 0.23082810347466706, "grad_norm": 0.6425878595430786, "learning_rate": 9.183896311469774e-06, "loss": 1.5223, "step": 3016 }, { "epoch": 0.23090463799173427, "grad_norm": 0.5077942100121299, "learning_rate": 9.183296023506064e-06, "loss": 0.9568, "step": 3017 }, { "epoch": 0.23098117250880146, "grad_norm": 0.9894307870341223, "learning_rate": 9.182695539951131e-06, "loss": 1.601, "step": 3018 }, { "epoch": 0.23105770702586867, "grad_norm": 0.5179897209070152, "learning_rate": 9.18209486084187e-06, "loss": 1.0687, "step": 3019 }, { "epoch": 0.23113424154293585, "grad_norm": 0.9671217808134823, "learning_rate": 9.181493986215186e-06, "loss": 1.6332, "step": 3020 }, { "epoch": 0.23121077606000306, "grad_norm": 0.7455282782709958, "learning_rate": 9.180892916108003e-06, "loss": 1.8617, "step": 3021 }, { "epoch": 0.23128731057707025, "grad_norm": 0.5615349362046993, "learning_rate": 9.18029165055725e-06, "loss": 1.3386, "step": 3022 }, { "epoch": 0.23136384509413746, "grad_norm": 1.3533147604393214, "learning_rate": 9.179690189599869e-06, "loss": 1.5846, "step": 3023 }, { "epoch": 0.23144037961120464, "grad_norm": 0.5861701059298099, "learning_rate": 9.179088533272818e-06, "loss": 1.3475, "step": 3024 }, { "epoch": 0.23151691412827186, "grad_norm": 0.7053634505095872, "learning_rate": 9.178486681613064e-06, "loss": 1.5568, "step": 3025 }, { "epoch": 0.23159344864533904, "grad_norm": 0.5986923836324153, "learning_rate": 9.177884634657586e-06, "loss": 1.475, "step": 3026 }, { "epoch": 0.23166998316240625, "grad_norm": 0.8828469046717905, "learning_rate": 9.177282392443377e-06, "loss": 1.7498, "step": 3027 }, { "epoch": 0.23174651767947343, "grad_norm": 0.7096372905428248, "learning_rate": 9.17667995500744e-06, "loss": 1.7032, "step": 3028 }, { "epoch": 0.23182305219654065, "grad_norm": 0.5485130353480059, "learning_rate": 9.176077322386788e-06, "loss": 1.3217, "step": 3029 }, { "epoch": 0.23189958671360783, "grad_norm": 0.48934155431973936, "learning_rate": 9.175474494618453e-06, "loss": 1.1041, "step": 3030 }, { "epoch": 0.23197612123067504, "grad_norm": 2.7808106226851232, "learning_rate": 9.17487147173947e-06, "loss": 1.6806, "step": 3031 }, { "epoch": 0.23205265574774223, "grad_norm": 1.835803276454149, "learning_rate": 9.174268253786894e-06, "loss": 1.9465, "step": 3032 }, { "epoch": 0.23212919026480944, "grad_norm": 0.6171732261761191, "learning_rate": 9.173664840797787e-06, "loss": 1.6142, "step": 3033 }, { "epoch": 0.23220572478187662, "grad_norm": 0.5824218553945947, "learning_rate": 9.173061232809223e-06, "loss": 1.5298, "step": 3034 }, { "epoch": 0.23228225929894383, "grad_norm": 0.4930961002573247, "learning_rate": 9.172457429858295e-06, "loss": 1.1061, "step": 3035 }, { "epoch": 0.23235879381601102, "grad_norm": 0.7455040568450183, "learning_rate": 9.171853431982096e-06, "loss": 1.6821, "step": 3036 }, { "epoch": 0.23243532833307823, "grad_norm": 0.6630880348352999, "learning_rate": 9.17124923921774e-06, "loss": 1.6551, "step": 3037 }, { "epoch": 0.2325118628501454, "grad_norm": 0.612990957468721, "learning_rate": 9.170644851602351e-06, "loss": 1.4617, "step": 3038 }, { "epoch": 0.23258839736721262, "grad_norm": 0.6279392656743779, "learning_rate": 9.170040269173064e-06, "loss": 1.3911, "step": 3039 }, { "epoch": 0.2326649318842798, "grad_norm": 0.6028397521466878, "learning_rate": 9.169435491967027e-06, "loss": 1.5403, "step": 3040 }, { "epoch": 0.23274146640134702, "grad_norm": 0.5843454645842991, "learning_rate": 9.168830520021397e-06, "loss": 1.4719, "step": 3041 }, { "epoch": 0.2328180009184142, "grad_norm": 0.5922003843721951, "learning_rate": 9.16822535337335e-06, "loss": 1.497, "step": 3042 }, { "epoch": 0.23289453543548141, "grad_norm": 0.5190656840388751, "learning_rate": 9.167619992060062e-06, "loss": 1.0307, "step": 3043 }, { "epoch": 0.2329710699525486, "grad_norm": 0.5743110912157243, "learning_rate": 9.167014436118735e-06, "loss": 1.3122, "step": 3044 }, { "epoch": 0.2330476044696158, "grad_norm": 0.9917770410172185, "learning_rate": 9.166408685586571e-06, "loss": 1.5024, "step": 3045 }, { "epoch": 0.233124138986683, "grad_norm": 0.7454613132040255, "learning_rate": 9.165802740500794e-06, "loss": 1.4946, "step": 3046 }, { "epoch": 0.23320067350375018, "grad_norm": 0.6715498070874352, "learning_rate": 9.165196600898631e-06, "loss": 1.4766, "step": 3047 }, { "epoch": 0.2332772080208174, "grad_norm": 0.608069876127437, "learning_rate": 9.164590266817328e-06, "loss": 1.6398, "step": 3048 }, { "epoch": 0.23335374253788457, "grad_norm": 0.49563986400697463, "learning_rate": 9.163983738294138e-06, "loss": 0.9083, "step": 3049 }, { "epoch": 0.23343027705495178, "grad_norm": 0.5281750246235306, "learning_rate": 9.163377015366331e-06, "loss": 1.0242, "step": 3050 }, { "epoch": 0.23350681157201897, "grad_norm": 1.1030484750979714, "learning_rate": 9.162770098071182e-06, "loss": 1.4922, "step": 3051 }, { "epoch": 0.23358334608908618, "grad_norm": 0.5910433614037904, "learning_rate": 9.162162986445981e-06, "loss": 1.2192, "step": 3052 }, { "epoch": 0.23365988060615336, "grad_norm": 0.613343146269707, "learning_rate": 9.161555680528034e-06, "loss": 1.4432, "step": 3053 }, { "epoch": 0.23373641512322058, "grad_norm": 0.6463279845619054, "learning_rate": 9.160948180354657e-06, "loss": 1.5396, "step": 3054 }, { "epoch": 0.23381294964028776, "grad_norm": 0.5423450042826602, "learning_rate": 9.16034048596317e-06, "loss": 1.3404, "step": 3055 }, { "epoch": 0.23388948415735497, "grad_norm": 0.5894893860598244, "learning_rate": 9.15973259739092e-06, "loss": 1.482, "step": 3056 }, { "epoch": 0.23396601867442215, "grad_norm": 0.6804000092352378, "learning_rate": 9.159124514675251e-06, "loss": 1.6906, "step": 3057 }, { "epoch": 0.23404255319148937, "grad_norm": 1.2557458022941907, "learning_rate": 9.15851623785353e-06, "loss": 1.8729, "step": 3058 }, { "epoch": 0.23411908770855655, "grad_norm": 0.7155650744106752, "learning_rate": 9.157907766963128e-06, "loss": 1.8599, "step": 3059 }, { "epoch": 0.23419562222562376, "grad_norm": 0.8153574578028134, "learning_rate": 9.157299102041433e-06, "loss": 1.7828, "step": 3060 }, { "epoch": 0.23427215674269095, "grad_norm": 0.4616611731005218, "learning_rate": 9.156690243125841e-06, "loss": 1.038, "step": 3061 }, { "epoch": 0.23434869125975816, "grad_norm": 0.6594094675349769, "learning_rate": 9.156081190253766e-06, "loss": 1.7044, "step": 3062 }, { "epoch": 0.23442522577682534, "grad_norm": 0.5762960873467764, "learning_rate": 9.155471943462627e-06, "loss": 1.4698, "step": 3063 }, { "epoch": 0.23450176029389255, "grad_norm": 0.6390799700022143, "learning_rate": 9.154862502789859e-06, "loss": 1.5681, "step": 3064 }, { "epoch": 0.23457829481095974, "grad_norm": 0.5775960330010619, "learning_rate": 9.154252868272907e-06, "loss": 1.3516, "step": 3065 }, { "epoch": 0.23465482932802695, "grad_norm": 0.5480615642556458, "learning_rate": 9.153643039949229e-06, "loss": 1.4086, "step": 3066 }, { "epoch": 0.23473136384509413, "grad_norm": 0.6410852424967012, "learning_rate": 9.153033017856294e-06, "loss": 1.3536, "step": 3067 }, { "epoch": 0.23480789836216134, "grad_norm": 0.7297831049646443, "learning_rate": 9.152422802031587e-06, "loss": 1.2589, "step": 3068 }, { "epoch": 0.23488443287922853, "grad_norm": 0.6644182369426759, "learning_rate": 9.151812392512597e-06, "loss": 1.5834, "step": 3069 }, { "epoch": 0.23496096739629574, "grad_norm": 0.5953950040007567, "learning_rate": 9.151201789336833e-06, "loss": 1.3881, "step": 3070 }, { "epoch": 0.23503750191336292, "grad_norm": 0.5356233292218051, "learning_rate": 9.15059099254181e-06, "loss": 1.195, "step": 3071 }, { "epoch": 0.23511403643043013, "grad_norm": 0.570358090244641, "learning_rate": 9.14998000216506e-06, "loss": 1.2606, "step": 3072 }, { "epoch": 0.23519057094749732, "grad_norm": 0.5844490084877373, "learning_rate": 9.14936881824412e-06, "loss": 1.4615, "step": 3073 }, { "epoch": 0.23526710546456453, "grad_norm": 0.6310683573377706, "learning_rate": 9.148757440816544e-06, "loss": 1.4165, "step": 3074 }, { "epoch": 0.2353436399816317, "grad_norm": 0.5744763049000733, "learning_rate": 9.1481458699199e-06, "loss": 1.4722, "step": 3075 }, { "epoch": 0.23542017449869893, "grad_norm": 0.5854287473513179, "learning_rate": 9.147534105591762e-06, "loss": 1.3296, "step": 3076 }, { "epoch": 0.2354967090157661, "grad_norm": 0.6208084976427715, "learning_rate": 9.146922147869719e-06, "loss": 1.3402, "step": 3077 }, { "epoch": 0.23557324353283332, "grad_norm": 0.5826265142211766, "learning_rate": 9.146309996791371e-06, "loss": 1.2305, "step": 3078 }, { "epoch": 0.2356497780499005, "grad_norm": 0.7718125818738732, "learning_rate": 9.145697652394332e-06, "loss": 1.4693, "step": 3079 }, { "epoch": 0.23572631256696772, "grad_norm": 0.6004668204480811, "learning_rate": 9.145085114716224e-06, "loss": 1.4351, "step": 3080 }, { "epoch": 0.2358028470840349, "grad_norm": 0.7091989545910758, "learning_rate": 9.144472383794688e-06, "loss": 1.5479, "step": 3081 }, { "epoch": 0.23587938160110208, "grad_norm": 0.6701296235131433, "learning_rate": 9.143859459667366e-06, "loss": 1.7188, "step": 3082 }, { "epoch": 0.2359559161181693, "grad_norm": 0.5520304552126182, "learning_rate": 9.14324634237192e-06, "loss": 1.3672, "step": 3083 }, { "epoch": 0.23603245063523648, "grad_norm": 0.5949118585138798, "learning_rate": 9.142633031946024e-06, "loss": 1.4386, "step": 3084 }, { "epoch": 0.2361089851523037, "grad_norm": 0.5555243005698862, "learning_rate": 9.142019528427357e-06, "loss": 1.2641, "step": 3085 }, { "epoch": 0.23618551966937087, "grad_norm": 0.630673122517327, "learning_rate": 9.141405831853621e-06, "loss": 1.768, "step": 3086 }, { "epoch": 0.2362620541864381, "grad_norm": 0.6944109119866917, "learning_rate": 9.140791942262516e-06, "loss": 1.5681, "step": 3087 }, { "epoch": 0.23633858870350527, "grad_norm": 0.6078767494080303, "learning_rate": 9.140177859691766e-06, "loss": 1.5322, "step": 3088 }, { "epoch": 0.23641512322057248, "grad_norm": 0.6100755075320545, "learning_rate": 9.139563584179103e-06, "loss": 1.473, "step": 3089 }, { "epoch": 0.23649165773763967, "grad_norm": 0.5199025165072514, "learning_rate": 9.138949115762267e-06, "loss": 1.1565, "step": 3090 }, { "epoch": 0.23656819225470688, "grad_norm": 0.580324813977237, "learning_rate": 9.138334454479012e-06, "loss": 1.2291, "step": 3091 }, { "epoch": 0.23664472677177406, "grad_norm": 0.6143923005507008, "learning_rate": 9.137719600367108e-06, "loss": 1.5195, "step": 3092 }, { "epoch": 0.23672126128884127, "grad_norm": 0.5086283246808706, "learning_rate": 9.137104553464332e-06, "loss": 1.2021, "step": 3093 }, { "epoch": 0.23679779580590846, "grad_norm": 0.6724474261860873, "learning_rate": 9.136489313808473e-06, "loss": 1.6346, "step": 3094 }, { "epoch": 0.23687433032297567, "grad_norm": 1.1728639489025616, "learning_rate": 9.135873881437335e-06, "loss": 1.2637, "step": 3095 }, { "epoch": 0.23695086484004285, "grad_norm": 0.5400517197120914, "learning_rate": 9.13525825638873e-06, "loss": 1.2009, "step": 3096 }, { "epoch": 0.23702739935711006, "grad_norm": 0.5218395671668188, "learning_rate": 9.134642438700487e-06, "loss": 1.2231, "step": 3097 }, { "epoch": 0.23710393387417725, "grad_norm": 0.5487260255218591, "learning_rate": 9.134026428410439e-06, "loss": 1.2837, "step": 3098 }, { "epoch": 0.23718046839124446, "grad_norm": 0.6494279799833653, "learning_rate": 9.13341022555644e-06, "loss": 1.4813, "step": 3099 }, { "epoch": 0.23725700290831164, "grad_norm": 0.5418866972309352, "learning_rate": 9.13279383017635e-06, "loss": 1.3979, "step": 3100 }, { "epoch": 0.23733353742537885, "grad_norm": 0.5424080426911753, "learning_rate": 9.132177242308041e-06, "loss": 1.1778, "step": 3101 }, { "epoch": 0.23741007194244604, "grad_norm": 0.541393840133471, "learning_rate": 9.1315604619894e-06, "loss": 1.2163, "step": 3102 }, { "epoch": 0.23748660645951325, "grad_norm": 0.6448211768580895, "learning_rate": 9.130943489258321e-06, "loss": 1.6085, "step": 3103 }, { "epoch": 0.23756314097658043, "grad_norm": 0.5887913906448454, "learning_rate": 9.130326324152715e-06, "loss": 1.3158, "step": 3104 }, { "epoch": 0.23763967549364765, "grad_norm": 0.5954176085428595, "learning_rate": 9.129708966710502e-06, "loss": 1.3738, "step": 3105 }, { "epoch": 0.23771621001071483, "grad_norm": 0.575924239465143, "learning_rate": 9.129091416969614e-06, "loss": 1.4091, "step": 3106 }, { "epoch": 0.23779274452778204, "grad_norm": 0.6371403734607605, "learning_rate": 9.128473674967994e-06, "loss": 1.4666, "step": 3107 }, { "epoch": 0.23786927904484922, "grad_norm": 0.5494589742078231, "learning_rate": 9.1278557407436e-06, "loss": 1.1362, "step": 3108 }, { "epoch": 0.23794581356191644, "grad_norm": 0.6332679692299988, "learning_rate": 9.127237614334398e-06, "loss": 1.5492, "step": 3109 }, { "epoch": 0.23802234807898362, "grad_norm": 0.5908747218135659, "learning_rate": 9.126619295778371e-06, "loss": 1.399, "step": 3110 }, { "epoch": 0.23809888259605083, "grad_norm": 0.5323095441732717, "learning_rate": 9.126000785113504e-06, "loss": 1.1907, "step": 3111 }, { "epoch": 0.23817541711311802, "grad_norm": 0.9660770805616796, "learning_rate": 9.125382082377808e-06, "loss": 1.7647, "step": 3112 }, { "epoch": 0.23825195163018523, "grad_norm": 0.5960580417460378, "learning_rate": 9.12476318760929e-06, "loss": 1.3177, "step": 3113 }, { "epoch": 0.2383284861472524, "grad_norm": 0.575257540307453, "learning_rate": 9.124144100845983e-06, "loss": 1.2129, "step": 3114 }, { "epoch": 0.2384050206643196, "grad_norm": 0.6960839844554212, "learning_rate": 9.123524822125922e-06, "loss": 1.7182, "step": 3115 }, { "epoch": 0.2384815551813868, "grad_norm": 0.6361313108823506, "learning_rate": 9.122905351487157e-06, "loss": 1.5592, "step": 3116 }, { "epoch": 0.238558089698454, "grad_norm": 0.5297151303616654, "learning_rate": 9.122285688967753e-06, "loss": 1.1717, "step": 3117 }, { "epoch": 0.2386346242155212, "grad_norm": 0.566583056154171, "learning_rate": 9.121665834605783e-06, "loss": 1.3273, "step": 3118 }, { "epoch": 0.23871115873258839, "grad_norm": 0.6185255297377522, "learning_rate": 9.12104578843933e-06, "loss": 1.4429, "step": 3119 }, { "epoch": 0.2387876932496556, "grad_norm": 0.7991467740727995, "learning_rate": 9.120425550506496e-06, "loss": 1.6218, "step": 3120 }, { "epoch": 0.23886422776672278, "grad_norm": 0.6437752679661424, "learning_rate": 9.119805120845385e-06, "loss": 1.5584, "step": 3121 }, { "epoch": 0.23894076228379, "grad_norm": 0.7043478492836373, "learning_rate": 9.119184499494123e-06, "loss": 1.9837, "step": 3122 }, { "epoch": 0.23901729680085718, "grad_norm": 0.6403275175747608, "learning_rate": 9.11856368649084e-06, "loss": 1.6149, "step": 3123 }, { "epoch": 0.2390938313179244, "grad_norm": 0.5666756229199282, "learning_rate": 9.117942681873682e-06, "loss": 1.4428, "step": 3124 }, { "epoch": 0.23917036583499157, "grad_norm": 0.5539106628170598, "learning_rate": 9.117321485680805e-06, "loss": 1.413, "step": 3125 }, { "epoch": 0.23924690035205878, "grad_norm": 0.6591776425251515, "learning_rate": 9.116700097950377e-06, "loss": 1.6762, "step": 3126 }, { "epoch": 0.23932343486912597, "grad_norm": 0.810502447783298, "learning_rate": 9.116078518720577e-06, "loss": 1.7021, "step": 3127 }, { "epoch": 0.23939996938619318, "grad_norm": 0.6233137020842391, "learning_rate": 9.1154567480296e-06, "loss": 1.5867, "step": 3128 }, { "epoch": 0.23947650390326036, "grad_norm": 2.6744189176635085, "learning_rate": 9.114834785915643e-06, "loss": 1.7412, "step": 3129 }, { "epoch": 0.23955303842032757, "grad_norm": 0.5429545241416844, "learning_rate": 9.114212632416927e-06, "loss": 1.3879, "step": 3130 }, { "epoch": 0.23962957293739476, "grad_norm": 0.7658097542287744, "learning_rate": 9.113590287571676e-06, "loss": 1.6039, "step": 3131 }, { "epoch": 0.23970610745446197, "grad_norm": 0.6129515168601047, "learning_rate": 9.112967751418132e-06, "loss": 1.2735, "step": 3132 }, { "epoch": 0.23978264197152915, "grad_norm": 0.5213725976778091, "learning_rate": 9.112345023994542e-06, "loss": 1.2309, "step": 3133 }, { "epoch": 0.23985917648859637, "grad_norm": 0.6291896930147709, "learning_rate": 9.111722105339168e-06, "loss": 1.4348, "step": 3134 }, { "epoch": 0.23993571100566355, "grad_norm": 0.5804116363551359, "learning_rate": 9.111098995490289e-06, "loss": 1.3026, "step": 3135 }, { "epoch": 0.24001224552273076, "grad_norm": 0.531023614594485, "learning_rate": 9.110475694486186e-06, "loss": 1.2148, "step": 3136 }, { "epoch": 0.24008878003979794, "grad_norm": 0.4629778789712716, "learning_rate": 9.109852202365157e-06, "loss": 0.9268, "step": 3137 }, { "epoch": 0.24016531455686516, "grad_norm": 0.5896129136293433, "learning_rate": 9.109228519165513e-06, "loss": 1.4156, "step": 3138 }, { "epoch": 0.24024184907393234, "grad_norm": 0.4609338078368986, "learning_rate": 9.108604644925572e-06, "loss": 0.9573, "step": 3139 }, { "epoch": 0.24031838359099955, "grad_norm": 0.7391583729403498, "learning_rate": 9.107980579683672e-06, "loss": 1.5502, "step": 3140 }, { "epoch": 0.24039491810806674, "grad_norm": 0.6631402346952806, "learning_rate": 9.107356323478152e-06, "loss": 1.6091, "step": 3141 }, { "epoch": 0.24047145262513395, "grad_norm": 0.5834214050098246, "learning_rate": 9.10673187634737e-06, "loss": 1.3219, "step": 3142 }, { "epoch": 0.24054798714220113, "grad_norm": 0.732808607949934, "learning_rate": 9.106107238329694e-06, "loss": 1.8006, "step": 3143 }, { "epoch": 0.24062452165926834, "grad_norm": 0.5303332378457858, "learning_rate": 9.105482409463506e-06, "loss": 1.2112, "step": 3144 }, { "epoch": 0.24070105617633553, "grad_norm": 0.5965692381378845, "learning_rate": 9.104857389787194e-06, "loss": 1.6779, "step": 3145 }, { "epoch": 0.24077759069340274, "grad_norm": 0.6060523027924233, "learning_rate": 9.104232179339161e-06, "loss": 1.381, "step": 3146 }, { "epoch": 0.24085412521046992, "grad_norm": 0.5912522865028086, "learning_rate": 9.103606778157824e-06, "loss": 1.3602, "step": 3147 }, { "epoch": 0.24093065972753713, "grad_norm": 0.589966306282011, "learning_rate": 9.102981186281607e-06, "loss": 1.3099, "step": 3148 }, { "epoch": 0.24100719424460432, "grad_norm": 0.6062973931843351, "learning_rate": 9.102355403748952e-06, "loss": 1.2495, "step": 3149 }, { "epoch": 0.2410837287616715, "grad_norm": 1.2853293805288188, "learning_rate": 9.101729430598305e-06, "loss": 1.4403, "step": 3150 }, { "epoch": 0.2411602632787387, "grad_norm": 0.6318484171975409, "learning_rate": 9.10110326686813e-06, "loss": 1.4002, "step": 3151 }, { "epoch": 0.2412367977958059, "grad_norm": 0.5846599404510086, "learning_rate": 9.1004769125969e-06, "loss": 1.3366, "step": 3152 }, { "epoch": 0.2413133323128731, "grad_norm": 0.6995631965620476, "learning_rate": 9.0998503678231e-06, "loss": 1.4169, "step": 3153 }, { "epoch": 0.2413898668299403, "grad_norm": 0.5425646844115817, "learning_rate": 9.099223632585225e-06, "loss": 1.1605, "step": 3154 }, { "epoch": 0.2414664013470075, "grad_norm": 0.5230366930440425, "learning_rate": 9.098596706921789e-06, "loss": 1.1746, "step": 3155 }, { "epoch": 0.2415429358640747, "grad_norm": 0.5243164523957232, "learning_rate": 9.097969590871306e-06, "loss": 1.1833, "step": 3156 }, { "epoch": 0.2416194703811419, "grad_norm": 0.6409129640370773, "learning_rate": 9.097342284472309e-06, "loss": 1.6321, "step": 3157 }, { "epoch": 0.24169600489820908, "grad_norm": 0.5532931327308588, "learning_rate": 9.096714787763344e-06, "loss": 1.4464, "step": 3158 }, { "epoch": 0.2417725394152763, "grad_norm": 0.6277188617918752, "learning_rate": 9.096087100782964e-06, "loss": 1.6613, "step": 3159 }, { "epoch": 0.24184907393234348, "grad_norm": 0.7300070210536035, "learning_rate": 9.095459223569738e-06, "loss": 1.5946, "step": 3160 }, { "epoch": 0.2419256084494107, "grad_norm": 0.6652557860567313, "learning_rate": 9.094831156162243e-06, "loss": 1.7019, "step": 3161 }, { "epoch": 0.24200214296647787, "grad_norm": 0.57260026667886, "learning_rate": 9.09420289859907e-06, "loss": 1.2855, "step": 3162 }, { "epoch": 0.24207867748354508, "grad_norm": 0.7795552084215531, "learning_rate": 9.093574450918822e-06, "loss": 1.68, "step": 3163 }, { "epoch": 0.24215521200061227, "grad_norm": 0.6039903836705921, "learning_rate": 9.092945813160112e-06, "loss": 1.4777, "step": 3164 }, { "epoch": 0.24223174651767948, "grad_norm": 0.8228312750251694, "learning_rate": 9.092316985361562e-06, "loss": 1.3519, "step": 3165 }, { "epoch": 0.24230828103474666, "grad_norm": 0.8632855731146798, "learning_rate": 9.091687967561815e-06, "loss": 1.5575, "step": 3166 }, { "epoch": 0.24238481555181388, "grad_norm": 0.6539550647376355, "learning_rate": 9.091058759799516e-06, "loss": 1.4072, "step": 3167 }, { "epoch": 0.24246135006888106, "grad_norm": 0.587713097490056, "learning_rate": 9.090429362113327e-06, "loss": 1.346, "step": 3168 }, { "epoch": 0.24253788458594827, "grad_norm": 0.6545959974045543, "learning_rate": 9.089799774541919e-06, "loss": 1.6976, "step": 3169 }, { "epoch": 0.24261441910301546, "grad_norm": 0.5575555334283373, "learning_rate": 9.089169997123975e-06, "loss": 1.339, "step": 3170 }, { "epoch": 0.24269095362008267, "grad_norm": 0.5686389875190114, "learning_rate": 9.088540029898193e-06, "loss": 1.2877, "step": 3171 }, { "epoch": 0.24276748813714985, "grad_norm": 0.660170035128635, "learning_rate": 9.087909872903278e-06, "loss": 1.6179, "step": 3172 }, { "epoch": 0.24284402265421706, "grad_norm": 0.803539025821149, "learning_rate": 9.087279526177949e-06, "loss": 1.7619, "step": 3173 }, { "epoch": 0.24292055717128425, "grad_norm": 0.619203652422892, "learning_rate": 9.086648989760938e-06, "loss": 1.5198, "step": 3174 }, { "epoch": 0.24299709168835146, "grad_norm": 0.781717504816201, "learning_rate": 9.086018263690984e-06, "loss": 1.5767, "step": 3175 }, { "epoch": 0.24307362620541864, "grad_norm": 0.5316091582174448, "learning_rate": 9.085387348006843e-06, "loss": 1.2654, "step": 3176 }, { "epoch": 0.24315016072248585, "grad_norm": 0.5433356394939403, "learning_rate": 9.08475624274728e-06, "loss": 1.3949, "step": 3177 }, { "epoch": 0.24322669523955304, "grad_norm": 0.5893836675889299, "learning_rate": 9.084124947951072e-06, "loss": 1.4509, "step": 3178 }, { "epoch": 0.24330322975662025, "grad_norm": 0.6361178384667795, "learning_rate": 9.083493463657007e-06, "loss": 1.5322, "step": 3179 }, { "epoch": 0.24337976427368743, "grad_norm": 0.6089517954252791, "learning_rate": 9.082861789903886e-06, "loss": 1.4185, "step": 3180 }, { "epoch": 0.24345629879075464, "grad_norm": 0.5474505673470077, "learning_rate": 9.08222992673052e-06, "loss": 1.2062, "step": 3181 }, { "epoch": 0.24353283330782183, "grad_norm": 0.8052195297544036, "learning_rate": 9.081597874175733e-06, "loss": 1.3022, "step": 3182 }, { "epoch": 0.243609367824889, "grad_norm": 0.6634224124400162, "learning_rate": 9.080965632278363e-06, "loss": 1.3626, "step": 3183 }, { "epoch": 0.24368590234195622, "grad_norm": 0.6763454354416232, "learning_rate": 9.080333201077252e-06, "loss": 1.6847, "step": 3184 }, { "epoch": 0.2437624368590234, "grad_norm": 0.516080274544247, "learning_rate": 9.079700580611262e-06, "loss": 1.2316, "step": 3185 }, { "epoch": 0.24383897137609062, "grad_norm": 0.5803781156077328, "learning_rate": 9.07906777091926e-06, "loss": 1.1967, "step": 3186 }, { "epoch": 0.2439155058931578, "grad_norm": 1.3692689758807615, "learning_rate": 9.07843477204013e-06, "loss": 1.5514, "step": 3187 }, { "epoch": 0.24399204041022501, "grad_norm": 0.6017192335466758, "learning_rate": 9.077801584012765e-06, "loss": 1.3172, "step": 3188 }, { "epoch": 0.2440685749272922, "grad_norm": 0.6780978568933004, "learning_rate": 9.07716820687607e-06, "loss": 1.6646, "step": 3189 }, { "epoch": 0.2441451094443594, "grad_norm": 0.5867931758524445, "learning_rate": 9.076534640668962e-06, "loss": 1.4975, "step": 3190 }, { "epoch": 0.2442216439614266, "grad_norm": 0.6066480080610349, "learning_rate": 9.075900885430366e-06, "loss": 1.2596, "step": 3191 }, { "epoch": 0.2442981784784938, "grad_norm": 0.6932794908254811, "learning_rate": 9.075266941199226e-06, "loss": 1.4306, "step": 3192 }, { "epoch": 0.244374712995561, "grad_norm": 0.6282763718500585, "learning_rate": 9.074632808014493e-06, "loss": 1.3005, "step": 3193 }, { "epoch": 0.2444512475126282, "grad_norm": 0.5552087591628135, "learning_rate": 9.073998485915129e-06, "loss": 1.2919, "step": 3194 }, { "epoch": 0.24452778202969538, "grad_norm": 0.592589129728964, "learning_rate": 9.073363974940107e-06, "loss": 1.334, "step": 3195 }, { "epoch": 0.2446043165467626, "grad_norm": 0.5376030274295786, "learning_rate": 9.072729275128415e-06, "loss": 1.2697, "step": 3196 }, { "epoch": 0.24468085106382978, "grad_norm": 0.5781965825217068, "learning_rate": 9.072094386519051e-06, "loss": 1.4081, "step": 3197 }, { "epoch": 0.244757385580897, "grad_norm": 0.4857195357852386, "learning_rate": 9.071459309151024e-06, "loss": 1.0045, "step": 3198 }, { "epoch": 0.24483392009796417, "grad_norm": 0.6080241768263784, "learning_rate": 9.070824043063355e-06, "loss": 1.439, "step": 3199 }, { "epoch": 0.2449104546150314, "grad_norm": 0.6989684890869625, "learning_rate": 9.070188588295077e-06, "loss": 1.7071, "step": 3200 }, { "epoch": 0.24498698913209857, "grad_norm": 0.5666115200095118, "learning_rate": 9.069552944885233e-06, "loss": 1.3237, "step": 3201 }, { "epoch": 0.24506352364916578, "grad_norm": 0.6474252201166979, "learning_rate": 9.068917112872882e-06, "loss": 1.5686, "step": 3202 }, { "epoch": 0.24514005816623297, "grad_norm": 0.6140623462788126, "learning_rate": 9.068281092297088e-06, "loss": 1.413, "step": 3203 }, { "epoch": 0.24521659268330018, "grad_norm": 0.6235999358095647, "learning_rate": 9.06764488319693e-06, "loss": 1.4265, "step": 3204 }, { "epoch": 0.24529312720036736, "grad_norm": 0.6108335167583451, "learning_rate": 9.067008485611504e-06, "loss": 1.4136, "step": 3205 }, { "epoch": 0.24536966171743457, "grad_norm": 0.598363090586888, "learning_rate": 9.066371899579904e-06, "loss": 1.2443, "step": 3206 }, { "epoch": 0.24544619623450176, "grad_norm": 0.6135943410120913, "learning_rate": 9.06573512514125e-06, "loss": 1.5368, "step": 3207 }, { "epoch": 0.24552273075156897, "grad_norm": 0.5442073136217579, "learning_rate": 9.065098162334662e-06, "loss": 1.2121, "step": 3208 }, { "epoch": 0.24559926526863615, "grad_norm": 1.0064047395337459, "learning_rate": 9.064461011199284e-06, "loss": 1.6959, "step": 3209 }, { "epoch": 0.24567579978570336, "grad_norm": 0.6006353599380782, "learning_rate": 9.063823671774257e-06, "loss": 1.4239, "step": 3210 }, { "epoch": 0.24575233430277055, "grad_norm": 0.5401545758946865, "learning_rate": 9.063186144098748e-06, "loss": 1.3449, "step": 3211 }, { "epoch": 0.24582886881983776, "grad_norm": 0.6952665433704399, "learning_rate": 9.062548428211923e-06, "loss": 1.3424, "step": 3212 }, { "epoch": 0.24590540333690494, "grad_norm": 0.6385827214006331, "learning_rate": 9.061910524152966e-06, "loss": 1.5141, "step": 3213 }, { "epoch": 0.24598193785397215, "grad_norm": 0.6536651745637228, "learning_rate": 9.061272431961076e-06, "loss": 1.1958, "step": 3214 }, { "epoch": 0.24605847237103934, "grad_norm": 0.6062102210554731, "learning_rate": 9.060634151675453e-06, "loss": 1.2294, "step": 3215 }, { "epoch": 0.24613500688810652, "grad_norm": 0.6265656926630601, "learning_rate": 9.05999568333532e-06, "loss": 1.5768, "step": 3216 }, { "epoch": 0.24621154140517373, "grad_norm": 0.5742262813811383, "learning_rate": 9.059357026979906e-06, "loss": 1.3974, "step": 3217 }, { "epoch": 0.24628807592224092, "grad_norm": 0.5493344515239547, "learning_rate": 9.05871818264845e-06, "loss": 1.2848, "step": 3218 }, { "epoch": 0.24636461043930813, "grad_norm": 0.5015838175707111, "learning_rate": 9.058079150380203e-06, "loss": 1.0239, "step": 3219 }, { "epoch": 0.2464411449563753, "grad_norm": 0.7300414771323662, "learning_rate": 9.057439930214433e-06, "loss": 1.6895, "step": 3220 }, { "epoch": 0.24651767947344252, "grad_norm": 0.7576160809691556, "learning_rate": 9.056800522190412e-06, "loss": 1.5839, "step": 3221 }, { "epoch": 0.2465942139905097, "grad_norm": 1.1126852000565486, "learning_rate": 9.05616092634743e-06, "loss": 1.4418, "step": 3222 }, { "epoch": 0.24667074850757692, "grad_norm": 0.6841090691643451, "learning_rate": 9.055521142724785e-06, "loss": 1.7123, "step": 3223 }, { "epoch": 0.2467472830246441, "grad_norm": 0.9665985387783297, "learning_rate": 9.054881171361784e-06, "loss": 1.7636, "step": 3224 }, { "epoch": 0.24682381754171132, "grad_norm": 0.6169624983579287, "learning_rate": 9.054241012297756e-06, "loss": 1.4803, "step": 3225 }, { "epoch": 0.2469003520587785, "grad_norm": 0.6141984547856034, "learning_rate": 9.053600665572026e-06, "loss": 1.5261, "step": 3226 }, { "epoch": 0.2469768865758457, "grad_norm": 0.5186204949687166, "learning_rate": 9.052960131223944e-06, "loss": 1.2582, "step": 3227 }, { "epoch": 0.2470534210929129, "grad_norm": 0.5676288129122636, "learning_rate": 9.052319409292864e-06, "loss": 1.284, "step": 3228 }, { "epoch": 0.2471299556099801, "grad_norm": 0.6392966165692108, "learning_rate": 9.051678499818157e-06, "loss": 1.3995, "step": 3229 }, { "epoch": 0.2472064901270473, "grad_norm": 0.6140573383158943, "learning_rate": 9.051037402839198e-06, "loss": 1.5505, "step": 3230 }, { "epoch": 0.2472830246441145, "grad_norm": 0.7129315864564049, "learning_rate": 9.05039611839538e-06, "loss": 1.4465, "step": 3231 }, { "epoch": 0.24735955916118169, "grad_norm": 0.6110332181084855, "learning_rate": 9.049754646526107e-06, "loss": 1.5602, "step": 3232 }, { "epoch": 0.2474360936782489, "grad_norm": 0.5927143849008228, "learning_rate": 9.04911298727079e-06, "loss": 1.4835, "step": 3233 }, { "epoch": 0.24751262819531608, "grad_norm": 0.551226140085256, "learning_rate": 9.048471140668858e-06, "loss": 1.1956, "step": 3234 }, { "epoch": 0.2475891627123833, "grad_norm": 0.5588592218116059, "learning_rate": 9.047829106759744e-06, "loss": 1.2377, "step": 3235 }, { "epoch": 0.24766569722945048, "grad_norm": 0.5793160869353187, "learning_rate": 9.0471868855829e-06, "loss": 1.5392, "step": 3236 }, { "epoch": 0.2477422317465177, "grad_norm": 0.7668747704581061, "learning_rate": 9.046544477177784e-06, "loss": 1.5876, "step": 3237 }, { "epoch": 0.24781876626358487, "grad_norm": 0.8628407747240295, "learning_rate": 9.045901881583865e-06, "loss": 1.5097, "step": 3238 }, { "epoch": 0.24789530078065208, "grad_norm": 1.5843090264289998, "learning_rate": 9.045259098840631e-06, "loss": 1.4534, "step": 3239 }, { "epoch": 0.24797183529771927, "grad_norm": 0.5981913790725477, "learning_rate": 9.044616128987577e-06, "loss": 1.3764, "step": 3240 }, { "epoch": 0.24804836981478648, "grad_norm": 0.5321385826966487, "learning_rate": 9.043972972064202e-06, "loss": 1.3653, "step": 3241 }, { "epoch": 0.24812490433185366, "grad_norm": 0.5470731002569674, "learning_rate": 9.043329628110031e-06, "loss": 1.3593, "step": 3242 }, { "epoch": 0.24820143884892087, "grad_norm": 0.5576958550235642, "learning_rate": 9.042686097164588e-06, "loss": 1.3581, "step": 3243 }, { "epoch": 0.24827797336598806, "grad_norm": 0.5748286721868735, "learning_rate": 9.042042379267417e-06, "loss": 1.4161, "step": 3244 }, { "epoch": 0.24835450788305527, "grad_norm": 0.5223406993717429, "learning_rate": 9.041398474458068e-06, "loss": 1.1778, "step": 3245 }, { "epoch": 0.24843104240012245, "grad_norm": 0.6048342486601034, "learning_rate": 9.040754382776102e-06, "loss": 1.6128, "step": 3246 }, { "epoch": 0.24850757691718967, "grad_norm": 0.9820640807418343, "learning_rate": 9.040110104261098e-06, "loss": 1.8466, "step": 3247 }, { "epoch": 0.24858411143425685, "grad_norm": 0.5360377116022859, "learning_rate": 9.039465638952642e-06, "loss": 1.0892, "step": 3248 }, { "epoch": 0.24866064595132406, "grad_norm": 0.5207013381568936, "learning_rate": 9.03882098689033e-06, "loss": 1.1334, "step": 3249 }, { "epoch": 0.24873718046839124, "grad_norm": 0.5877565545497926, "learning_rate": 9.038176148113774e-06, "loss": 1.3322, "step": 3250 }, { "epoch": 0.24881371498545843, "grad_norm": 0.5952722775701375, "learning_rate": 9.037531122662591e-06, "loss": 1.212, "step": 3251 }, { "epoch": 0.24889024950252564, "grad_norm": 0.57792756018947, "learning_rate": 9.036885910576416e-06, "loss": 1.2491, "step": 3252 }, { "epoch": 0.24896678401959282, "grad_norm": 0.6472642854820897, "learning_rate": 9.036240511894893e-06, "loss": 1.5585, "step": 3253 }, { "epoch": 0.24904331853666004, "grad_norm": 0.6783649607170928, "learning_rate": 9.035594926657675e-06, "loss": 1.7125, "step": 3254 }, { "epoch": 0.24911985305372722, "grad_norm": 0.5285371321005123, "learning_rate": 9.034949154904431e-06, "loss": 1.3578, "step": 3255 }, { "epoch": 0.24919638757079443, "grad_norm": 0.8117662663057703, "learning_rate": 9.034303196674836e-06, "loss": 1.6899, "step": 3256 }, { "epoch": 0.24927292208786161, "grad_norm": 0.6449509328736384, "learning_rate": 9.033657052008584e-06, "loss": 1.6284, "step": 3257 }, { "epoch": 0.24934945660492883, "grad_norm": 0.6190618139738437, "learning_rate": 9.033010720945371e-06, "loss": 1.4106, "step": 3258 }, { "epoch": 0.249425991121996, "grad_norm": 0.7071242978536956, "learning_rate": 9.032364203524916e-06, "loss": 1.5161, "step": 3259 }, { "epoch": 0.24950252563906322, "grad_norm": 0.7201402927097185, "learning_rate": 9.031717499786939e-06, "loss": 1.5529, "step": 3260 }, { "epoch": 0.2495790601561304, "grad_norm": 0.8245063990495857, "learning_rate": 9.031070609771174e-06, "loss": 1.8898, "step": 3261 }, { "epoch": 0.24965559467319762, "grad_norm": 0.6156600635679942, "learning_rate": 9.03042353351737e-06, "loss": 1.5222, "step": 3262 }, { "epoch": 0.2497321291902648, "grad_norm": 0.7479621987907202, "learning_rate": 9.029776271065285e-06, "loss": 1.5828, "step": 3263 }, { "epoch": 0.249808663707332, "grad_norm": 0.5566877159630692, "learning_rate": 9.029128822454688e-06, "loss": 1.3147, "step": 3264 }, { "epoch": 0.2498851982243992, "grad_norm": 0.7215942956663132, "learning_rate": 9.02848118772536e-06, "loss": 1.6669, "step": 3265 }, { "epoch": 0.2499617327414664, "grad_norm": 0.6360552313738286, "learning_rate": 9.027833366917097e-06, "loss": 1.2008, "step": 3266 }, { "epoch": 0.2500382672585336, "grad_norm": 0.6484312366301113, "learning_rate": 9.027185360069699e-06, "loss": 1.5683, "step": 3267 }, { "epoch": 0.2501148017756008, "grad_norm": 0.6321521614807967, "learning_rate": 9.026537167222981e-06, "loss": 1.3577, "step": 3268 }, { "epoch": 0.250191336292668, "grad_norm": 0.623787048368203, "learning_rate": 9.025888788416776e-06, "loss": 1.4525, "step": 3269 }, { "epoch": 0.25026787080973517, "grad_norm": 0.5865370260810124, "learning_rate": 9.025240223690916e-06, "loss": 1.5723, "step": 3270 }, { "epoch": 0.2503444053268024, "grad_norm": 0.5644570290624109, "learning_rate": 9.024591473085252e-06, "loss": 1.2362, "step": 3271 }, { "epoch": 0.2504209398438696, "grad_norm": 0.5386053739410984, "learning_rate": 9.023942536639647e-06, "loss": 1.1009, "step": 3272 }, { "epoch": 0.2504974743609368, "grad_norm": 0.7572762309336014, "learning_rate": 9.023293414393972e-06, "loss": 1.9037, "step": 3273 }, { "epoch": 0.25057400887800396, "grad_norm": 0.800051231404303, "learning_rate": 9.022644106388113e-06, "loss": 1.5909, "step": 3274 }, { "epoch": 0.2506505433950712, "grad_norm": 0.5222282311839674, "learning_rate": 9.021994612661964e-06, "loss": 1.1736, "step": 3275 }, { "epoch": 0.2507270779121384, "grad_norm": 1.3166271489354033, "learning_rate": 9.021344933255431e-06, "loss": 1.6208, "step": 3276 }, { "epoch": 0.25080361242920557, "grad_norm": 0.5871296322896524, "learning_rate": 9.020695068208434e-06, "loss": 1.4305, "step": 3277 }, { "epoch": 0.25088014694627275, "grad_norm": 0.7163227928151628, "learning_rate": 9.0200450175609e-06, "loss": 1.6181, "step": 3278 }, { "epoch": 0.25095668146334, "grad_norm": 0.8876494702762585, "learning_rate": 9.019394781352775e-06, "loss": 1.3441, "step": 3279 }, { "epoch": 0.2510332159804072, "grad_norm": 0.6091829814179424, "learning_rate": 9.018744359624005e-06, "loss": 1.533, "step": 3280 }, { "epoch": 0.25110975049747436, "grad_norm": 0.6282642715364671, "learning_rate": 9.018093752414561e-06, "loss": 1.3375, "step": 3281 }, { "epoch": 0.25118628501454154, "grad_norm": 0.5628337390541509, "learning_rate": 9.017442959764412e-06, "loss": 1.4649, "step": 3282 }, { "epoch": 0.2512628195316088, "grad_norm": 0.6254291975134005, "learning_rate": 9.016791981713546e-06, "loss": 1.4483, "step": 3283 }, { "epoch": 0.25133935404867597, "grad_norm": 0.6584262999088419, "learning_rate": 9.016140818301965e-06, "loss": 1.5717, "step": 3284 }, { "epoch": 0.25141588856574315, "grad_norm": 0.6682503785961129, "learning_rate": 9.015489469569674e-06, "loss": 1.6109, "step": 3285 }, { "epoch": 0.25149242308281033, "grad_norm": 0.6096778511105624, "learning_rate": 9.014837935556696e-06, "loss": 1.2651, "step": 3286 }, { "epoch": 0.2515689575998775, "grad_norm": 0.6510342041213915, "learning_rate": 9.014186216303062e-06, "loss": 1.534, "step": 3287 }, { "epoch": 0.25164549211694476, "grad_norm": 0.5945882176351136, "learning_rate": 9.013534311848816e-06, "loss": 1.3807, "step": 3288 }, { "epoch": 0.25172202663401194, "grad_norm": 0.8447296678921453, "learning_rate": 9.012882222234014e-06, "loss": 1.2351, "step": 3289 }, { "epoch": 0.2517985611510791, "grad_norm": 0.6253668899586378, "learning_rate": 9.012229947498721e-06, "loss": 1.3626, "step": 3290 }, { "epoch": 0.2518750956681463, "grad_norm": 0.6835721406939164, "learning_rate": 9.011577487683015e-06, "loss": 1.6599, "step": 3291 }, { "epoch": 0.25195163018521355, "grad_norm": 0.5113814435446744, "learning_rate": 9.010924842826984e-06, "loss": 1.1118, "step": 3292 }, { "epoch": 0.25202816470228073, "grad_norm": 0.5513821771879607, "learning_rate": 9.010272012970732e-06, "loss": 1.2706, "step": 3293 }, { "epoch": 0.2521046992193479, "grad_norm": 0.5714620217982561, "learning_rate": 9.009618998154367e-06, "loss": 1.2919, "step": 3294 }, { "epoch": 0.2521812337364151, "grad_norm": 0.7263633982027927, "learning_rate": 9.008965798418015e-06, "loss": 1.5457, "step": 3295 }, { "epoch": 0.25225776825348234, "grad_norm": 2.3182780022560117, "learning_rate": 9.008312413801808e-06, "loss": 1.5664, "step": 3296 }, { "epoch": 0.2523343027705495, "grad_norm": 0.7103326164672831, "learning_rate": 9.007658844345895e-06, "loss": 1.8613, "step": 3297 }, { "epoch": 0.2524108372876167, "grad_norm": 0.5631124208725424, "learning_rate": 9.00700509009043e-06, "loss": 1.3274, "step": 3298 }, { "epoch": 0.2524873718046839, "grad_norm": 0.6378075541470317, "learning_rate": 9.006351151075584e-06, "loss": 1.5491, "step": 3299 }, { "epoch": 0.25256390632175113, "grad_norm": 0.5131143076469415, "learning_rate": 9.005697027341533e-06, "loss": 1.1226, "step": 3300 }, { "epoch": 0.2526404408388183, "grad_norm": 0.6319721550605131, "learning_rate": 9.005042718928475e-06, "loss": 1.4469, "step": 3301 }, { "epoch": 0.2527169753558855, "grad_norm": 0.8552289539992665, "learning_rate": 9.004388225876608e-06, "loss": 1.7555, "step": 3302 }, { "epoch": 0.2527935098729527, "grad_norm": 2.861538801972237, "learning_rate": 9.003733548226146e-06, "loss": 1.6346, "step": 3303 }, { "epoch": 0.2528700443900199, "grad_norm": 0.4873387699810874, "learning_rate": 9.003078686017315e-06, "loss": 0.9098, "step": 3304 }, { "epoch": 0.2529465789070871, "grad_norm": 0.6679989176210899, "learning_rate": 9.002423639290353e-06, "loss": 1.531, "step": 3305 }, { "epoch": 0.2530231134241543, "grad_norm": 0.562264867838976, "learning_rate": 9.001768408085507e-06, "loss": 1.4225, "step": 3306 }, { "epoch": 0.2530996479412215, "grad_norm": 0.7853037002070502, "learning_rate": 9.001112992443036e-06, "loss": 1.4486, "step": 3307 }, { "epoch": 0.2531761824582887, "grad_norm": 0.47812989729090927, "learning_rate": 9.000457392403212e-06, "loss": 0.8921, "step": 3308 }, { "epoch": 0.2532527169753559, "grad_norm": 0.5844594092847442, "learning_rate": 8.999801608006315e-06, "loss": 1.2837, "step": 3309 }, { "epoch": 0.2533292514924231, "grad_norm": 0.578924203716155, "learning_rate": 8.999145639292641e-06, "loss": 1.4125, "step": 3310 }, { "epoch": 0.25340578600949026, "grad_norm": 0.6600459919022038, "learning_rate": 8.998489486302491e-06, "loss": 1.6798, "step": 3311 }, { "epoch": 0.2534823205265575, "grad_norm": 0.6085443516847856, "learning_rate": 8.997833149076185e-06, "loss": 1.4481, "step": 3312 }, { "epoch": 0.2535588550436247, "grad_norm": 0.49325712723463283, "learning_rate": 8.997176627654048e-06, "loss": 1.0097, "step": 3313 }, { "epoch": 0.25363538956069187, "grad_norm": 0.717838196337139, "learning_rate": 8.99651992207642e-06, "loss": 1.37, "step": 3314 }, { "epoch": 0.25371192407775905, "grad_norm": 0.5626368318947889, "learning_rate": 8.995863032383649e-06, "loss": 1.1762, "step": 3315 }, { "epoch": 0.2537884585948263, "grad_norm": 0.6316450763550079, "learning_rate": 8.995205958616098e-06, "loss": 1.5176, "step": 3316 }, { "epoch": 0.2538649931118935, "grad_norm": 0.8110391780609597, "learning_rate": 8.994548700814139e-06, "loss": 2.0103, "step": 3317 }, { "epoch": 0.25394152762896066, "grad_norm": 0.6041650437875522, "learning_rate": 8.993891259018155e-06, "loss": 1.526, "step": 3318 }, { "epoch": 0.25401806214602785, "grad_norm": 0.7658007771465808, "learning_rate": 8.993233633268541e-06, "loss": 1.6033, "step": 3319 }, { "epoch": 0.25409459666309503, "grad_norm": 0.8993256302149156, "learning_rate": 8.992575823605707e-06, "loss": 1.6966, "step": 3320 }, { "epoch": 0.25417113118016227, "grad_norm": 0.6586180472246821, "learning_rate": 8.991917830070066e-06, "loss": 1.447, "step": 3321 }, { "epoch": 0.25424766569722945, "grad_norm": 0.5777320866104343, "learning_rate": 8.991259652702052e-06, "loss": 1.2093, "step": 3322 }, { "epoch": 0.25432420021429664, "grad_norm": 0.6778239790280142, "learning_rate": 8.9906012915421e-06, "loss": 1.8146, "step": 3323 }, { "epoch": 0.2544007347313638, "grad_norm": 0.5723745400622686, "learning_rate": 8.989942746630666e-06, "loss": 1.1535, "step": 3324 }, { "epoch": 0.25447726924843106, "grad_norm": 0.5602307300868434, "learning_rate": 8.989284018008212e-06, "loss": 1.5304, "step": 3325 }, { "epoch": 0.25455380376549824, "grad_norm": 0.6048970235904882, "learning_rate": 8.988625105715211e-06, "loss": 1.4522, "step": 3326 }, { "epoch": 0.2546303382825654, "grad_norm": 0.620508476219885, "learning_rate": 8.987966009792148e-06, "loss": 1.4883, "step": 3327 }, { "epoch": 0.2547068727996326, "grad_norm": 0.5765190381552868, "learning_rate": 8.987306730279524e-06, "loss": 1.4423, "step": 3328 }, { "epoch": 0.25478340731669985, "grad_norm": 0.6502880409083325, "learning_rate": 8.986647267217841e-06, "loss": 1.5883, "step": 3329 }, { "epoch": 0.25485994183376703, "grad_norm": 0.6093751161116732, "learning_rate": 8.985987620647623e-06, "loss": 1.2817, "step": 3330 }, { "epoch": 0.2549364763508342, "grad_norm": 1.622286499712984, "learning_rate": 8.9853277906094e-06, "loss": 1.2062, "step": 3331 }, { "epoch": 0.2550130108679014, "grad_norm": 0.6111577888102662, "learning_rate": 8.98466777714371e-06, "loss": 1.3896, "step": 3332 }, { "epoch": 0.25508954538496864, "grad_norm": 0.5948837773431797, "learning_rate": 8.984007580291112e-06, "loss": 1.4248, "step": 3333 }, { "epoch": 0.2551660799020358, "grad_norm": 0.5904806416827256, "learning_rate": 8.983347200092167e-06, "loss": 1.4063, "step": 3334 }, { "epoch": 0.255242614419103, "grad_norm": 0.636997242300718, "learning_rate": 8.982686636587454e-06, "loss": 1.6004, "step": 3335 }, { "epoch": 0.2553191489361702, "grad_norm": 0.5327474341694292, "learning_rate": 8.982025889817554e-06, "loss": 1.226, "step": 3336 }, { "epoch": 0.25539568345323743, "grad_norm": 0.5205814578136969, "learning_rate": 8.98136495982307e-06, "loss": 1.2036, "step": 3337 }, { "epoch": 0.2554722179703046, "grad_norm": 0.6443644339804638, "learning_rate": 8.98070384664461e-06, "loss": 1.5698, "step": 3338 }, { "epoch": 0.2555487524873718, "grad_norm": 0.6300002851811847, "learning_rate": 8.980042550322795e-06, "loss": 1.5707, "step": 3339 }, { "epoch": 0.255625287004439, "grad_norm": 0.6065189675163757, "learning_rate": 8.979381070898257e-06, "loss": 1.1835, "step": 3340 }, { "epoch": 0.2557018215215062, "grad_norm": 0.7419027361112494, "learning_rate": 8.978719408411641e-06, "loss": 1.5368, "step": 3341 }, { "epoch": 0.2557783560385734, "grad_norm": 0.521093917046281, "learning_rate": 8.978057562903597e-06, "loss": 1.2077, "step": 3342 }, { "epoch": 0.2558548905556406, "grad_norm": 0.5711683907981333, "learning_rate": 8.977395534414797e-06, "loss": 1.3357, "step": 3343 }, { "epoch": 0.2559314250727078, "grad_norm": 0.5797222137444634, "learning_rate": 8.976733322985913e-06, "loss": 1.1944, "step": 3344 }, { "epoch": 0.256007959589775, "grad_norm": 0.613540779533733, "learning_rate": 8.976070928657635e-06, "loss": 1.5809, "step": 3345 }, { "epoch": 0.2560844941068422, "grad_norm": 0.522761341603768, "learning_rate": 8.97540835147066e-06, "loss": 1.0765, "step": 3346 }, { "epoch": 0.2561610286239094, "grad_norm": 0.6405714827768698, "learning_rate": 8.974745591465706e-06, "loss": 1.5487, "step": 3347 }, { "epoch": 0.25623756314097657, "grad_norm": 0.5209848233187651, "learning_rate": 8.974082648683486e-06, "loss": 1.2636, "step": 3348 }, { "epoch": 0.2563140976580438, "grad_norm": 0.6249935737767129, "learning_rate": 8.97341952316474e-06, "loss": 1.4513, "step": 3349 }, { "epoch": 0.256390632175111, "grad_norm": 0.607002931136032, "learning_rate": 8.972756214950207e-06, "loss": 1.4266, "step": 3350 }, { "epoch": 0.2564671666921782, "grad_norm": 0.5472475640470145, "learning_rate": 8.972092724080646e-06, "loss": 1.2411, "step": 3351 }, { "epoch": 0.25654370120924536, "grad_norm": 0.5761638510273275, "learning_rate": 8.971429050596825e-06, "loss": 1.405, "step": 3352 }, { "epoch": 0.25662023572631254, "grad_norm": 0.9027824549916551, "learning_rate": 8.970765194539519e-06, "loss": 1.7125, "step": 3353 }, { "epoch": 0.2566967702433798, "grad_norm": 0.7892620699958762, "learning_rate": 8.97010115594952e-06, "loss": 1.3341, "step": 3354 }, { "epoch": 0.25677330476044696, "grad_norm": 0.5664362197146878, "learning_rate": 8.969436934867624e-06, "loss": 1.3625, "step": 3355 }, { "epoch": 0.25684983927751415, "grad_norm": 0.4884039994550569, "learning_rate": 8.968772531334647e-06, "loss": 1.0395, "step": 3356 }, { "epoch": 0.25692637379458133, "grad_norm": 0.8234203511311835, "learning_rate": 8.968107945391411e-06, "loss": 1.5617, "step": 3357 }, { "epoch": 0.25700290831164857, "grad_norm": 0.5867050362077992, "learning_rate": 8.967443177078749e-06, "loss": 1.4019, "step": 3358 }, { "epoch": 0.25707944282871575, "grad_norm": 0.640938002465678, "learning_rate": 8.966778226437511e-06, "loss": 1.6372, "step": 3359 }, { "epoch": 0.25715597734578294, "grad_norm": 1.0347715241654072, "learning_rate": 8.966113093508545e-06, "loss": 1.635, "step": 3360 }, { "epoch": 0.2572325118628501, "grad_norm": 0.812906901001404, "learning_rate": 8.965447778332726e-06, "loss": 1.214, "step": 3361 }, { "epoch": 0.25730904637991736, "grad_norm": 0.8700116367395473, "learning_rate": 8.96478228095093e-06, "loss": 1.5174, "step": 3362 }, { "epoch": 0.25738558089698454, "grad_norm": 0.5151690320847487, "learning_rate": 8.964116601404048e-06, "loss": 1.128, "step": 3363 }, { "epoch": 0.25746211541405173, "grad_norm": 0.6265163134796894, "learning_rate": 8.963450739732982e-06, "loss": 1.5662, "step": 3364 }, { "epoch": 0.2575386499311189, "grad_norm": 0.5953967913911674, "learning_rate": 8.96278469597864e-06, "loss": 1.2925, "step": 3365 }, { "epoch": 0.25761518444818615, "grad_norm": 0.595232077612484, "learning_rate": 8.962118470181952e-06, "loss": 1.4863, "step": 3366 }, { "epoch": 0.25769171896525334, "grad_norm": 0.646636176674663, "learning_rate": 8.96145206238385e-06, "loss": 1.6386, "step": 3367 }, { "epoch": 0.2577682534823205, "grad_norm": 0.6440477093431428, "learning_rate": 8.96078547262528e-06, "loss": 1.6693, "step": 3368 }, { "epoch": 0.2578447879993877, "grad_norm": 0.6036365949130488, "learning_rate": 8.9601187009472e-06, "loss": 1.3268, "step": 3369 }, { "epoch": 0.25792132251645494, "grad_norm": 0.6225935624046032, "learning_rate": 8.959451747390577e-06, "loss": 1.5836, "step": 3370 }, { "epoch": 0.2579978570335221, "grad_norm": 0.5462729002806985, "learning_rate": 8.958784611996391e-06, "loss": 1.175, "step": 3371 }, { "epoch": 0.2580743915505893, "grad_norm": 1.7094168450321536, "learning_rate": 8.958117294805634e-06, "loss": 1.8992, "step": 3372 }, { "epoch": 0.2581509260676565, "grad_norm": 0.6458442077888142, "learning_rate": 8.957449795859308e-06, "loss": 1.6888, "step": 3373 }, { "epoch": 0.25822746058472373, "grad_norm": 1.060614768984823, "learning_rate": 8.956782115198425e-06, "loss": 1.4386, "step": 3374 }, { "epoch": 0.2583039951017909, "grad_norm": 0.6060152631263467, "learning_rate": 8.956114252864011e-06, "loss": 1.6694, "step": 3375 }, { "epoch": 0.2583805296188581, "grad_norm": 0.5623939053276723, "learning_rate": 8.955446208897098e-06, "loss": 1.3339, "step": 3376 }, { "epoch": 0.2584570641359253, "grad_norm": 0.6478992419351856, "learning_rate": 8.954777983338736e-06, "loss": 1.2456, "step": 3377 }, { "epoch": 0.2585335986529925, "grad_norm": 0.616167898034332, "learning_rate": 8.95410957622998e-06, "loss": 1.4985, "step": 3378 }, { "epoch": 0.2586101331700597, "grad_norm": 0.5437734612979134, "learning_rate": 8.953440987611903e-06, "loss": 1.1746, "step": 3379 }, { "epoch": 0.2586866676871269, "grad_norm": 0.6796338329748484, "learning_rate": 8.952772217525581e-06, "loss": 1.7983, "step": 3380 }, { "epoch": 0.2587632022041941, "grad_norm": 0.5017086719601757, "learning_rate": 8.952103266012108e-06, "loss": 1.1183, "step": 3381 }, { "epoch": 0.2588397367212613, "grad_norm": 0.5065417596692893, "learning_rate": 8.951434133112585e-06, "loss": 1.2243, "step": 3382 }, { "epoch": 0.2589162712383285, "grad_norm": 0.6419873090875938, "learning_rate": 8.950764818868125e-06, "loss": 1.6572, "step": 3383 }, { "epoch": 0.2589928057553957, "grad_norm": 0.8387911749665797, "learning_rate": 8.950095323319854e-06, "loss": 1.656, "step": 3384 }, { "epoch": 0.25906934027246287, "grad_norm": 0.6610795256673254, "learning_rate": 8.949425646508906e-06, "loss": 1.615, "step": 3385 }, { "epoch": 0.25914587478953005, "grad_norm": 0.691808618325749, "learning_rate": 8.948755788476431e-06, "loss": 1.4106, "step": 3386 }, { "epoch": 0.2592224093065973, "grad_norm": 0.6228364352503721, "learning_rate": 8.948085749263585e-06, "loss": 1.4255, "step": 3387 }, { "epoch": 0.2592989438236645, "grad_norm": 0.5871694002395471, "learning_rate": 8.947415528911538e-06, "loss": 1.2626, "step": 3388 }, { "epoch": 0.25937547834073166, "grad_norm": 0.6183524596772697, "learning_rate": 8.946745127461469e-06, "loss": 1.2748, "step": 3389 }, { "epoch": 0.25945201285779884, "grad_norm": 0.630529088376259, "learning_rate": 8.946074544954569e-06, "loss": 1.6062, "step": 3390 }, { "epoch": 0.2595285473748661, "grad_norm": 0.8455790596470869, "learning_rate": 8.945403781432042e-06, "loss": 1.8431, "step": 3391 }, { "epoch": 0.25960508189193326, "grad_norm": 0.6276045647582295, "learning_rate": 8.944732836935103e-06, "loss": 1.5645, "step": 3392 }, { "epoch": 0.25968161640900045, "grad_norm": 0.6313839706920654, "learning_rate": 8.944061711504975e-06, "loss": 1.3078, "step": 3393 }, { "epoch": 0.25975815092606763, "grad_norm": 0.6054932283048413, "learning_rate": 8.943390405182895e-06, "loss": 1.4009, "step": 3394 }, { "epoch": 0.25983468544313487, "grad_norm": 0.7726568423497437, "learning_rate": 8.942718918010108e-06, "loss": 1.6474, "step": 3395 }, { "epoch": 0.25991121996020206, "grad_norm": 0.599494826410889, "learning_rate": 8.942047250027874e-06, "loss": 1.514, "step": 3396 }, { "epoch": 0.25998775447726924, "grad_norm": 0.6782208688536101, "learning_rate": 8.941375401277463e-06, "loss": 1.732, "step": 3397 }, { "epoch": 0.2600642889943364, "grad_norm": 1.155948867696403, "learning_rate": 8.940703371800154e-06, "loss": 1.6858, "step": 3398 }, { "epoch": 0.26014082351140366, "grad_norm": 0.6489153501821443, "learning_rate": 8.940031161637236e-06, "loss": 1.4727, "step": 3399 }, { "epoch": 0.26021735802847085, "grad_norm": 0.9380391915139475, "learning_rate": 8.939358770830017e-06, "loss": 1.481, "step": 3400 }, { "epoch": 0.26029389254553803, "grad_norm": 0.5866337232494885, "learning_rate": 8.938686199419807e-06, "loss": 1.3215, "step": 3401 }, { "epoch": 0.2603704270626052, "grad_norm": 0.8422749616936513, "learning_rate": 8.938013447447935e-06, "loss": 1.4252, "step": 3402 }, { "epoch": 0.26044696157967245, "grad_norm": 0.610949188911165, "learning_rate": 8.937340514955731e-06, "loss": 1.4931, "step": 3403 }, { "epoch": 0.26052349609673964, "grad_norm": 0.5769613629378636, "learning_rate": 8.936667401984546e-06, "loss": 1.5145, "step": 3404 }, { "epoch": 0.2606000306138068, "grad_norm": 0.5585429328550094, "learning_rate": 8.935994108575737e-06, "loss": 1.3312, "step": 3405 }, { "epoch": 0.260676565130874, "grad_norm": 0.539246967330039, "learning_rate": 8.93532063477067e-06, "loss": 1.2345, "step": 3406 }, { "epoch": 0.26075309964794124, "grad_norm": 0.4889035754747771, "learning_rate": 8.934646980610732e-06, "loss": 1.0171, "step": 3407 }, { "epoch": 0.26082963416500843, "grad_norm": 0.6288083669385178, "learning_rate": 8.93397314613731e-06, "loss": 1.3763, "step": 3408 }, { "epoch": 0.2609061686820756, "grad_norm": 0.5522592683972382, "learning_rate": 8.933299131391805e-06, "loss": 1.2584, "step": 3409 }, { "epoch": 0.2609827031991428, "grad_norm": 0.522119027801266, "learning_rate": 8.932624936415634e-06, "loss": 1.1648, "step": 3410 }, { "epoch": 0.26105923771621004, "grad_norm": 0.7400130107484161, "learning_rate": 8.93195056125022e-06, "loss": 1.6857, "step": 3411 }, { "epoch": 0.2611357722332772, "grad_norm": 0.5569242603492037, "learning_rate": 8.931276005936998e-06, "loss": 1.0805, "step": 3412 }, { "epoch": 0.2612123067503444, "grad_norm": 0.9385284111486153, "learning_rate": 8.930601270517415e-06, "loss": 1.3612, "step": 3413 }, { "epoch": 0.2612888412674116, "grad_norm": 0.7545126249174545, "learning_rate": 8.92992635503293e-06, "loss": 1.2978, "step": 3414 }, { "epoch": 0.2613653757844788, "grad_norm": 0.5690562087627564, "learning_rate": 8.92925125952501e-06, "loss": 1.277, "step": 3415 }, { "epoch": 0.261441910301546, "grad_norm": 0.5132265389394406, "learning_rate": 8.928575984035135e-06, "loss": 1.2515, "step": 3416 }, { "epoch": 0.2615184448186132, "grad_norm": 0.7385348383095829, "learning_rate": 8.9279005286048e-06, "loss": 1.4733, "step": 3417 }, { "epoch": 0.2615949793356804, "grad_norm": 1.266365227082782, "learning_rate": 8.927224893275502e-06, "loss": 1.7572, "step": 3418 }, { "epoch": 0.26167151385274756, "grad_norm": 0.8012300513117667, "learning_rate": 8.926549078088756e-06, "loss": 1.5734, "step": 3419 }, { "epoch": 0.2617480483698148, "grad_norm": 0.5854521696365306, "learning_rate": 8.925873083086085e-06, "loss": 1.4476, "step": 3420 }, { "epoch": 0.261824582886882, "grad_norm": 0.5791602486832522, "learning_rate": 8.925196908309026e-06, "loss": 1.32, "step": 3421 }, { "epoch": 0.26190111740394917, "grad_norm": 0.5788114630554966, "learning_rate": 8.924520553799125e-06, "loss": 1.1939, "step": 3422 }, { "epoch": 0.26197765192101635, "grad_norm": 0.6905864040461394, "learning_rate": 8.923844019597936e-06, "loss": 1.7711, "step": 3423 }, { "epoch": 0.2620541864380836, "grad_norm": 0.5384354813551414, "learning_rate": 8.923167305747031e-06, "loss": 1.0823, "step": 3424 }, { "epoch": 0.2621307209551508, "grad_norm": 0.7771744591140217, "learning_rate": 8.92249041228799e-06, "loss": 1.5723, "step": 3425 }, { "epoch": 0.26220725547221796, "grad_norm": 0.5789142854821234, "learning_rate": 8.9218133392624e-06, "loss": 1.3808, "step": 3426 }, { "epoch": 0.26228378998928514, "grad_norm": 0.6445129760559685, "learning_rate": 8.921136086711864e-06, "loss": 1.4051, "step": 3427 }, { "epoch": 0.2623603245063524, "grad_norm": 0.5601072750876006, "learning_rate": 8.920458654677993e-06, "loss": 1.3729, "step": 3428 }, { "epoch": 0.26243685902341957, "grad_norm": 0.5537660094420971, "learning_rate": 8.919781043202414e-06, "loss": 1.3283, "step": 3429 }, { "epoch": 0.26251339354048675, "grad_norm": 0.7085245350183556, "learning_rate": 8.919103252326757e-06, "loss": 1.6511, "step": 3430 }, { "epoch": 0.26258992805755393, "grad_norm": 0.6080830832331413, "learning_rate": 8.91842528209267e-06, "loss": 1.3274, "step": 3431 }, { "epoch": 0.2626664625746212, "grad_norm": 0.5782127257541769, "learning_rate": 8.917747132541813e-06, "loss": 1.4602, "step": 3432 }, { "epoch": 0.26274299709168836, "grad_norm": 0.613366237453535, "learning_rate": 8.917068803715845e-06, "loss": 1.4115, "step": 3433 }, { "epoch": 0.26281953160875554, "grad_norm": 0.7356043196928024, "learning_rate": 8.916390295656451e-06, "loss": 1.8858, "step": 3434 }, { "epoch": 0.2628960661258227, "grad_norm": 0.593812435388459, "learning_rate": 8.91571160840532e-06, "loss": 1.4967, "step": 3435 }, { "epoch": 0.26297260064288996, "grad_norm": 0.6562525448407238, "learning_rate": 8.915032742004151e-06, "loss": 1.5176, "step": 3436 }, { "epoch": 0.26304913515995715, "grad_norm": 0.579424981504042, "learning_rate": 8.914353696494656e-06, "loss": 1.6537, "step": 3437 }, { "epoch": 0.26312566967702433, "grad_norm": 0.6058621995122673, "learning_rate": 8.913674471918557e-06, "loss": 1.6543, "step": 3438 }, { "epoch": 0.2632022041940915, "grad_norm": 0.6525414916645165, "learning_rate": 8.912995068317587e-06, "loss": 1.3558, "step": 3439 }, { "epoch": 0.26327873871115876, "grad_norm": 0.7122626598292496, "learning_rate": 8.912315485733494e-06, "loss": 1.6399, "step": 3440 }, { "epoch": 0.26335527322822594, "grad_norm": 0.5769747299496463, "learning_rate": 8.91163572420803e-06, "loss": 1.3511, "step": 3441 }, { "epoch": 0.2634318077452931, "grad_norm": 0.566987657292196, "learning_rate": 8.910955783782964e-06, "loss": 1.3892, "step": 3442 }, { "epoch": 0.2635083422623603, "grad_norm": 0.5020332447971544, "learning_rate": 8.910275664500073e-06, "loss": 1.0082, "step": 3443 }, { "epoch": 0.26358487677942755, "grad_norm": 0.5831901997560142, "learning_rate": 8.909595366401144e-06, "loss": 1.42, "step": 3444 }, { "epoch": 0.26366141129649473, "grad_norm": 0.5719104742098355, "learning_rate": 8.908914889527976e-06, "loss": 1.3746, "step": 3445 }, { "epoch": 0.2637379458135619, "grad_norm": 0.5805612641549518, "learning_rate": 8.908234233922386e-06, "loss": 1.389, "step": 3446 }, { "epoch": 0.2638144803306291, "grad_norm": 0.7130793537981377, "learning_rate": 8.907553399626188e-06, "loss": 1.2954, "step": 3447 }, { "epoch": 0.26389101484769634, "grad_norm": 0.4841055186168499, "learning_rate": 8.906872386681216e-06, "loss": 0.995, "step": 3448 }, { "epoch": 0.2639675493647635, "grad_norm": 0.5213516721253582, "learning_rate": 8.906191195129315e-06, "loss": 1.113, "step": 3449 }, { "epoch": 0.2640440838818307, "grad_norm": 0.48120745007008164, "learning_rate": 8.90550982501234e-06, "loss": 1.0392, "step": 3450 }, { "epoch": 0.2641206183988979, "grad_norm": 0.5599510953950011, "learning_rate": 8.904828276372154e-06, "loss": 1.2489, "step": 3451 }, { "epoch": 0.2641971529159651, "grad_norm": 0.679248268417149, "learning_rate": 8.904146549250638e-06, "loss": 1.3991, "step": 3452 }, { "epoch": 0.2642736874330323, "grad_norm": 0.6303563721283776, "learning_rate": 8.903464643689675e-06, "loss": 1.6302, "step": 3453 }, { "epoch": 0.2643502219500995, "grad_norm": 0.6592049693602878, "learning_rate": 8.902782559731162e-06, "loss": 1.4151, "step": 3454 }, { "epoch": 0.2644267564671667, "grad_norm": 0.5506038805114946, "learning_rate": 8.902100297417014e-06, "loss": 1.2951, "step": 3455 }, { "epoch": 0.26450329098423386, "grad_norm": 0.5670747155917847, "learning_rate": 8.901417856789148e-06, "loss": 1.2991, "step": 3456 }, { "epoch": 0.2645798255013011, "grad_norm": 0.5893532306664044, "learning_rate": 8.900735237889493e-06, "loss": 1.3896, "step": 3457 }, { "epoch": 0.2646563600183683, "grad_norm": 0.6023878624159764, "learning_rate": 8.900052440759996e-06, "loss": 1.3936, "step": 3458 }, { "epoch": 0.26473289453543547, "grad_norm": 0.5312067539014457, "learning_rate": 8.899369465442606e-06, "loss": 1.2845, "step": 3459 }, { "epoch": 0.26480942905250265, "grad_norm": 1.0146772894082274, "learning_rate": 8.898686311979287e-06, "loss": 1.2439, "step": 3460 }, { "epoch": 0.2648859635695699, "grad_norm": 1.0672890947335725, "learning_rate": 8.89800298041202e-06, "loss": 1.522, "step": 3461 }, { "epoch": 0.2649624980866371, "grad_norm": 0.5643786434615116, "learning_rate": 8.897319470782783e-06, "loss": 1.329, "step": 3462 }, { "epoch": 0.26503903260370426, "grad_norm": 0.5951958785034942, "learning_rate": 8.896635783133576e-06, "loss": 1.43, "step": 3463 }, { "epoch": 0.26511556712077144, "grad_norm": 0.6072768871775489, "learning_rate": 8.895951917506409e-06, "loss": 1.4926, "step": 3464 }, { "epoch": 0.2651921016378387, "grad_norm": 0.5594926917936035, "learning_rate": 8.8952678739433e-06, "loss": 1.2353, "step": 3465 }, { "epoch": 0.26526863615490587, "grad_norm": 0.540457727982046, "learning_rate": 8.894583652486276e-06, "loss": 1.2829, "step": 3466 }, { "epoch": 0.26534517067197305, "grad_norm": 0.7894321003404909, "learning_rate": 8.893899253177377e-06, "loss": 1.2405, "step": 3467 }, { "epoch": 0.26542170518904024, "grad_norm": 0.6192461672882728, "learning_rate": 8.89321467605866e-06, "loss": 1.4913, "step": 3468 }, { "epoch": 0.2654982397061075, "grad_norm": 0.5178474862912659, "learning_rate": 8.892529921172184e-06, "loss": 1.1311, "step": 3469 }, { "epoch": 0.26557477422317466, "grad_norm": 0.6886636551049277, "learning_rate": 8.891844988560022e-06, "loss": 1.4526, "step": 3470 }, { "epoch": 0.26565130874024184, "grad_norm": 0.8475857273400127, "learning_rate": 8.891159878264258e-06, "loss": 1.4843, "step": 3471 }, { "epoch": 0.265727843257309, "grad_norm": 0.560912125121099, "learning_rate": 8.89047459032699e-06, "loss": 1.4325, "step": 3472 }, { "epoch": 0.26580437777437627, "grad_norm": 0.7302689100931841, "learning_rate": 8.88978912479032e-06, "loss": 1.5751, "step": 3473 }, { "epoch": 0.26588091229144345, "grad_norm": 0.797764169705971, "learning_rate": 8.889103481696369e-06, "loss": 1.44, "step": 3474 }, { "epoch": 0.26595744680851063, "grad_norm": 0.609447740244025, "learning_rate": 8.888417661087265e-06, "loss": 1.412, "step": 3475 }, { "epoch": 0.2660339813255778, "grad_norm": 0.6429322462731517, "learning_rate": 8.887731663005142e-06, "loss": 1.6576, "step": 3476 }, { "epoch": 0.26611051584264506, "grad_norm": 0.6153248933137965, "learning_rate": 8.887045487492153e-06, "loss": 1.386, "step": 3477 }, { "epoch": 0.26618705035971224, "grad_norm": 0.6176043536483721, "learning_rate": 8.88635913459046e-06, "loss": 1.666, "step": 3478 }, { "epoch": 0.2662635848767794, "grad_norm": 0.5539907442269266, "learning_rate": 8.885672604342233e-06, "loss": 1.2657, "step": 3479 }, { "epoch": 0.2663401193938466, "grad_norm": 0.5508790576141662, "learning_rate": 8.884985896789656e-06, "loss": 1.1739, "step": 3480 }, { "epoch": 0.26641665391091385, "grad_norm": 0.5592517803838281, "learning_rate": 8.88429901197492e-06, "loss": 1.2616, "step": 3481 }, { "epoch": 0.26649318842798103, "grad_norm": 0.6851060632751401, "learning_rate": 8.883611949940228e-06, "loss": 1.7552, "step": 3482 }, { "epoch": 0.2665697229450482, "grad_norm": 0.6109579522645324, "learning_rate": 8.8829247107278e-06, "loss": 1.4643, "step": 3483 }, { "epoch": 0.2666462574621154, "grad_norm": 0.9134907611198402, "learning_rate": 8.882237294379859e-06, "loss": 1.5658, "step": 3484 }, { "epoch": 0.26672279197918264, "grad_norm": 0.5682618405070584, "learning_rate": 8.881549700938643e-06, "loss": 1.4818, "step": 3485 }, { "epoch": 0.2667993264962498, "grad_norm": 0.5745650049622986, "learning_rate": 8.880861930446397e-06, "loss": 1.2968, "step": 3486 }, { "epoch": 0.266875861013317, "grad_norm": 0.6327468571761793, "learning_rate": 8.880173982945384e-06, "loss": 1.683, "step": 3487 }, { "epoch": 0.2669523955303842, "grad_norm": 0.5616286989738482, "learning_rate": 8.879485858477873e-06, "loss": 1.4913, "step": 3488 }, { "epoch": 0.2670289300474514, "grad_norm": 0.5766109613983665, "learning_rate": 8.878797557086138e-06, "loss": 1.3202, "step": 3489 }, { "epoch": 0.2671054645645186, "grad_norm": 0.6560556046620124, "learning_rate": 8.878109078812477e-06, "loss": 1.2779, "step": 3490 }, { "epoch": 0.2671819990815858, "grad_norm": 0.6380338909073573, "learning_rate": 8.877420423699193e-06, "loss": 1.4865, "step": 3491 }, { "epoch": 0.267258533598653, "grad_norm": 0.6255429572083387, "learning_rate": 8.876731591788594e-06, "loss": 1.532, "step": 3492 }, { "epoch": 0.26733506811572016, "grad_norm": 0.5396283717711022, "learning_rate": 8.876042583123006e-06, "loss": 1.3376, "step": 3493 }, { "epoch": 0.2674116026327874, "grad_norm": 0.8138679734784323, "learning_rate": 8.875353397744766e-06, "loss": 1.5204, "step": 3494 }, { "epoch": 0.2674881371498546, "grad_norm": 0.6023385069616198, "learning_rate": 8.874664035696215e-06, "loss": 1.2153, "step": 3495 }, { "epoch": 0.26756467166692177, "grad_norm": 0.5818366815323276, "learning_rate": 8.873974497019712e-06, "loss": 1.4898, "step": 3496 }, { "epoch": 0.26764120618398896, "grad_norm": 0.5476732408928123, "learning_rate": 8.873284781757625e-06, "loss": 1.3617, "step": 3497 }, { "epoch": 0.2677177407010562, "grad_norm": 0.6079620433362374, "learning_rate": 8.872594889952333e-06, "loss": 1.4398, "step": 3498 }, { "epoch": 0.2677942752181234, "grad_norm": 0.6012229948420516, "learning_rate": 8.871904821646223e-06, "loss": 1.5797, "step": 3499 }, { "epoch": 0.26787080973519056, "grad_norm": 0.5512921999576286, "learning_rate": 8.871214576881693e-06, "loss": 1.2971, "step": 3500 }, { "epoch": 0.26794734425225775, "grad_norm": 0.9326371630428554, "learning_rate": 8.870524155701159e-06, "loss": 1.963, "step": 3501 }, { "epoch": 0.268023878769325, "grad_norm": 0.5410028004854016, "learning_rate": 8.869833558147039e-06, "loss": 1.2819, "step": 3502 }, { "epoch": 0.26810041328639217, "grad_norm": 0.6255740274912025, "learning_rate": 8.869142784261764e-06, "loss": 1.4808, "step": 3503 }, { "epoch": 0.26817694780345935, "grad_norm": 0.5363740320861168, "learning_rate": 8.86845183408778e-06, "loss": 1.0029, "step": 3504 }, { "epoch": 0.26825348232052654, "grad_norm": 0.8660756295307892, "learning_rate": 8.867760707667542e-06, "loss": 1.6577, "step": 3505 }, { "epoch": 0.2683300168375938, "grad_norm": 0.5019996334577125, "learning_rate": 8.867069405043514e-06, "loss": 1.0742, "step": 3506 }, { "epoch": 0.26840655135466096, "grad_norm": 0.5289857630836305, "learning_rate": 8.866377926258167e-06, "loss": 1.1688, "step": 3507 }, { "epoch": 0.26848308587172814, "grad_norm": 0.5073471563990157, "learning_rate": 8.865686271353995e-06, "loss": 1.1122, "step": 3508 }, { "epoch": 0.26855962038879533, "grad_norm": 1.0547999592154154, "learning_rate": 8.86499444037349e-06, "loss": 1.7205, "step": 3509 }, { "epoch": 0.26863615490586257, "grad_norm": 0.48020690603997196, "learning_rate": 8.864302433359162e-06, "loss": 1.0186, "step": 3510 }, { "epoch": 0.26871268942292975, "grad_norm": 0.8896268132912883, "learning_rate": 8.86361025035353e-06, "loss": 1.6548, "step": 3511 }, { "epoch": 0.26878922393999694, "grad_norm": 0.6044843870986601, "learning_rate": 8.862917891399123e-06, "loss": 1.3939, "step": 3512 }, { "epoch": 0.2688657584570641, "grad_norm": 0.5331701269845696, "learning_rate": 8.862225356538483e-06, "loss": 1.1944, "step": 3513 }, { "epoch": 0.26894229297413136, "grad_norm": 0.5909514625666088, "learning_rate": 8.86153264581416e-06, "loss": 1.4865, "step": 3514 }, { "epoch": 0.26901882749119854, "grad_norm": 0.6604732752425165, "learning_rate": 8.860839759268718e-06, "loss": 1.481, "step": 3515 }, { "epoch": 0.2690953620082657, "grad_norm": 0.5918216519542152, "learning_rate": 8.860146696944729e-06, "loss": 1.4079, "step": 3516 }, { "epoch": 0.2691718965253329, "grad_norm": 0.7121282404667701, "learning_rate": 8.859453458884775e-06, "loss": 1.5623, "step": 3517 }, { "epoch": 0.26924843104240015, "grad_norm": 0.5765381462143397, "learning_rate": 8.858760045131453e-06, "loss": 1.4511, "step": 3518 }, { "epoch": 0.26932496555946733, "grad_norm": 0.5863902040859339, "learning_rate": 8.858066455727369e-06, "loss": 1.4412, "step": 3519 }, { "epoch": 0.2694015000765345, "grad_norm": 0.6379716031418247, "learning_rate": 8.857372690715137e-06, "loss": 1.5714, "step": 3520 }, { "epoch": 0.2694780345936017, "grad_norm": 0.48987632662174996, "learning_rate": 8.856678750137385e-06, "loss": 1.0162, "step": 3521 }, { "epoch": 0.2695545691106689, "grad_norm": 0.9876253035687975, "learning_rate": 8.85598463403675e-06, "loss": 1.4545, "step": 3522 }, { "epoch": 0.2696311036277361, "grad_norm": 0.5465516290988456, "learning_rate": 8.855290342455884e-06, "loss": 1.1968, "step": 3523 }, { "epoch": 0.2697076381448033, "grad_norm": 0.64794372414627, "learning_rate": 8.854595875437441e-06, "loss": 1.6367, "step": 3524 }, { "epoch": 0.2697841726618705, "grad_norm": 0.7677040354276707, "learning_rate": 8.853901233024094e-06, "loss": 1.572, "step": 3525 }, { "epoch": 0.2698607071789377, "grad_norm": 0.6083310841261927, "learning_rate": 8.853206415258526e-06, "loss": 1.5131, "step": 3526 }, { "epoch": 0.2699372416960049, "grad_norm": 0.6110940216186909, "learning_rate": 8.852511422183424e-06, "loss": 1.5126, "step": 3527 }, { "epoch": 0.2700137762130721, "grad_norm": 1.291716425981629, "learning_rate": 8.851816253841493e-06, "loss": 1.3122, "step": 3528 }, { "epoch": 0.2700903107301393, "grad_norm": 0.5474408317016691, "learning_rate": 8.851120910275446e-06, "loss": 1.3086, "step": 3529 }, { "epoch": 0.27016684524720647, "grad_norm": 0.9142657105856322, "learning_rate": 8.850425391528008e-06, "loss": 1.8054, "step": 3530 }, { "epoch": 0.2702433797642737, "grad_norm": 0.6847823769993207, "learning_rate": 8.849729697641913e-06, "loss": 1.5238, "step": 3531 }, { "epoch": 0.2703199142813409, "grad_norm": 0.6271895270452204, "learning_rate": 8.849033828659905e-06, "loss": 1.5519, "step": 3532 }, { "epoch": 0.2703964487984081, "grad_norm": 0.5870602697674999, "learning_rate": 8.848337784624743e-06, "loss": 1.368, "step": 3533 }, { "epoch": 0.27047298331547526, "grad_norm": 0.7106295340521336, "learning_rate": 8.84764156557919e-06, "loss": 1.4441, "step": 3534 }, { "epoch": 0.2705495178325425, "grad_norm": 0.6050851265818868, "learning_rate": 8.846945171566029e-06, "loss": 1.4876, "step": 3535 }, { "epoch": 0.2706260523496097, "grad_norm": 0.6176342624731553, "learning_rate": 8.846248602628046e-06, "loss": 1.4838, "step": 3536 }, { "epoch": 0.27070258686667686, "grad_norm": 0.6537884827993979, "learning_rate": 8.84555185880804e-06, "loss": 1.4408, "step": 3537 }, { "epoch": 0.27077912138374405, "grad_norm": 0.6516210650847598, "learning_rate": 8.844854940148819e-06, "loss": 1.4993, "step": 3538 }, { "epoch": 0.2708556559008113, "grad_norm": 0.5162080584672749, "learning_rate": 8.844157846693207e-06, "loss": 1.2618, "step": 3539 }, { "epoch": 0.27093219041787847, "grad_norm": 0.6274870308767259, "learning_rate": 8.843460578484036e-06, "loss": 1.5697, "step": 3540 }, { "epoch": 0.27100872493494566, "grad_norm": 0.5043670847360606, "learning_rate": 8.842763135564147e-06, "loss": 1.1038, "step": 3541 }, { "epoch": 0.27108525945201284, "grad_norm": 0.6732899007045421, "learning_rate": 8.84206551797639e-06, "loss": 1.6331, "step": 3542 }, { "epoch": 0.2711617939690801, "grad_norm": 0.5871193307243329, "learning_rate": 8.841367725763634e-06, "loss": 1.4495, "step": 3543 }, { "epoch": 0.27123832848614726, "grad_norm": 0.5833439362738877, "learning_rate": 8.84066975896875e-06, "loss": 1.5327, "step": 3544 }, { "epoch": 0.27131486300321445, "grad_norm": 0.5650267343691712, "learning_rate": 8.839971617634623e-06, "loss": 1.239, "step": 3545 }, { "epoch": 0.27139139752028163, "grad_norm": 0.7661255270821923, "learning_rate": 8.839273301804151e-06, "loss": 1.2947, "step": 3546 }, { "epoch": 0.27146793203734887, "grad_norm": 0.5584705301817857, "learning_rate": 8.83857481152024e-06, "loss": 1.2718, "step": 3547 }, { "epoch": 0.27154446655441605, "grad_norm": 0.6488234437216428, "learning_rate": 8.837876146825806e-06, "loss": 1.4357, "step": 3548 }, { "epoch": 0.27162100107148324, "grad_norm": 0.6371027813591514, "learning_rate": 8.837177307763778e-06, "loss": 1.5715, "step": 3549 }, { "epoch": 0.2716975355885504, "grad_norm": 0.6000978143216956, "learning_rate": 8.836478294377093e-06, "loss": 1.2705, "step": 3550 }, { "epoch": 0.27177407010561766, "grad_norm": 0.5911265649982782, "learning_rate": 8.835779106708704e-06, "loss": 1.5186, "step": 3551 }, { "epoch": 0.27185060462268484, "grad_norm": 0.7212793807576058, "learning_rate": 8.83507974480157e-06, "loss": 1.5134, "step": 3552 }, { "epoch": 0.271927139139752, "grad_norm": 0.7912941071239109, "learning_rate": 8.83438020869866e-06, "loss": 1.1798, "step": 3553 }, { "epoch": 0.2720036736568192, "grad_norm": 0.5835514083347395, "learning_rate": 8.833680498442958e-06, "loss": 1.2827, "step": 3554 }, { "epoch": 0.2720802081738864, "grad_norm": 0.723911251243632, "learning_rate": 8.832980614077454e-06, "loss": 1.5874, "step": 3555 }, { "epoch": 0.27215674269095363, "grad_norm": 0.5415441766792516, "learning_rate": 8.832280555645152e-06, "loss": 1.291, "step": 3556 }, { "epoch": 0.2722332772080208, "grad_norm": 0.6084209598307635, "learning_rate": 8.831580323189066e-06, "loss": 1.4031, "step": 3557 }, { "epoch": 0.272309811725088, "grad_norm": 0.5610162919220906, "learning_rate": 8.83087991675222e-06, "loss": 1.1995, "step": 3558 }, { "epoch": 0.2723863462421552, "grad_norm": 0.7521090277645529, "learning_rate": 8.830179336377649e-06, "loss": 1.7528, "step": 3559 }, { "epoch": 0.2724628807592224, "grad_norm": 0.5905763900616012, "learning_rate": 8.8294785821084e-06, "loss": 1.2013, "step": 3560 }, { "epoch": 0.2725394152762896, "grad_norm": 0.6024507099269946, "learning_rate": 8.828777653987528e-06, "loss": 1.455, "step": 3561 }, { "epoch": 0.2726159497933568, "grad_norm": 0.5682970042099119, "learning_rate": 8.8280765520581e-06, "loss": 1.4885, "step": 3562 }, { "epoch": 0.272692484310424, "grad_norm": 0.58006467554575, "learning_rate": 8.827375276363194e-06, "loss": 1.3719, "step": 3563 }, { "epoch": 0.2727690188274912, "grad_norm": 0.6232255035609012, "learning_rate": 8.826673826945899e-06, "loss": 1.5137, "step": 3564 }, { "epoch": 0.2728455533445584, "grad_norm": 0.5530576085867999, "learning_rate": 8.825972203849316e-06, "loss": 1.3568, "step": 3565 }, { "epoch": 0.2729220878616256, "grad_norm": 0.6024951246791647, "learning_rate": 8.825270407116548e-06, "loss": 1.3658, "step": 3566 }, { "epoch": 0.27299862237869277, "grad_norm": 0.5773168526733049, "learning_rate": 8.824568436790724e-06, "loss": 1.4403, "step": 3567 }, { "epoch": 0.27307515689576, "grad_norm": 0.5780750988905827, "learning_rate": 8.823866292914971e-06, "loss": 1.2789, "step": 3568 }, { "epoch": 0.2731516914128272, "grad_norm": 0.534636943208935, "learning_rate": 8.82316397553243e-06, "loss": 1.0364, "step": 3569 }, { "epoch": 0.2732282259298944, "grad_norm": 0.7169318683621608, "learning_rate": 8.822461484686256e-06, "loss": 1.6904, "step": 3570 }, { "epoch": 0.27330476044696156, "grad_norm": 0.5922116226180532, "learning_rate": 8.821758820419612e-06, "loss": 1.3453, "step": 3571 }, { "epoch": 0.2733812949640288, "grad_norm": 0.7030193757144565, "learning_rate": 8.821055982775667e-06, "loss": 1.5509, "step": 3572 }, { "epoch": 0.273457829481096, "grad_norm": 0.6216818143742466, "learning_rate": 8.820352971797612e-06, "loss": 1.568, "step": 3573 }, { "epoch": 0.27353436399816317, "grad_norm": 0.6303024932229521, "learning_rate": 8.819649787528637e-06, "loss": 1.5611, "step": 3574 }, { "epoch": 0.27361089851523035, "grad_norm": 0.5823656608454506, "learning_rate": 8.818946430011951e-06, "loss": 1.4215, "step": 3575 }, { "epoch": 0.2736874330322976, "grad_norm": 0.6550631189157534, "learning_rate": 8.81824289929077e-06, "loss": 1.6193, "step": 3576 }, { "epoch": 0.2737639675493648, "grad_norm": 0.5626036595092203, "learning_rate": 8.81753919540832e-06, "loss": 1.2192, "step": 3577 }, { "epoch": 0.27384050206643196, "grad_norm": 0.7472385795745617, "learning_rate": 8.81683531840784e-06, "loss": 1.5166, "step": 3578 }, { "epoch": 0.27391703658349914, "grad_norm": 0.5281545952906993, "learning_rate": 8.816131268332575e-06, "loss": 1.1332, "step": 3579 }, { "epoch": 0.2739935711005664, "grad_norm": 0.6669292636506378, "learning_rate": 8.815427045225787e-06, "loss": 1.2956, "step": 3580 }, { "epoch": 0.27407010561763356, "grad_norm": 0.7160119506920407, "learning_rate": 8.814722649130745e-06, "loss": 1.4459, "step": 3581 }, { "epoch": 0.27414664013470075, "grad_norm": 0.5979683644328537, "learning_rate": 8.814018080090728e-06, "loss": 1.4968, "step": 3582 }, { "epoch": 0.27422317465176793, "grad_norm": 0.5923766620828622, "learning_rate": 8.81331333814903e-06, "loss": 1.4896, "step": 3583 }, { "epoch": 0.27429970916883517, "grad_norm": 0.7020191805968746, "learning_rate": 8.812608423348949e-06, "loss": 1.2201, "step": 3584 }, { "epoch": 0.27437624368590235, "grad_norm": 0.6246874409147168, "learning_rate": 8.811903335733798e-06, "loss": 1.5009, "step": 3585 }, { "epoch": 0.27445277820296954, "grad_norm": 0.5412271387783371, "learning_rate": 8.8111980753469e-06, "loss": 1.1988, "step": 3586 }, { "epoch": 0.2745293127200367, "grad_norm": 0.527415201027386, "learning_rate": 8.81049264223159e-06, "loss": 1.208, "step": 3587 }, { "epoch": 0.27460584723710396, "grad_norm": 0.5634251140031026, "learning_rate": 8.809787036431208e-06, "loss": 1.3525, "step": 3588 }, { "epoch": 0.27468238175417115, "grad_norm": 0.6574479058206769, "learning_rate": 8.809081257989114e-06, "loss": 1.526, "step": 3589 }, { "epoch": 0.27475891627123833, "grad_norm": 0.6466184016753238, "learning_rate": 8.808375306948667e-06, "loss": 1.2697, "step": 3590 }, { "epoch": 0.2748354507883055, "grad_norm": 0.5311590423351233, "learning_rate": 8.807669183353247e-06, "loss": 1.2723, "step": 3591 }, { "epoch": 0.2749119853053727, "grad_norm": 0.6997838003880911, "learning_rate": 8.80696288724624e-06, "loss": 1.5939, "step": 3592 }, { "epoch": 0.27498851982243994, "grad_norm": 0.7104200775581856, "learning_rate": 8.806256418671039e-06, "loss": 1.7841, "step": 3593 }, { "epoch": 0.2750650543395071, "grad_norm": 0.5944059281608299, "learning_rate": 8.805549777671057e-06, "loss": 1.5348, "step": 3594 }, { "epoch": 0.2751415888565743, "grad_norm": 0.5206178878265881, "learning_rate": 8.80484296428971e-06, "loss": 1.2007, "step": 3595 }, { "epoch": 0.2752181233736415, "grad_norm": 0.6757008505386013, "learning_rate": 8.804135978570423e-06, "loss": 1.3494, "step": 3596 }, { "epoch": 0.2752946578907087, "grad_norm": 0.6411946814826576, "learning_rate": 8.803428820556641e-06, "loss": 1.4401, "step": 3597 }, { "epoch": 0.2753711924077759, "grad_norm": 0.5677211218446411, "learning_rate": 8.802721490291813e-06, "loss": 1.3279, "step": 3598 }, { "epoch": 0.2754477269248431, "grad_norm": 0.5835699728919611, "learning_rate": 8.802013987819398e-06, "loss": 1.2958, "step": 3599 }, { "epoch": 0.2755242614419103, "grad_norm": 0.6103489343587111, "learning_rate": 8.801306313182866e-06, "loss": 1.5695, "step": 3600 }, { "epoch": 0.2756007959589775, "grad_norm": 0.5123203250962035, "learning_rate": 8.800598466425699e-06, "loss": 1.2813, "step": 3601 }, { "epoch": 0.2756773304760447, "grad_norm": 0.5734417132394409, "learning_rate": 8.79989044759139e-06, "loss": 1.4799, "step": 3602 }, { "epoch": 0.2757538649931119, "grad_norm": 0.6103871134871939, "learning_rate": 8.799182256723443e-06, "loss": 1.7049, "step": 3603 }, { "epoch": 0.27583039951017907, "grad_norm": 0.6181320192437647, "learning_rate": 8.798473893865368e-06, "loss": 1.4168, "step": 3604 }, { "epoch": 0.2759069340272463, "grad_norm": 0.5985655819889756, "learning_rate": 8.797765359060695e-06, "loss": 1.4022, "step": 3605 }, { "epoch": 0.2759834685443135, "grad_norm": 0.8849962027225373, "learning_rate": 8.797056652352952e-06, "loss": 1.3894, "step": 3606 }, { "epoch": 0.2760600030613807, "grad_norm": 1.0611407135056434, "learning_rate": 8.796347773785688e-06, "loss": 1.596, "step": 3607 }, { "epoch": 0.27613653757844786, "grad_norm": 0.635383724634486, "learning_rate": 8.795638723402456e-06, "loss": 1.428, "step": 3608 }, { "epoch": 0.2762130720955151, "grad_norm": 0.5790241111766913, "learning_rate": 8.794929501246824e-06, "loss": 1.2684, "step": 3609 }, { "epoch": 0.2762896066125823, "grad_norm": 0.5747352621422426, "learning_rate": 8.794220107362367e-06, "loss": 1.397, "step": 3610 }, { "epoch": 0.27636614112964947, "grad_norm": 0.5576091042326117, "learning_rate": 8.793510541792675e-06, "loss": 1.2913, "step": 3611 }, { "epoch": 0.27644267564671665, "grad_norm": 0.5208965639944735, "learning_rate": 8.792800804581345e-06, "loss": 1.2282, "step": 3612 }, { "epoch": 0.2765192101637839, "grad_norm": 0.5278812780260157, "learning_rate": 8.792090895771983e-06, "loss": 1.3026, "step": 3613 }, { "epoch": 0.2765957446808511, "grad_norm": 0.588122201789881, "learning_rate": 8.79138081540821e-06, "loss": 1.2683, "step": 3614 }, { "epoch": 0.27667227919791826, "grad_norm": 0.6030070167018412, "learning_rate": 8.790670563533654e-06, "loss": 1.4113, "step": 3615 }, { "epoch": 0.27674881371498544, "grad_norm": 0.7340413996983781, "learning_rate": 8.789960140191958e-06, "loss": 1.5156, "step": 3616 }, { "epoch": 0.2768253482320527, "grad_norm": 0.6162373471750664, "learning_rate": 8.789249545426771e-06, "loss": 1.274, "step": 3617 }, { "epoch": 0.27690188274911987, "grad_norm": 0.5713566819199289, "learning_rate": 8.788538779281753e-06, "loss": 1.3107, "step": 3618 }, { "epoch": 0.27697841726618705, "grad_norm": 0.5923789976123058, "learning_rate": 8.787827841800575e-06, "loss": 1.3073, "step": 3619 }, { "epoch": 0.27705495178325423, "grad_norm": 0.5688373403040591, "learning_rate": 8.787116733026922e-06, "loss": 1.2133, "step": 3620 }, { "epoch": 0.2771314863003215, "grad_norm": 0.6106182503928569, "learning_rate": 8.786405453004482e-06, "loss": 1.5069, "step": 3621 }, { "epoch": 0.27720802081738866, "grad_norm": 0.521544405694256, "learning_rate": 8.785694001776965e-06, "loss": 1.2655, "step": 3622 }, { "epoch": 0.27728455533445584, "grad_norm": 0.5481086440314781, "learning_rate": 8.78498237938808e-06, "loss": 1.2382, "step": 3623 }, { "epoch": 0.277361089851523, "grad_norm": 0.5978940479439943, "learning_rate": 8.78427058588155e-06, "loss": 1.387, "step": 3624 }, { "epoch": 0.2774376243685902, "grad_norm": 0.6540964168768322, "learning_rate": 8.783558621301113e-06, "loss": 1.3007, "step": 3625 }, { "epoch": 0.27751415888565745, "grad_norm": 0.5995443213572583, "learning_rate": 8.782846485690514e-06, "loss": 1.5573, "step": 3626 }, { "epoch": 0.27759069340272463, "grad_norm": 0.6625015468070091, "learning_rate": 8.782134179093507e-06, "loss": 1.6742, "step": 3627 }, { "epoch": 0.2776672279197918, "grad_norm": 0.509347392860829, "learning_rate": 8.781421701553859e-06, "loss": 1.1773, "step": 3628 }, { "epoch": 0.277743762436859, "grad_norm": 0.6762770011562699, "learning_rate": 8.780709053115347e-06, "loss": 1.8477, "step": 3629 }, { "epoch": 0.27782029695392624, "grad_norm": 0.7126489833052773, "learning_rate": 8.779996233821757e-06, "loss": 1.5183, "step": 3630 }, { "epoch": 0.2778968314709934, "grad_norm": 0.6150790140176577, "learning_rate": 8.77928324371689e-06, "loss": 1.3841, "step": 3631 }, { "epoch": 0.2779733659880606, "grad_norm": 0.5175463740506744, "learning_rate": 8.77857008284455e-06, "loss": 1.2448, "step": 3632 }, { "epoch": 0.2780499005051278, "grad_norm": 0.5822356326769346, "learning_rate": 8.77785675124856e-06, "loss": 1.3173, "step": 3633 }, { "epoch": 0.27812643502219503, "grad_norm": 0.4414127872292318, "learning_rate": 8.777143248972745e-06, "loss": 0.9699, "step": 3634 }, { "epoch": 0.2782029695392622, "grad_norm": 0.709010292985821, "learning_rate": 8.776429576060947e-06, "loss": 1.269, "step": 3635 }, { "epoch": 0.2782795040563294, "grad_norm": 0.5400505593853092, "learning_rate": 8.775715732557016e-06, "loss": 1.3316, "step": 3636 }, { "epoch": 0.2783560385733966, "grad_norm": 0.5674742409195827, "learning_rate": 8.775001718504813e-06, "loss": 1.2854, "step": 3637 }, { "epoch": 0.2784325730904638, "grad_norm": 0.6105263618948713, "learning_rate": 8.774287533948209e-06, "loss": 1.437, "step": 3638 }, { "epoch": 0.278509107607531, "grad_norm": 0.7116127955931304, "learning_rate": 8.773573178931085e-06, "loss": 1.9078, "step": 3639 }, { "epoch": 0.2785856421245982, "grad_norm": 0.6111858694530059, "learning_rate": 8.772858653497332e-06, "loss": 1.4542, "step": 3640 }, { "epoch": 0.27866217664166537, "grad_norm": 0.6707278188558881, "learning_rate": 8.772143957690856e-06, "loss": 1.5367, "step": 3641 }, { "epoch": 0.2787387111587326, "grad_norm": 0.6057108398700497, "learning_rate": 8.771429091555567e-06, "loss": 1.4596, "step": 3642 }, { "epoch": 0.2788152456757998, "grad_norm": 0.5598806693037085, "learning_rate": 8.77071405513539e-06, "loss": 1.3753, "step": 3643 }, { "epoch": 0.278891780192867, "grad_norm": 0.5611803692168602, "learning_rate": 8.769998848474259e-06, "loss": 1.2545, "step": 3644 }, { "epoch": 0.27896831470993416, "grad_norm": 0.5885117028605641, "learning_rate": 8.769283471616116e-06, "loss": 1.515, "step": 3645 }, { "epoch": 0.2790448492270014, "grad_norm": 0.6070916347608392, "learning_rate": 8.76856792460492e-06, "loss": 1.4344, "step": 3646 }, { "epoch": 0.2791213837440686, "grad_norm": 0.5306445445265184, "learning_rate": 8.767852207484635e-06, "loss": 1.2789, "step": 3647 }, { "epoch": 0.27919791826113577, "grad_norm": 1.559111643231933, "learning_rate": 8.767136320299234e-06, "loss": 1.6159, "step": 3648 }, { "epoch": 0.27927445277820295, "grad_norm": 1.5705180946824453, "learning_rate": 8.766420263092707e-06, "loss": 1.7729, "step": 3649 }, { "epoch": 0.2793509872952702, "grad_norm": 0.5894811699448765, "learning_rate": 8.765704035909048e-06, "loss": 1.3991, "step": 3650 }, { "epoch": 0.2794275218123374, "grad_norm": 0.6383779374752037, "learning_rate": 8.764987638792267e-06, "loss": 1.3143, "step": 3651 }, { "epoch": 0.27950405632940456, "grad_norm": 0.6678533459861249, "learning_rate": 8.764271071786378e-06, "loss": 1.6121, "step": 3652 }, { "epoch": 0.27958059084647174, "grad_norm": 0.5308138666848184, "learning_rate": 8.76355433493541e-06, "loss": 1.0665, "step": 3653 }, { "epoch": 0.279657125363539, "grad_norm": 0.5839280911579253, "learning_rate": 8.762837428283404e-06, "loss": 1.5408, "step": 3654 }, { "epoch": 0.27973365988060617, "grad_norm": 0.8038861000096386, "learning_rate": 8.762120351874405e-06, "loss": 1.6893, "step": 3655 }, { "epoch": 0.27981019439767335, "grad_norm": 0.6545720093286296, "learning_rate": 8.761403105752476e-06, "loss": 1.4542, "step": 3656 }, { "epoch": 0.27988672891474053, "grad_norm": 0.5654379061372935, "learning_rate": 8.760685689961685e-06, "loss": 1.3708, "step": 3657 }, { "epoch": 0.2799632634318077, "grad_norm": 1.206689928318509, "learning_rate": 8.759968104546112e-06, "loss": 1.3713, "step": 3658 }, { "epoch": 0.28003979794887496, "grad_norm": 0.4888781788165021, "learning_rate": 8.759250349549848e-06, "loss": 1.0885, "step": 3659 }, { "epoch": 0.28011633246594214, "grad_norm": 0.5737965905469711, "learning_rate": 8.758532425016994e-06, "loss": 1.2631, "step": 3660 }, { "epoch": 0.2801928669830093, "grad_norm": 0.7135121349337896, "learning_rate": 8.75781433099166e-06, "loss": 1.8132, "step": 3661 }, { "epoch": 0.2802694015000765, "grad_norm": 0.5286142383021356, "learning_rate": 8.757096067517972e-06, "loss": 1.2324, "step": 3662 }, { "epoch": 0.28034593601714375, "grad_norm": 0.6144143558809743, "learning_rate": 8.75637763464006e-06, "loss": 1.3466, "step": 3663 }, { "epoch": 0.28042247053421093, "grad_norm": 0.7474853282343621, "learning_rate": 8.755659032402062e-06, "loss": 1.4364, "step": 3664 }, { "epoch": 0.2804990050512781, "grad_norm": 0.5959628784416016, "learning_rate": 8.754940260848139e-06, "loss": 1.4149, "step": 3665 }, { "epoch": 0.2805755395683453, "grad_norm": 0.5820565742076965, "learning_rate": 8.754221320022448e-06, "loss": 1.4045, "step": 3666 }, { "epoch": 0.28065207408541254, "grad_norm": 0.7468429511631482, "learning_rate": 8.753502209969167e-06, "loss": 1.5058, "step": 3667 }, { "epoch": 0.2807286086024797, "grad_norm": 0.5305776146850415, "learning_rate": 8.752782930732477e-06, "loss": 1.1897, "step": 3668 }, { "epoch": 0.2808051431195469, "grad_norm": 0.601357738179021, "learning_rate": 8.752063482356577e-06, "loss": 1.3817, "step": 3669 }, { "epoch": 0.2808816776366141, "grad_norm": 0.6370953577338682, "learning_rate": 8.751343864885669e-06, "loss": 1.5076, "step": 3670 }, { "epoch": 0.28095821215368133, "grad_norm": 0.6988845779583842, "learning_rate": 8.750624078363968e-06, "loss": 1.5855, "step": 3671 }, { "epoch": 0.2810347466707485, "grad_norm": 0.6119770612990132, "learning_rate": 8.7499041228357e-06, "loss": 1.3832, "step": 3672 }, { "epoch": 0.2811112811878157, "grad_norm": 0.6018308738536003, "learning_rate": 8.749183998345103e-06, "loss": 1.5203, "step": 3673 }, { "epoch": 0.2811878157048829, "grad_norm": 0.5959511337750982, "learning_rate": 8.748463704936424e-06, "loss": 1.3389, "step": 3674 }, { "epoch": 0.2812643502219501, "grad_norm": 0.7467707253168461, "learning_rate": 8.747743242653916e-06, "loss": 1.4521, "step": 3675 }, { "epoch": 0.2813408847390173, "grad_norm": 1.3728546093453937, "learning_rate": 8.747022611541851e-06, "loss": 1.5706, "step": 3676 }, { "epoch": 0.2814174192560845, "grad_norm": 0.5779422308447663, "learning_rate": 8.746301811644506e-06, "loss": 1.3439, "step": 3677 }, { "epoch": 0.2814939537731517, "grad_norm": 0.8520005257232307, "learning_rate": 8.745580843006165e-06, "loss": 1.54, "step": 3678 }, { "epoch": 0.2815704882902189, "grad_norm": 0.6691464739116886, "learning_rate": 8.74485970567113e-06, "loss": 1.4978, "step": 3679 }, { "epoch": 0.2816470228072861, "grad_norm": 0.7480429671942278, "learning_rate": 8.74413839968371e-06, "loss": 1.4635, "step": 3680 }, { "epoch": 0.2817235573243533, "grad_norm": 0.5348065992817821, "learning_rate": 8.743416925088221e-06, "loss": 1.3496, "step": 3681 }, { "epoch": 0.28180009184142046, "grad_norm": 0.8199045618506375, "learning_rate": 8.742695281928997e-06, "loss": 1.5374, "step": 3682 }, { "epoch": 0.2818766263584877, "grad_norm": 0.5266511285915523, "learning_rate": 8.741973470250376e-06, "loss": 1.1109, "step": 3683 }, { "epoch": 0.2819531608755549, "grad_norm": 0.598320183476823, "learning_rate": 8.741251490096709e-06, "loss": 1.4113, "step": 3684 }, { "epoch": 0.28202969539262207, "grad_norm": 1.1236667367264088, "learning_rate": 8.740529341512354e-06, "loss": 1.6434, "step": 3685 }, { "epoch": 0.28210622990968925, "grad_norm": 0.5632041356232158, "learning_rate": 8.739807024541683e-06, "loss": 1.3025, "step": 3686 }, { "epoch": 0.2821827644267565, "grad_norm": 0.5534714359242461, "learning_rate": 8.739084539229077e-06, "loss": 1.2665, "step": 3687 }, { "epoch": 0.2822592989438237, "grad_norm": 0.6731479646246024, "learning_rate": 8.738361885618933e-06, "loss": 1.504, "step": 3688 }, { "epoch": 0.28233583346089086, "grad_norm": 0.565637416970439, "learning_rate": 8.737639063755646e-06, "loss": 1.2893, "step": 3689 }, { "epoch": 0.28241236797795805, "grad_norm": 0.5479935753663517, "learning_rate": 8.73691607368363e-06, "loss": 1.2797, "step": 3690 }, { "epoch": 0.28248890249502523, "grad_norm": 0.548286245020332, "learning_rate": 8.73619291544731e-06, "loss": 1.2728, "step": 3691 }, { "epoch": 0.28256543701209247, "grad_norm": 0.6657172779940426, "learning_rate": 8.735469589091116e-06, "loss": 1.3523, "step": 3692 }, { "epoch": 0.28264197152915965, "grad_norm": 1.3582105308328702, "learning_rate": 8.734746094659495e-06, "loss": 1.6709, "step": 3693 }, { "epoch": 0.28271850604622684, "grad_norm": 0.6013426428892562, "learning_rate": 8.734022432196898e-06, "loss": 1.3897, "step": 3694 }, { "epoch": 0.282795040563294, "grad_norm": 0.5348676594204039, "learning_rate": 8.733298601747789e-06, "loss": 1.2449, "step": 3695 }, { "epoch": 0.28287157508036126, "grad_norm": 0.6054852824241213, "learning_rate": 8.732574603356643e-06, "loss": 1.3656, "step": 3696 }, { "epoch": 0.28294810959742844, "grad_norm": 0.6082892404148224, "learning_rate": 8.731850437067944e-06, "loss": 1.3766, "step": 3697 }, { "epoch": 0.2830246441144956, "grad_norm": 0.5867805508169249, "learning_rate": 8.731126102926188e-06, "loss": 1.3663, "step": 3698 }, { "epoch": 0.2831011786315628, "grad_norm": 0.6811592149795286, "learning_rate": 8.73040160097588e-06, "loss": 1.33, "step": 3699 }, { "epoch": 0.28317771314863005, "grad_norm": 0.6230453180313732, "learning_rate": 8.729676931261534e-06, "loss": 1.4607, "step": 3700 }, { "epoch": 0.28325424766569723, "grad_norm": 0.589046070905386, "learning_rate": 8.72895209382768e-06, "loss": 1.3518, "step": 3701 }, { "epoch": 0.2833307821827644, "grad_norm": 0.6075346334099744, "learning_rate": 8.728227088718849e-06, "loss": 1.4443, "step": 3702 }, { "epoch": 0.2834073166998316, "grad_norm": 0.588767610840727, "learning_rate": 8.727501915979591e-06, "loss": 1.4245, "step": 3703 }, { "epoch": 0.28348385121689884, "grad_norm": 1.2404632918685814, "learning_rate": 8.72677657565446e-06, "loss": 1.4499, "step": 3704 }, { "epoch": 0.283560385733966, "grad_norm": 0.6515667152789805, "learning_rate": 8.726051067788027e-06, "loss": 1.6106, "step": 3705 }, { "epoch": 0.2836369202510332, "grad_norm": 0.732586909974566, "learning_rate": 8.725325392424865e-06, "loss": 1.5348, "step": 3706 }, { "epoch": 0.2837134547681004, "grad_norm": 0.6608847276077866, "learning_rate": 8.724599549609564e-06, "loss": 1.4651, "step": 3707 }, { "epoch": 0.28378998928516763, "grad_norm": 0.5616336442079984, "learning_rate": 8.723873539386722e-06, "loss": 1.0844, "step": 3708 }, { "epoch": 0.2838665238022348, "grad_norm": 0.6049926693319626, "learning_rate": 8.723147361800948e-06, "loss": 1.4178, "step": 3709 }, { "epoch": 0.283943058319302, "grad_norm": 0.5404274935987547, "learning_rate": 8.722421016896858e-06, "loss": 1.2646, "step": 3710 }, { "epoch": 0.2840195928363692, "grad_norm": 0.5706539299346446, "learning_rate": 8.721694504719083e-06, "loss": 1.5595, "step": 3711 }, { "epoch": 0.2840961273534364, "grad_norm": 0.8634263543799305, "learning_rate": 8.720967825312263e-06, "loss": 1.2747, "step": 3712 }, { "epoch": 0.2841726618705036, "grad_norm": 0.70906741981063, "learning_rate": 8.720240978721042e-06, "loss": 1.2532, "step": 3713 }, { "epoch": 0.2842491963875708, "grad_norm": 0.5369774777017984, "learning_rate": 8.719513964990084e-06, "loss": 1.1913, "step": 3714 }, { "epoch": 0.284325730904638, "grad_norm": 0.5826594546865405, "learning_rate": 8.71878678416406e-06, "loss": 1.2834, "step": 3715 }, { "epoch": 0.2844022654217052, "grad_norm": 0.61789890639959, "learning_rate": 8.718059436287648e-06, "loss": 1.3696, "step": 3716 }, { "epoch": 0.2844787999387724, "grad_norm": 0.5691397417945164, "learning_rate": 8.717331921405539e-06, "loss": 1.3539, "step": 3717 }, { "epoch": 0.2845553344558396, "grad_norm": 0.5242795758462084, "learning_rate": 8.716604239562432e-06, "loss": 1.3288, "step": 3718 }, { "epoch": 0.28463186897290677, "grad_norm": 0.6947717354563508, "learning_rate": 8.71587639080304e-06, "loss": 1.4297, "step": 3719 }, { "epoch": 0.284708403489974, "grad_norm": 0.6198338289637177, "learning_rate": 8.715148375172084e-06, "loss": 1.2113, "step": 3720 }, { "epoch": 0.2847849380070412, "grad_norm": 0.6092200858070954, "learning_rate": 8.714420192714296e-06, "loss": 1.4141, "step": 3721 }, { "epoch": 0.2848614725241084, "grad_norm": 0.6088369108746245, "learning_rate": 8.713691843474416e-06, "loss": 1.5784, "step": 3722 }, { "epoch": 0.28493800704117556, "grad_norm": 0.6092193712951847, "learning_rate": 8.712963327497196e-06, "loss": 1.5671, "step": 3723 }, { "epoch": 0.28501454155824274, "grad_norm": 0.5481628822980937, "learning_rate": 8.712234644827398e-06, "loss": 1.2301, "step": 3724 }, { "epoch": 0.28509107607531, "grad_norm": 0.620007533062021, "learning_rate": 8.711505795509796e-06, "loss": 1.5862, "step": 3725 }, { "epoch": 0.28516761059237716, "grad_norm": 0.9132100922988459, "learning_rate": 8.710776779589174e-06, "loss": 1.7631, "step": 3726 }, { "epoch": 0.28524414510944435, "grad_norm": 0.5860607647174906, "learning_rate": 8.710047597110323e-06, "loss": 1.4103, "step": 3727 }, { "epoch": 0.28532067962651153, "grad_norm": 0.9928428294695825, "learning_rate": 8.709318248118043e-06, "loss": 1.71, "step": 3728 }, { "epoch": 0.28539721414357877, "grad_norm": 0.657651049575163, "learning_rate": 8.708588732657154e-06, "loss": 1.5138, "step": 3729 }, { "epoch": 0.28547374866064595, "grad_norm": 0.6581073630189497, "learning_rate": 8.707859050772474e-06, "loss": 1.1655, "step": 3730 }, { "epoch": 0.28555028317771314, "grad_norm": 0.5039231839137421, "learning_rate": 8.707129202508841e-06, "loss": 0.9629, "step": 3731 }, { "epoch": 0.2856268176947803, "grad_norm": 0.5939935037684403, "learning_rate": 8.706399187911094e-06, "loss": 1.3741, "step": 3732 }, { "epoch": 0.28570335221184756, "grad_norm": 0.5456098318719169, "learning_rate": 8.705669007024092e-06, "loss": 1.2595, "step": 3733 }, { "epoch": 0.28577988672891474, "grad_norm": 0.6031595598370497, "learning_rate": 8.704938659892697e-06, "loss": 1.4424, "step": 3734 }, { "epoch": 0.28585642124598193, "grad_norm": 0.5641242841736288, "learning_rate": 8.704208146561786e-06, "loss": 1.3427, "step": 3735 }, { "epoch": 0.2859329557630491, "grad_norm": 0.5847178337826162, "learning_rate": 8.703477467076242e-06, "loss": 1.2933, "step": 3736 }, { "epoch": 0.28600949028011635, "grad_norm": 0.5051984070563652, "learning_rate": 8.702746621480962e-06, "loss": 1.0443, "step": 3737 }, { "epoch": 0.28608602479718354, "grad_norm": 0.4970363752397225, "learning_rate": 8.702015609820847e-06, "loss": 0.9305, "step": 3738 }, { "epoch": 0.2861625593142507, "grad_norm": 0.6697180123170673, "learning_rate": 8.701284432140816e-06, "loss": 1.7062, "step": 3739 }, { "epoch": 0.2862390938313179, "grad_norm": 0.8684629269154807, "learning_rate": 8.700553088485794e-06, "loss": 1.5866, "step": 3740 }, { "epoch": 0.28631562834838514, "grad_norm": 0.5779801896977085, "learning_rate": 8.699821578900719e-06, "loss": 1.4419, "step": 3741 }, { "epoch": 0.2863921628654523, "grad_norm": 0.9189349141532294, "learning_rate": 8.699089903430533e-06, "loss": 1.9324, "step": 3742 }, { "epoch": 0.2864686973825195, "grad_norm": 0.6212480247774612, "learning_rate": 8.698358062120197e-06, "loss": 1.5261, "step": 3743 }, { "epoch": 0.2865452318995867, "grad_norm": 0.5630404422973094, "learning_rate": 8.697626055014674e-06, "loss": 1.3215, "step": 3744 }, { "epoch": 0.28662176641665393, "grad_norm": 0.5133022012692801, "learning_rate": 8.696893882158942e-06, "loss": 1.1338, "step": 3745 }, { "epoch": 0.2866983009337211, "grad_norm": 0.5972755659209805, "learning_rate": 8.696161543597988e-06, "loss": 1.322, "step": 3746 }, { "epoch": 0.2867748354507883, "grad_norm": 0.5782078806477287, "learning_rate": 8.695429039376807e-06, "loss": 1.2501, "step": 3747 }, { "epoch": 0.2868513699678555, "grad_norm": 0.6076763720343973, "learning_rate": 8.69469636954041e-06, "loss": 1.4992, "step": 3748 }, { "epoch": 0.2869279044849227, "grad_norm": 0.5752393722387898, "learning_rate": 8.693963534133812e-06, "loss": 1.3283, "step": 3749 }, { "epoch": 0.2870044390019899, "grad_norm": 0.5585111947788148, "learning_rate": 8.69323053320204e-06, "loss": 1.3263, "step": 3750 }, { "epoch": 0.2870809735190571, "grad_norm": 0.6062549431800373, "learning_rate": 8.692497366790132e-06, "loss": 1.4636, "step": 3751 }, { "epoch": 0.2871575080361243, "grad_norm": 0.6867390993709146, "learning_rate": 8.691764034943138e-06, "loss": 1.5298, "step": 3752 }, { "epoch": 0.2872340425531915, "grad_norm": 0.5147002020208563, "learning_rate": 8.691030537706116e-06, "loss": 1.1588, "step": 3753 }, { "epoch": 0.2873105770702587, "grad_norm": 0.6004249203536406, "learning_rate": 8.690296875124132e-06, "loss": 1.4864, "step": 3754 }, { "epoch": 0.2873871115873259, "grad_norm": 0.6129270609025497, "learning_rate": 8.689563047242266e-06, "loss": 1.5219, "step": 3755 }, { "epoch": 0.28746364610439307, "grad_norm": 0.7479284582022618, "learning_rate": 8.688829054105607e-06, "loss": 1.7591, "step": 3756 }, { "epoch": 0.2875401806214603, "grad_norm": 0.7381802765753273, "learning_rate": 8.688094895759252e-06, "loss": 1.6509, "step": 3757 }, { "epoch": 0.2876167151385275, "grad_norm": 0.5405443926189567, "learning_rate": 8.687360572248308e-06, "loss": 1.226, "step": 3758 }, { "epoch": 0.2876932496555947, "grad_norm": 0.5707022940316212, "learning_rate": 8.6866260836179e-06, "loss": 1.3882, "step": 3759 }, { "epoch": 0.28776978417266186, "grad_norm": 0.47954941415140423, "learning_rate": 8.685891429913154e-06, "loss": 1.0006, "step": 3760 }, { "epoch": 0.28784631868972904, "grad_norm": 0.6603735273904074, "learning_rate": 8.685156611179207e-06, "loss": 1.8635, "step": 3761 }, { "epoch": 0.2879228532067963, "grad_norm": 0.5305610675102772, "learning_rate": 8.684421627461213e-06, "loss": 1.2095, "step": 3762 }, { "epoch": 0.28799938772386346, "grad_norm": 0.7003193979614027, "learning_rate": 8.683686478804328e-06, "loss": 1.4934, "step": 3763 }, { "epoch": 0.28807592224093065, "grad_norm": 0.5518676015531665, "learning_rate": 8.682951165253723e-06, "loss": 1.3425, "step": 3764 }, { "epoch": 0.28815245675799783, "grad_norm": 0.6984336468108876, "learning_rate": 8.682215686854577e-06, "loss": 1.4898, "step": 3765 }, { "epoch": 0.28822899127506507, "grad_norm": 0.9802181728507325, "learning_rate": 8.681480043652084e-06, "loss": 1.4872, "step": 3766 }, { "epoch": 0.28830552579213226, "grad_norm": 0.7146055699459799, "learning_rate": 8.680744235691439e-06, "loss": 1.5769, "step": 3767 }, { "epoch": 0.28838206030919944, "grad_norm": 0.6110045612152266, "learning_rate": 8.680008263017855e-06, "loss": 1.5929, "step": 3768 }, { "epoch": 0.2884585948262666, "grad_norm": 0.5880023117116432, "learning_rate": 8.67927212567655e-06, "loss": 1.4247, "step": 3769 }, { "epoch": 0.28853512934333386, "grad_norm": 0.5935532358412893, "learning_rate": 8.678535823712759e-06, "loss": 1.3182, "step": 3770 }, { "epoch": 0.28861166386040105, "grad_norm": 0.5839084031904265, "learning_rate": 8.677799357171718e-06, "loss": 1.3094, "step": 3771 }, { "epoch": 0.28868819837746823, "grad_norm": 0.5001293386063321, "learning_rate": 8.677062726098678e-06, "loss": 1.1574, "step": 3772 }, { "epoch": 0.2887647328945354, "grad_norm": 0.6282072143100563, "learning_rate": 8.676325930538904e-06, "loss": 1.5552, "step": 3773 }, { "epoch": 0.28884126741160265, "grad_norm": 0.5664560441945093, "learning_rate": 8.675588970537662e-06, "loss": 1.1949, "step": 3774 }, { "epoch": 0.28891780192866984, "grad_norm": 0.5345463564866261, "learning_rate": 8.674851846140237e-06, "loss": 1.303, "step": 3775 }, { "epoch": 0.288994336445737, "grad_norm": 0.7488873830598296, "learning_rate": 8.674114557391917e-06, "loss": 1.7178, "step": 3776 }, { "epoch": 0.2890708709628042, "grad_norm": 0.6296864538734944, "learning_rate": 8.673377104338004e-06, "loss": 1.7282, "step": 3777 }, { "epoch": 0.28914740547987144, "grad_norm": 0.6693033101502, "learning_rate": 8.672639487023811e-06, "loss": 1.3172, "step": 3778 }, { "epoch": 0.28922393999693863, "grad_norm": 0.5436303446229925, "learning_rate": 8.671901705494657e-06, "loss": 1.2632, "step": 3779 }, { "epoch": 0.2893004745140058, "grad_norm": 0.6321319047926872, "learning_rate": 8.671163759795875e-06, "loss": 1.3915, "step": 3780 }, { "epoch": 0.289377009031073, "grad_norm": 0.668574754715371, "learning_rate": 8.670425649972807e-06, "loss": 1.7638, "step": 3781 }, { "epoch": 0.28945354354814024, "grad_norm": 0.6211705856390566, "learning_rate": 8.669687376070804e-06, "loss": 1.505, "step": 3782 }, { "epoch": 0.2895300780652074, "grad_norm": 0.6334597633761015, "learning_rate": 8.668948938135228e-06, "loss": 1.5819, "step": 3783 }, { "epoch": 0.2896066125822746, "grad_norm": 0.6490923505081908, "learning_rate": 8.668210336211451e-06, "loss": 1.79, "step": 3784 }, { "epoch": 0.2896831470993418, "grad_norm": 0.5111869538639318, "learning_rate": 8.667471570344851e-06, "loss": 1.0435, "step": 3785 }, { "epoch": 0.289759681616409, "grad_norm": 0.5648656415951381, "learning_rate": 8.666732640580828e-06, "loss": 1.4447, "step": 3786 }, { "epoch": 0.2898362161334762, "grad_norm": 0.5690228188407059, "learning_rate": 8.665993546964779e-06, "loss": 1.4963, "step": 3787 }, { "epoch": 0.2899127506505434, "grad_norm": 0.6531940878163828, "learning_rate": 8.665254289542113e-06, "loss": 1.5111, "step": 3788 }, { "epoch": 0.2899892851676106, "grad_norm": 0.6110122472163924, "learning_rate": 8.664514868358258e-06, "loss": 1.3546, "step": 3789 }, { "epoch": 0.2900658196846778, "grad_norm": 0.6540761122833701, "learning_rate": 8.663775283458646e-06, "loss": 1.3712, "step": 3790 }, { "epoch": 0.290142354201745, "grad_norm": 0.632323117607764, "learning_rate": 8.663035534888717e-06, "loss": 1.43, "step": 3791 }, { "epoch": 0.2902188887188122, "grad_norm": 0.6394779766793734, "learning_rate": 8.662295622693921e-06, "loss": 1.7012, "step": 3792 }, { "epoch": 0.29029542323587937, "grad_norm": 0.543609933314997, "learning_rate": 8.661555546919725e-06, "loss": 1.2142, "step": 3793 }, { "epoch": 0.29037195775294655, "grad_norm": 0.5833848022496033, "learning_rate": 8.660815307611598e-06, "loss": 1.2665, "step": 3794 }, { "epoch": 0.2904484922700138, "grad_norm": 0.5780003405897753, "learning_rate": 8.660074904815027e-06, "loss": 1.3659, "step": 3795 }, { "epoch": 0.290525026787081, "grad_norm": 0.5380732814852017, "learning_rate": 8.6593343385755e-06, "loss": 1.1978, "step": 3796 }, { "epoch": 0.29060156130414816, "grad_norm": 0.6868895088005885, "learning_rate": 8.658593608938523e-06, "loss": 1.4584, "step": 3797 }, { "epoch": 0.29067809582121534, "grad_norm": 0.5469326151025119, "learning_rate": 8.657852715949604e-06, "loss": 1.3007, "step": 3798 }, { "epoch": 0.2907546303382826, "grad_norm": 0.5329718150453722, "learning_rate": 8.65711165965427e-06, "loss": 1.2735, "step": 3799 }, { "epoch": 0.29083116485534977, "grad_norm": 0.5513595616391868, "learning_rate": 8.656370440098054e-06, "loss": 1.4093, "step": 3800 }, { "epoch": 0.29090769937241695, "grad_norm": 0.6067632312232921, "learning_rate": 8.655629057326495e-06, "loss": 1.4576, "step": 3801 }, { "epoch": 0.29098423388948413, "grad_norm": 0.7882150915430364, "learning_rate": 8.65488751138515e-06, "loss": 1.8516, "step": 3802 }, { "epoch": 0.2910607684065514, "grad_norm": 0.5024271346242811, "learning_rate": 8.654145802319578e-06, "loss": 1.1322, "step": 3803 }, { "epoch": 0.29113730292361856, "grad_norm": 0.6872607653522823, "learning_rate": 8.653403930175354e-06, "loss": 1.6046, "step": 3804 }, { "epoch": 0.29121383744068574, "grad_norm": 0.5890030919853909, "learning_rate": 8.65266189499806e-06, "loss": 1.2761, "step": 3805 }, { "epoch": 0.2912903719577529, "grad_norm": 0.6261400913436037, "learning_rate": 8.65191969683329e-06, "loss": 1.6092, "step": 3806 }, { "epoch": 0.29136690647482016, "grad_norm": 0.49406300676343706, "learning_rate": 8.651177335726647e-06, "loss": 1.1091, "step": 3807 }, { "epoch": 0.29144344099188735, "grad_norm": 0.648955319152215, "learning_rate": 8.650434811723741e-06, "loss": 1.5437, "step": 3808 }, { "epoch": 0.29151997550895453, "grad_norm": 0.6187330779543058, "learning_rate": 8.649692124870198e-06, "loss": 1.5207, "step": 3809 }, { "epoch": 0.2915965100260217, "grad_norm": 0.8016123309155921, "learning_rate": 8.648949275211649e-06, "loss": 1.579, "step": 3810 }, { "epoch": 0.29167304454308896, "grad_norm": 0.6277411895064672, "learning_rate": 8.648206262793736e-06, "loss": 1.5616, "step": 3811 }, { "epoch": 0.29174957906015614, "grad_norm": 1.0305229342469033, "learning_rate": 8.647463087662117e-06, "loss": 1.6038, "step": 3812 }, { "epoch": 0.2918261135772233, "grad_norm": 0.5244477292859103, "learning_rate": 8.646719749862449e-06, "loss": 1.1136, "step": 3813 }, { "epoch": 0.2919026480942905, "grad_norm": 0.7223927354054808, "learning_rate": 8.64597624944041e-06, "loss": 1.2522, "step": 3814 }, { "epoch": 0.29197918261135775, "grad_norm": 0.6590343795409499, "learning_rate": 8.645232586441678e-06, "loss": 1.6179, "step": 3815 }, { "epoch": 0.29205571712842493, "grad_norm": 0.5854343274401055, "learning_rate": 8.644488760911949e-06, "loss": 1.5027, "step": 3816 }, { "epoch": 0.2921322516454921, "grad_norm": 0.5951717437273784, "learning_rate": 8.643744772896924e-06, "loss": 1.4803, "step": 3817 }, { "epoch": 0.2922087861625593, "grad_norm": 0.667882713575183, "learning_rate": 8.643000622442317e-06, "loss": 1.7545, "step": 3818 }, { "epoch": 0.29228532067962654, "grad_norm": 0.5340698771449409, "learning_rate": 8.642256309593852e-06, "loss": 1.2565, "step": 3819 }, { "epoch": 0.2923618551966937, "grad_norm": 0.5587200050433053, "learning_rate": 8.64151183439726e-06, "loss": 1.105, "step": 3820 }, { "epoch": 0.2924383897137609, "grad_norm": 0.5422029356368192, "learning_rate": 8.640767196898284e-06, "loss": 1.224, "step": 3821 }, { "epoch": 0.2925149242308281, "grad_norm": 0.6026823912804673, "learning_rate": 8.640022397142679e-06, "loss": 1.6248, "step": 3822 }, { "epoch": 0.2925914587478953, "grad_norm": 0.6937649890038309, "learning_rate": 8.639277435176203e-06, "loss": 1.8446, "step": 3823 }, { "epoch": 0.2926679932649625, "grad_norm": 0.6692565742864086, "learning_rate": 8.638532311044631e-06, "loss": 1.5893, "step": 3824 }, { "epoch": 0.2927445277820297, "grad_norm": 0.5598601910818835, "learning_rate": 8.637787024793749e-06, "loss": 1.4326, "step": 3825 }, { "epoch": 0.2928210622990969, "grad_norm": 0.6844576702513679, "learning_rate": 8.637041576469344e-06, "loss": 1.5243, "step": 3826 }, { "epoch": 0.29289759681616406, "grad_norm": 0.6174387708162203, "learning_rate": 8.636295966117223e-06, "loss": 1.4212, "step": 3827 }, { "epoch": 0.2929741313332313, "grad_norm": 0.6660813668966216, "learning_rate": 8.635550193783197e-06, "loss": 1.8009, "step": 3828 }, { "epoch": 0.2930506658502985, "grad_norm": 0.6413334396454486, "learning_rate": 8.634804259513087e-06, "loss": 1.6723, "step": 3829 }, { "epoch": 0.29312720036736567, "grad_norm": 0.6396172922535909, "learning_rate": 8.634058163352727e-06, "loss": 1.5073, "step": 3830 }, { "epoch": 0.29320373488443285, "grad_norm": 0.6659430575301949, "learning_rate": 8.63331190534796e-06, "loss": 1.5581, "step": 3831 }, { "epoch": 0.2932802694015001, "grad_norm": 1.1711085443360483, "learning_rate": 8.632565485544637e-06, "loss": 1.7771, "step": 3832 }, { "epoch": 0.2933568039185673, "grad_norm": 0.6168381688217491, "learning_rate": 8.63181890398862e-06, "loss": 1.503, "step": 3833 }, { "epoch": 0.29343333843563446, "grad_norm": 0.5352127120020854, "learning_rate": 8.631072160725784e-06, "loss": 1.3666, "step": 3834 }, { "epoch": 0.29350987295270164, "grad_norm": 0.5874611818799794, "learning_rate": 8.630325255802006e-06, "loss": 1.4283, "step": 3835 }, { "epoch": 0.2935864074697689, "grad_norm": 0.5181236334132876, "learning_rate": 8.629578189263184e-06, "loss": 1.1428, "step": 3836 }, { "epoch": 0.29366294198683607, "grad_norm": 0.6095947023252191, "learning_rate": 8.628830961155214e-06, "loss": 1.4717, "step": 3837 }, { "epoch": 0.29373947650390325, "grad_norm": 0.6009245727053596, "learning_rate": 8.628083571524014e-06, "loss": 1.3795, "step": 3838 }, { "epoch": 0.29381601102097044, "grad_norm": 0.5682969724696744, "learning_rate": 8.627336020415503e-06, "loss": 1.3614, "step": 3839 }, { "epoch": 0.2938925455380377, "grad_norm": 0.4792030557814177, "learning_rate": 8.626588307875609e-06, "loss": 1.0622, "step": 3840 }, { "epoch": 0.29396908005510486, "grad_norm": 0.6655199253458147, "learning_rate": 8.62584043395028e-06, "loss": 1.5692, "step": 3841 }, { "epoch": 0.29404561457217204, "grad_norm": 0.5823903028693784, "learning_rate": 8.625092398685465e-06, "loss": 1.5003, "step": 3842 }, { "epoch": 0.2941221490892392, "grad_norm": 0.7017171347747112, "learning_rate": 8.624344202127125e-06, "loss": 1.7268, "step": 3843 }, { "epoch": 0.29419868360630647, "grad_norm": 0.6848241727608534, "learning_rate": 8.623595844321232e-06, "loss": 1.5248, "step": 3844 }, { "epoch": 0.29427521812337365, "grad_norm": 0.6488556789498775, "learning_rate": 8.622847325313768e-06, "loss": 1.669, "step": 3845 }, { "epoch": 0.29435175264044083, "grad_norm": 0.7540173406983993, "learning_rate": 8.622098645150723e-06, "loss": 1.6017, "step": 3846 }, { "epoch": 0.294428287157508, "grad_norm": 0.5397828516954186, "learning_rate": 8.621349803878098e-06, "loss": 1.2896, "step": 3847 }, { "epoch": 0.29450482167457526, "grad_norm": 0.6236655733568989, "learning_rate": 8.620600801541905e-06, "loss": 1.5286, "step": 3848 }, { "epoch": 0.29458135619164244, "grad_norm": 0.5952193067288596, "learning_rate": 8.619851638188165e-06, "loss": 1.3085, "step": 3849 }, { "epoch": 0.2946578907087096, "grad_norm": 0.5498375330592687, "learning_rate": 8.619102313862908e-06, "loss": 1.1742, "step": 3850 }, { "epoch": 0.2947344252257768, "grad_norm": 0.6055859004774233, "learning_rate": 8.618352828612177e-06, "loss": 1.468, "step": 3851 }, { "epoch": 0.29481095974284405, "grad_norm": 0.7087829687078339, "learning_rate": 8.617603182482019e-06, "loss": 1.8051, "step": 3852 }, { "epoch": 0.29488749425991123, "grad_norm": 0.6250309906925446, "learning_rate": 8.616853375518496e-06, "loss": 1.3877, "step": 3853 }, { "epoch": 0.2949640287769784, "grad_norm": 1.1124839876016308, "learning_rate": 8.61610340776768e-06, "loss": 1.5309, "step": 3854 }, { "epoch": 0.2950405632940456, "grad_norm": 0.5429511175135999, "learning_rate": 8.615353279275648e-06, "loss": 1.2715, "step": 3855 }, { "epoch": 0.29511709781111284, "grad_norm": 0.5223802332337621, "learning_rate": 8.614602990088495e-06, "loss": 1.2885, "step": 3856 }, { "epoch": 0.29519363232818, "grad_norm": 0.6553034336065423, "learning_rate": 8.613852540252316e-06, "loss": 1.6515, "step": 3857 }, { "epoch": 0.2952701668452472, "grad_norm": 0.61732595318518, "learning_rate": 8.613101929813223e-06, "loss": 1.5893, "step": 3858 }, { "epoch": 0.2953467013623144, "grad_norm": 0.6834780868927792, "learning_rate": 8.612351158817338e-06, "loss": 1.7587, "step": 3859 }, { "epoch": 0.2954232358793816, "grad_norm": 0.7515668456017371, "learning_rate": 8.611600227310786e-06, "loss": 1.4241, "step": 3860 }, { "epoch": 0.2954997703964488, "grad_norm": 0.9099215237989489, "learning_rate": 8.61084913533971e-06, "loss": 1.6753, "step": 3861 }, { "epoch": 0.295576304913516, "grad_norm": 0.5048977243260254, "learning_rate": 8.610097882950257e-06, "loss": 1.0284, "step": 3862 }, { "epoch": 0.2956528394305832, "grad_norm": 0.8698645783272234, "learning_rate": 8.609346470188587e-06, "loss": 1.8474, "step": 3863 }, { "epoch": 0.29572937394765036, "grad_norm": 0.5288729795413105, "learning_rate": 8.608594897100873e-06, "loss": 1.2791, "step": 3864 }, { "epoch": 0.2958059084647176, "grad_norm": 0.5984171447808992, "learning_rate": 8.607843163733287e-06, "loss": 1.4029, "step": 3865 }, { "epoch": 0.2958824429817848, "grad_norm": 0.8416450773753796, "learning_rate": 8.60709127013202e-06, "loss": 1.6261, "step": 3866 }, { "epoch": 0.29595897749885197, "grad_norm": 0.5913670506372063, "learning_rate": 8.606339216343275e-06, "loss": 1.445, "step": 3867 }, { "epoch": 0.29603551201591916, "grad_norm": 0.5566023323064693, "learning_rate": 8.605587002413257e-06, "loss": 1.3555, "step": 3868 }, { "epoch": 0.2961120465329864, "grad_norm": 0.5810175599900945, "learning_rate": 8.60483462838818e-06, "loss": 1.3357, "step": 3869 }, { "epoch": 0.2961885810500536, "grad_norm": 1.0286740085782065, "learning_rate": 8.604082094314281e-06, "loss": 1.3319, "step": 3870 }, { "epoch": 0.29626511556712076, "grad_norm": 0.5570838704350073, "learning_rate": 8.603329400237793e-06, "loss": 1.2281, "step": 3871 }, { "epoch": 0.29634165008418795, "grad_norm": 0.7188568254621768, "learning_rate": 8.602576546204961e-06, "loss": 1.7893, "step": 3872 }, { "epoch": 0.2964181846012552, "grad_norm": 0.5925809497414375, "learning_rate": 8.601823532262047e-06, "loss": 1.4346, "step": 3873 }, { "epoch": 0.29649471911832237, "grad_norm": 0.618082036193073, "learning_rate": 8.601070358455318e-06, "loss": 1.6288, "step": 3874 }, { "epoch": 0.29657125363538955, "grad_norm": 0.6546187762594536, "learning_rate": 8.600317024831052e-06, "loss": 1.7468, "step": 3875 }, { "epoch": 0.29664778815245674, "grad_norm": 0.6921653686813082, "learning_rate": 8.599563531435532e-06, "loss": 1.575, "step": 3876 }, { "epoch": 0.296724322669524, "grad_norm": 0.674124254118077, "learning_rate": 8.59880987831506e-06, "loss": 1.3785, "step": 3877 }, { "epoch": 0.29680085718659116, "grad_norm": 0.5755892229361229, "learning_rate": 8.598056065515938e-06, "loss": 1.5192, "step": 3878 }, { "epoch": 0.29687739170365834, "grad_norm": 0.5957370215619502, "learning_rate": 8.597302093084484e-06, "loss": 1.3526, "step": 3879 }, { "epoch": 0.29695392622072553, "grad_norm": 0.5180876502057801, "learning_rate": 8.596547961067026e-06, "loss": 1.0967, "step": 3880 }, { "epoch": 0.29703046073779277, "grad_norm": 0.5323326291636999, "learning_rate": 8.5957936695099e-06, "loss": 1.1174, "step": 3881 }, { "epoch": 0.29710699525485995, "grad_norm": 0.559659351238107, "learning_rate": 8.59503921845945e-06, "loss": 1.3117, "step": 3882 }, { "epoch": 0.29718352977192714, "grad_norm": 0.530667576552079, "learning_rate": 8.594284607962033e-06, "loss": 1.2026, "step": 3883 }, { "epoch": 0.2972600642889943, "grad_norm": 0.4757475690821142, "learning_rate": 8.593529838064015e-06, "loss": 0.8788, "step": 3884 }, { "epoch": 0.29733659880606156, "grad_norm": 0.5896856440473038, "learning_rate": 8.592774908811769e-06, "loss": 1.4323, "step": 3885 }, { "epoch": 0.29741313332312874, "grad_norm": 0.5832958340628257, "learning_rate": 8.592019820251682e-06, "loss": 1.2747, "step": 3886 }, { "epoch": 0.2974896678401959, "grad_norm": 0.49602195515506914, "learning_rate": 8.591264572430148e-06, "loss": 1.1422, "step": 3887 }, { "epoch": 0.2975662023572631, "grad_norm": 0.6132763632798324, "learning_rate": 8.590509165393574e-06, "loss": 1.5384, "step": 3888 }, { "epoch": 0.29764273687433035, "grad_norm": 0.587339627987338, "learning_rate": 8.589753599188371e-06, "loss": 1.433, "step": 3889 }, { "epoch": 0.29771927139139753, "grad_norm": 0.6868074293285539, "learning_rate": 8.588997873860966e-06, "loss": 1.2111, "step": 3890 }, { "epoch": 0.2977958059084647, "grad_norm": 0.5368947337224408, "learning_rate": 8.588241989457792e-06, "loss": 1.339, "step": 3891 }, { "epoch": 0.2978723404255319, "grad_norm": 1.1982217111222393, "learning_rate": 8.587485946025292e-06, "loss": 1.6334, "step": 3892 }, { "epoch": 0.2979488749425991, "grad_norm": 0.5365077407137788, "learning_rate": 8.586729743609922e-06, "loss": 1.1026, "step": 3893 }, { "epoch": 0.2980254094596663, "grad_norm": 0.516984162213276, "learning_rate": 8.585973382258144e-06, "loss": 1.1342, "step": 3894 }, { "epoch": 0.2981019439767335, "grad_norm": 0.5980184441251883, "learning_rate": 8.585216862016427e-06, "loss": 1.4804, "step": 3895 }, { "epoch": 0.2981784784938007, "grad_norm": 0.7261296561816519, "learning_rate": 8.584460182931262e-06, "loss": 1.314, "step": 3896 }, { "epoch": 0.2982550130108679, "grad_norm": 0.5916337255153182, "learning_rate": 8.583703345049136e-06, "loss": 1.2979, "step": 3897 }, { "epoch": 0.2983315475279351, "grad_norm": 0.898206959677812, "learning_rate": 8.582946348416552e-06, "loss": 1.8169, "step": 3898 }, { "epoch": 0.2984080820450023, "grad_norm": 0.7061096905494493, "learning_rate": 8.582189193080022e-06, "loss": 1.5059, "step": 3899 }, { "epoch": 0.2984846165620695, "grad_norm": 0.6496049931094201, "learning_rate": 8.58143187908607e-06, "loss": 1.4013, "step": 3900 }, { "epoch": 0.29856115107913667, "grad_norm": 0.5693763562072043, "learning_rate": 8.580674406481226e-06, "loss": 1.3105, "step": 3901 }, { "epoch": 0.2986376855962039, "grad_norm": 0.6417578140421832, "learning_rate": 8.579916775312031e-06, "loss": 1.666, "step": 3902 }, { "epoch": 0.2987142201132711, "grad_norm": 0.6256306047242187, "learning_rate": 8.579158985625038e-06, "loss": 1.4961, "step": 3903 }, { "epoch": 0.2987907546303383, "grad_norm": 0.5380048708763958, "learning_rate": 8.578401037466807e-06, "loss": 1.3324, "step": 3904 }, { "epoch": 0.29886728914740546, "grad_norm": 0.5764450723438643, "learning_rate": 8.577642930883907e-06, "loss": 1.5667, "step": 3905 }, { "epoch": 0.2989438236644727, "grad_norm": 0.559448944622855, "learning_rate": 8.576884665922921e-06, "loss": 1.2083, "step": 3906 }, { "epoch": 0.2990203581815399, "grad_norm": 0.6452612859800898, "learning_rate": 8.576126242630437e-06, "loss": 1.7044, "step": 3907 }, { "epoch": 0.29909689269860706, "grad_norm": 0.5499062297511386, "learning_rate": 8.575367661053054e-06, "loss": 1.2494, "step": 3908 }, { "epoch": 0.29917342721567425, "grad_norm": 0.5996712467880405, "learning_rate": 8.574608921237385e-06, "loss": 1.5162, "step": 3909 }, { "epoch": 0.2992499617327415, "grad_norm": 0.9024300781302368, "learning_rate": 8.573850023230047e-06, "loss": 1.3426, "step": 3910 }, { "epoch": 0.29932649624980867, "grad_norm": 0.6661602147503909, "learning_rate": 8.573090967077669e-06, "loss": 1.6463, "step": 3911 }, { "epoch": 0.29940303076687586, "grad_norm": 0.6065981392850711, "learning_rate": 8.57233175282689e-06, "loss": 1.4171, "step": 3912 }, { "epoch": 0.29947956528394304, "grad_norm": 0.509839424372573, "learning_rate": 8.571572380524357e-06, "loss": 1.1317, "step": 3913 }, { "epoch": 0.2995560998010103, "grad_norm": 0.5888023047413016, "learning_rate": 8.570812850216733e-06, "loss": 1.2548, "step": 3914 }, { "epoch": 0.29963263431807746, "grad_norm": 0.7208445274230706, "learning_rate": 8.57005316195068e-06, "loss": 1.8437, "step": 3915 }, { "epoch": 0.29970916883514465, "grad_norm": 0.61386755116628, "learning_rate": 8.569293315772877e-06, "loss": 1.4855, "step": 3916 }, { "epoch": 0.29978570335221183, "grad_norm": 0.9127988947808061, "learning_rate": 8.568533311730012e-06, "loss": 1.4149, "step": 3917 }, { "epoch": 0.29986223786927907, "grad_norm": 0.5776743824570507, "learning_rate": 8.567773149868782e-06, "loss": 1.3519, "step": 3918 }, { "epoch": 0.29993877238634625, "grad_norm": 0.6232585433135712, "learning_rate": 8.567012830235893e-06, "loss": 1.8464, "step": 3919 }, { "epoch": 0.30001530690341344, "grad_norm": 0.7099263120494999, "learning_rate": 8.566252352878064e-06, "loss": 1.6664, "step": 3920 }, { "epoch": 0.3000918414204806, "grad_norm": 0.9385178132882772, "learning_rate": 8.565491717842017e-06, "loss": 1.8633, "step": 3921 }, { "epoch": 0.30016837593754786, "grad_norm": 0.5684554536137181, "learning_rate": 8.56473092517449e-06, "loss": 1.444, "step": 3922 }, { "epoch": 0.30024491045461504, "grad_norm": 0.6076572329748415, "learning_rate": 8.563969974922227e-06, "loss": 1.3663, "step": 3923 }, { "epoch": 0.3003214449716822, "grad_norm": 0.6170170292514775, "learning_rate": 8.563208867131985e-06, "loss": 1.1749, "step": 3924 }, { "epoch": 0.3003979794887494, "grad_norm": 0.59937816821963, "learning_rate": 8.562447601850528e-06, "loss": 1.4965, "step": 3925 }, { "epoch": 0.30047451400581665, "grad_norm": 0.6082266790787796, "learning_rate": 8.56168617912463e-06, "loss": 1.4221, "step": 3926 }, { "epoch": 0.30055104852288383, "grad_norm": 0.6458107782562159, "learning_rate": 8.560924599001075e-06, "loss": 1.5123, "step": 3927 }, { "epoch": 0.300627583039951, "grad_norm": 0.7396415753930646, "learning_rate": 8.560162861526656e-06, "loss": 1.4162, "step": 3928 }, { "epoch": 0.3007041175570182, "grad_norm": 0.7447406971264575, "learning_rate": 8.559400966748177e-06, "loss": 1.5291, "step": 3929 }, { "epoch": 0.3007806520740854, "grad_norm": 0.5565073945282611, "learning_rate": 8.558638914712452e-06, "loss": 1.3431, "step": 3930 }, { "epoch": 0.3008571865911526, "grad_norm": 0.5794447839828555, "learning_rate": 8.557876705466306e-06, "loss": 1.3014, "step": 3931 }, { "epoch": 0.3009337211082198, "grad_norm": 1.8579261620835212, "learning_rate": 8.557114339056565e-06, "loss": 1.3716, "step": 3932 }, { "epoch": 0.301010255625287, "grad_norm": 0.6192528302836845, "learning_rate": 8.556351815530075e-06, "loss": 1.4628, "step": 3933 }, { "epoch": 0.3010867901423542, "grad_norm": 0.6392102491276295, "learning_rate": 8.55558913493369e-06, "loss": 1.5918, "step": 3934 }, { "epoch": 0.3011633246594214, "grad_norm": 0.6035228140736651, "learning_rate": 8.554826297314268e-06, "loss": 1.3753, "step": 3935 }, { "epoch": 0.3012398591764886, "grad_norm": 0.6271632082508507, "learning_rate": 8.554063302718678e-06, "loss": 1.6839, "step": 3936 }, { "epoch": 0.3013163936935558, "grad_norm": 0.5524514728450319, "learning_rate": 8.553300151193806e-06, "loss": 1.3214, "step": 3937 }, { "epoch": 0.30139292821062297, "grad_norm": 0.5640838215439568, "learning_rate": 8.552536842786539e-06, "loss": 1.2624, "step": 3938 }, { "epoch": 0.3014694627276902, "grad_norm": 0.556870277665857, "learning_rate": 8.551773377543777e-06, "loss": 1.2547, "step": 3939 }, { "epoch": 0.3015459972447574, "grad_norm": 0.542279056066123, "learning_rate": 8.551009755512433e-06, "loss": 1.2892, "step": 3940 }, { "epoch": 0.3016225317618246, "grad_norm": 0.5719312934466881, "learning_rate": 8.55024597673942e-06, "loss": 1.0987, "step": 3941 }, { "epoch": 0.30169906627889176, "grad_norm": 0.5882948168554879, "learning_rate": 8.549482041271673e-06, "loss": 1.3158, "step": 3942 }, { "epoch": 0.301775600795959, "grad_norm": 0.509793689989769, "learning_rate": 8.548717949156128e-06, "loss": 1.1677, "step": 3943 }, { "epoch": 0.3018521353130262, "grad_norm": 0.6498106880067513, "learning_rate": 8.547953700439732e-06, "loss": 1.382, "step": 3944 }, { "epoch": 0.30192866983009337, "grad_norm": 0.6022468555622043, "learning_rate": 8.547189295169442e-06, "loss": 1.395, "step": 3945 }, { "epoch": 0.30200520434716055, "grad_norm": 0.5610864261756763, "learning_rate": 8.54642473339223e-06, "loss": 1.097, "step": 3946 }, { "epoch": 0.3020817388642278, "grad_norm": 0.6769764811198609, "learning_rate": 8.545660015155069e-06, "loss": 1.6219, "step": 3947 }, { "epoch": 0.302158273381295, "grad_norm": 0.6529796705852268, "learning_rate": 8.544895140504948e-06, "loss": 1.5528, "step": 3948 }, { "epoch": 0.30223480789836216, "grad_norm": 0.8251835074464103, "learning_rate": 8.54413010948886e-06, "loss": 1.7082, "step": 3949 }, { "epoch": 0.30231134241542934, "grad_norm": 0.6585695237411253, "learning_rate": 8.543364922153813e-06, "loss": 1.4863, "step": 3950 }, { "epoch": 0.3023878769324966, "grad_norm": 0.585819895677379, "learning_rate": 8.542599578546821e-06, "loss": 1.2984, "step": 3951 }, { "epoch": 0.30246441144956376, "grad_norm": 0.7760772297142522, "learning_rate": 8.54183407871491e-06, "loss": 1.4733, "step": 3952 }, { "epoch": 0.30254094596663095, "grad_norm": 0.5426896096698054, "learning_rate": 8.541068422705117e-06, "loss": 1.1948, "step": 3953 }, { "epoch": 0.30261748048369813, "grad_norm": 0.6502912316291899, "learning_rate": 8.540302610564482e-06, "loss": 1.5294, "step": 3954 }, { "epoch": 0.30269401500076537, "grad_norm": 0.6343365644087671, "learning_rate": 8.53953664234006e-06, "loss": 1.3962, "step": 3955 }, { "epoch": 0.30277054951783255, "grad_norm": 1.5555624067519194, "learning_rate": 8.538770518078915e-06, "loss": 1.2138, "step": 3956 }, { "epoch": 0.30284708403489974, "grad_norm": 0.5749922736025674, "learning_rate": 8.538004237828122e-06, "loss": 1.3093, "step": 3957 }, { "epoch": 0.3029236185519669, "grad_norm": 0.6030169805113846, "learning_rate": 8.53723780163476e-06, "loss": 1.5919, "step": 3958 }, { "epoch": 0.30300015306903416, "grad_norm": 0.5729841992624559, "learning_rate": 8.536471209545922e-06, "loss": 1.2498, "step": 3959 }, { "epoch": 0.30307668758610135, "grad_norm": 0.5553029351537802, "learning_rate": 8.53570446160871e-06, "loss": 1.3201, "step": 3960 }, { "epoch": 0.30315322210316853, "grad_norm": 0.7079071987149759, "learning_rate": 8.534937557870237e-06, "loss": 1.4058, "step": 3961 }, { "epoch": 0.3032297566202357, "grad_norm": 0.6535599311652175, "learning_rate": 8.534170498377622e-06, "loss": 1.6387, "step": 3962 }, { "epoch": 0.3033062911373029, "grad_norm": 0.5342264195648487, "learning_rate": 8.533403283177995e-06, "loss": 1.2744, "step": 3963 }, { "epoch": 0.30338282565437014, "grad_norm": 0.6999261519958371, "learning_rate": 8.532635912318497e-06, "loss": 1.4025, "step": 3964 }, { "epoch": 0.3034593601714373, "grad_norm": 0.6397692753575245, "learning_rate": 8.531868385846279e-06, "loss": 1.522, "step": 3965 }, { "epoch": 0.3035358946885045, "grad_norm": 0.5137629494413081, "learning_rate": 8.531100703808497e-06, "loss": 1.038, "step": 3966 }, { "epoch": 0.3036124292055717, "grad_norm": 0.7128165561297463, "learning_rate": 8.530332866252323e-06, "loss": 1.8415, "step": 3967 }, { "epoch": 0.3036889637226389, "grad_norm": 0.5653799346951923, "learning_rate": 8.529564873224931e-06, "loss": 1.2672, "step": 3968 }, { "epoch": 0.3037654982397061, "grad_norm": 0.5348807896955141, "learning_rate": 8.528796724773512e-06, "loss": 1.3157, "step": 3969 }, { "epoch": 0.3038420327567733, "grad_norm": 0.5924437773699348, "learning_rate": 8.528028420945263e-06, "loss": 1.3759, "step": 3970 }, { "epoch": 0.3039185672738405, "grad_norm": 0.5714278889773653, "learning_rate": 8.52725996178739e-06, "loss": 1.4153, "step": 3971 }, { "epoch": 0.3039951017909077, "grad_norm": 0.6228438475956116, "learning_rate": 8.52649134734711e-06, "loss": 1.4345, "step": 3972 }, { "epoch": 0.3040716363079749, "grad_norm": 0.618928821621021, "learning_rate": 8.52572257767165e-06, "loss": 1.4075, "step": 3973 }, { "epoch": 0.3041481708250421, "grad_norm": 0.9976574201507562, "learning_rate": 8.524953652808244e-06, "loss": 1.4962, "step": 3974 }, { "epoch": 0.30422470534210927, "grad_norm": 1.1061336213790092, "learning_rate": 8.524184572804135e-06, "loss": 1.6299, "step": 3975 }, { "epoch": 0.3043012398591765, "grad_norm": 0.59956197359586, "learning_rate": 8.523415337706583e-06, "loss": 1.406, "step": 3976 }, { "epoch": 0.3043777743762437, "grad_norm": 0.5488057428619996, "learning_rate": 8.522645947562848e-06, "loss": 1.2179, "step": 3977 }, { "epoch": 0.3044543088933109, "grad_norm": 0.5859530523663137, "learning_rate": 8.521876402420203e-06, "loss": 1.336, "step": 3978 }, { "epoch": 0.30453084341037806, "grad_norm": 0.6507296031520056, "learning_rate": 8.521106702325935e-06, "loss": 1.3796, "step": 3979 }, { "epoch": 0.3046073779274453, "grad_norm": 0.6616363440249207, "learning_rate": 8.520336847327336e-06, "loss": 1.6784, "step": 3980 }, { "epoch": 0.3046839124445125, "grad_norm": 0.642697353044471, "learning_rate": 8.519566837471703e-06, "loss": 1.4249, "step": 3981 }, { "epoch": 0.30476044696157967, "grad_norm": 0.5952308904576313, "learning_rate": 8.518796672806354e-06, "loss": 1.4465, "step": 3982 }, { "epoch": 0.30483698147864685, "grad_norm": 0.5265433812021052, "learning_rate": 8.518026353378605e-06, "loss": 1.0988, "step": 3983 }, { "epoch": 0.3049135159957141, "grad_norm": 0.6163179009780396, "learning_rate": 8.517255879235792e-06, "loss": 1.5531, "step": 3984 }, { "epoch": 0.3049900505127813, "grad_norm": 0.6182144621522343, "learning_rate": 8.516485250425251e-06, "loss": 1.4954, "step": 3985 }, { "epoch": 0.30506658502984846, "grad_norm": 0.6944510824351718, "learning_rate": 8.515714466994335e-06, "loss": 1.8157, "step": 3986 }, { "epoch": 0.30514311954691564, "grad_norm": 0.5682834332101553, "learning_rate": 8.514943528990402e-06, "loss": 1.4183, "step": 3987 }, { "epoch": 0.3052196540639829, "grad_norm": 0.6290301664187333, "learning_rate": 8.514172436460817e-06, "loss": 1.326, "step": 3988 }, { "epoch": 0.30529618858105007, "grad_norm": 0.5260755673506999, "learning_rate": 8.513401189452964e-06, "loss": 1.2048, "step": 3989 }, { "epoch": 0.30537272309811725, "grad_norm": 0.5501104564436665, "learning_rate": 8.512629788014227e-06, "loss": 1.5473, "step": 3990 }, { "epoch": 0.30544925761518443, "grad_norm": 0.588678842523762, "learning_rate": 8.511858232192008e-06, "loss": 1.4251, "step": 3991 }, { "epoch": 0.3055257921322517, "grad_norm": 0.5568449212209106, "learning_rate": 8.511086522033707e-06, "loss": 1.2517, "step": 3992 }, { "epoch": 0.30560232664931886, "grad_norm": 0.5173582082786771, "learning_rate": 8.510314657586743e-06, "loss": 1.3228, "step": 3993 }, { "epoch": 0.30567886116638604, "grad_norm": 0.5305014236114577, "learning_rate": 8.509542638898543e-06, "loss": 1.1937, "step": 3994 }, { "epoch": 0.3057553956834532, "grad_norm": 0.5879000163511513, "learning_rate": 8.50877046601654e-06, "loss": 1.4864, "step": 3995 }, { "epoch": 0.3058319302005204, "grad_norm": 0.5325494697679537, "learning_rate": 8.507998138988183e-06, "loss": 1.2216, "step": 3996 }, { "epoch": 0.30590846471758765, "grad_norm": 0.5704398953512231, "learning_rate": 8.507225657860919e-06, "loss": 1.3824, "step": 3997 }, { "epoch": 0.30598499923465483, "grad_norm": 0.5054621454128664, "learning_rate": 8.506453022682218e-06, "loss": 1.1119, "step": 3998 }, { "epoch": 0.306061533751722, "grad_norm": 0.5657610620122316, "learning_rate": 8.505680233499549e-06, "loss": 1.2176, "step": 3999 }, { "epoch": 0.3061380682687892, "grad_norm": 0.6602523385770169, "learning_rate": 8.504907290360397e-06, "loss": 1.5691, "step": 4000 }, { "epoch": 0.30621460278585644, "grad_norm": 0.9464104559690034, "learning_rate": 8.504134193312253e-06, "loss": 1.5153, "step": 4001 }, { "epoch": 0.3062911373029236, "grad_norm": 0.6311477212567572, "learning_rate": 8.503360942402616e-06, "loss": 1.644, "step": 4002 }, { "epoch": 0.3063676718199908, "grad_norm": 0.7520063961383615, "learning_rate": 8.502587537679002e-06, "loss": 1.6745, "step": 4003 }, { "epoch": 0.306444206337058, "grad_norm": 0.6668221385450491, "learning_rate": 8.501813979188926e-06, "loss": 1.1057, "step": 4004 }, { "epoch": 0.30652074085412523, "grad_norm": 0.6419135767260228, "learning_rate": 8.501040266979921e-06, "loss": 1.6632, "step": 4005 }, { "epoch": 0.3065972753711924, "grad_norm": 0.7698228867447293, "learning_rate": 8.500266401099526e-06, "loss": 1.4246, "step": 4006 }, { "epoch": 0.3066738098882596, "grad_norm": 0.6422989648029283, "learning_rate": 8.499492381595289e-06, "loss": 1.6658, "step": 4007 }, { "epoch": 0.3067503444053268, "grad_norm": 0.5677376303605552, "learning_rate": 8.498718208514766e-06, "loss": 1.3667, "step": 4008 }, { "epoch": 0.306826878922394, "grad_norm": 0.6122787569582879, "learning_rate": 8.497943881905529e-06, "loss": 1.5122, "step": 4009 }, { "epoch": 0.3069034134394612, "grad_norm": 0.6202341615142088, "learning_rate": 8.497169401815151e-06, "loss": 1.5055, "step": 4010 }, { "epoch": 0.3069799479565284, "grad_norm": 0.6411513774535579, "learning_rate": 8.49639476829122e-06, "loss": 1.7144, "step": 4011 }, { "epoch": 0.30705648247359557, "grad_norm": 0.6355276864536031, "learning_rate": 8.495619981381332e-06, "loss": 1.5575, "step": 4012 }, { "epoch": 0.3071330169906628, "grad_norm": 0.7446985998638967, "learning_rate": 8.49484504113309e-06, "loss": 1.8188, "step": 4013 }, { "epoch": 0.30720955150773, "grad_norm": 0.8730413675637649, "learning_rate": 8.494069947594114e-06, "loss": 1.6066, "step": 4014 }, { "epoch": 0.3072860860247972, "grad_norm": 0.6228058892961797, "learning_rate": 8.493294700812022e-06, "loss": 1.4152, "step": 4015 }, { "epoch": 0.30736262054186436, "grad_norm": 0.4995902696661319, "learning_rate": 8.492519300834452e-06, "loss": 1.1299, "step": 4016 }, { "epoch": 0.3074391550589316, "grad_norm": 0.7729575002084843, "learning_rate": 8.491743747709042e-06, "loss": 1.9919, "step": 4017 }, { "epoch": 0.3075156895759988, "grad_norm": 0.5406848997779622, "learning_rate": 8.49096804148345e-06, "loss": 1.2143, "step": 4018 }, { "epoch": 0.30759222409306597, "grad_norm": 0.6010306957586127, "learning_rate": 8.490192182205332e-06, "loss": 1.3923, "step": 4019 }, { "epoch": 0.30766875861013315, "grad_norm": 0.5486904006072041, "learning_rate": 8.489416169922366e-06, "loss": 1.352, "step": 4020 }, { "epoch": 0.3077452931272004, "grad_norm": 0.5941239447635012, "learning_rate": 8.488640004682225e-06, "loss": 1.3762, "step": 4021 }, { "epoch": 0.3078218276442676, "grad_norm": 0.9789476602722632, "learning_rate": 8.487863686532605e-06, "loss": 1.6513, "step": 4022 }, { "epoch": 0.30789836216133476, "grad_norm": 0.6369594608593958, "learning_rate": 8.487087215521203e-06, "loss": 1.4154, "step": 4023 }, { "epoch": 0.30797489667840194, "grad_norm": 0.5450396447198237, "learning_rate": 8.486310591695725e-06, "loss": 1.3431, "step": 4024 }, { "epoch": 0.3080514311954692, "grad_norm": 0.48441854978215126, "learning_rate": 8.485533815103892e-06, "loss": 0.8822, "step": 4025 }, { "epoch": 0.30812796571253637, "grad_norm": 0.6946992820467667, "learning_rate": 8.484756885793433e-06, "loss": 1.5312, "step": 4026 }, { "epoch": 0.30820450022960355, "grad_norm": 0.8094137183714607, "learning_rate": 8.483979803812083e-06, "loss": 1.553, "step": 4027 }, { "epoch": 0.30828103474667073, "grad_norm": 0.6143069406846207, "learning_rate": 8.483202569207586e-06, "loss": 1.5342, "step": 4028 }, { "epoch": 0.3083575692637379, "grad_norm": 0.5494442127207285, "learning_rate": 8.482425182027701e-06, "loss": 1.2774, "step": 4029 }, { "epoch": 0.30843410378080516, "grad_norm": 0.5195612406839936, "learning_rate": 8.48164764232019e-06, "loss": 1.1963, "step": 4030 }, { "epoch": 0.30851063829787234, "grad_norm": 0.5903981204924249, "learning_rate": 8.480869950132833e-06, "loss": 1.4735, "step": 4031 }, { "epoch": 0.3085871728149395, "grad_norm": 0.9075497110183214, "learning_rate": 8.480092105513407e-06, "loss": 1.6703, "step": 4032 }, { "epoch": 0.3086637073320067, "grad_norm": 0.5832798299088336, "learning_rate": 8.479314108509708e-06, "loss": 1.4198, "step": 4033 }, { "epoch": 0.30874024184907395, "grad_norm": 0.5836374767797575, "learning_rate": 8.478535959169541e-06, "loss": 1.3419, "step": 4034 }, { "epoch": 0.30881677636614113, "grad_norm": 2.4038862154382032, "learning_rate": 8.477757657540712e-06, "loss": 1.6428, "step": 4035 }, { "epoch": 0.3088933108832083, "grad_norm": 0.5448467588038062, "learning_rate": 8.476979203671047e-06, "loss": 1.2999, "step": 4036 }, { "epoch": 0.3089698454002755, "grad_norm": 0.6402485786711514, "learning_rate": 8.476200597608375e-06, "loss": 1.4593, "step": 4037 }, { "epoch": 0.30904637991734274, "grad_norm": 0.5585772365422629, "learning_rate": 8.475421839400536e-06, "loss": 1.1766, "step": 4038 }, { "epoch": 0.3091229144344099, "grad_norm": 0.6244904666390959, "learning_rate": 8.474642929095378e-06, "loss": 1.5385, "step": 4039 }, { "epoch": 0.3091994489514771, "grad_norm": 0.5625678594219304, "learning_rate": 8.473863866740762e-06, "loss": 1.491, "step": 4040 }, { "epoch": 0.3092759834685443, "grad_norm": 0.5294250172389056, "learning_rate": 8.473084652384554e-06, "loss": 1.1888, "step": 4041 }, { "epoch": 0.30935251798561153, "grad_norm": 0.5903603744813471, "learning_rate": 8.472305286074631e-06, "loss": 1.4498, "step": 4042 }, { "epoch": 0.3094290525026787, "grad_norm": 0.6844472785323329, "learning_rate": 8.471525767858882e-06, "loss": 1.5908, "step": 4043 }, { "epoch": 0.3095055870197459, "grad_norm": 0.6569774016375974, "learning_rate": 8.470746097785201e-06, "loss": 1.5739, "step": 4044 }, { "epoch": 0.3095821215368131, "grad_norm": 0.5546316422689344, "learning_rate": 8.469966275901493e-06, "loss": 1.4092, "step": 4045 }, { "epoch": 0.3096586560538803, "grad_norm": 0.7889204820701504, "learning_rate": 8.469186302255672e-06, "loss": 1.7806, "step": 4046 }, { "epoch": 0.3097351905709475, "grad_norm": 0.5649552354513144, "learning_rate": 8.468406176895667e-06, "loss": 1.3783, "step": 4047 }, { "epoch": 0.3098117250880147, "grad_norm": 0.7108976908123161, "learning_rate": 8.467625899869406e-06, "loss": 1.4152, "step": 4048 }, { "epoch": 0.3098882596050819, "grad_norm": 0.5297107508465495, "learning_rate": 8.46684547122483e-06, "loss": 1.0776, "step": 4049 }, { "epoch": 0.3099647941221491, "grad_norm": 0.5791366372899523, "learning_rate": 8.466064891009895e-06, "loss": 1.3997, "step": 4050 }, { "epoch": 0.3100413286392163, "grad_norm": 0.5382822846192351, "learning_rate": 8.465284159272562e-06, "loss": 1.3226, "step": 4051 }, { "epoch": 0.3101178631562835, "grad_norm": 0.6387395740555446, "learning_rate": 8.4645032760608e-06, "loss": 1.2895, "step": 4052 }, { "epoch": 0.31019439767335066, "grad_norm": 0.6701960314057185, "learning_rate": 8.46372224142259e-06, "loss": 1.5609, "step": 4053 }, { "epoch": 0.3102709321904179, "grad_norm": 0.6504592007579273, "learning_rate": 8.46294105540592e-06, "loss": 1.5071, "step": 4054 }, { "epoch": 0.3103474667074851, "grad_norm": 0.6230418988165015, "learning_rate": 8.462159718058787e-06, "loss": 1.3455, "step": 4055 }, { "epoch": 0.31042400122455227, "grad_norm": 0.6803101527641047, "learning_rate": 8.461378229429204e-06, "loss": 1.341, "step": 4056 }, { "epoch": 0.31050053574161945, "grad_norm": 0.4946668025963241, "learning_rate": 8.46059658956518e-06, "loss": 1.031, "step": 4057 }, { "epoch": 0.3105770702586867, "grad_norm": 0.6764968621510753, "learning_rate": 8.459814798514749e-06, "loss": 1.3837, "step": 4058 }, { "epoch": 0.3106536047757539, "grad_norm": 0.7398546943745138, "learning_rate": 8.459032856325943e-06, "loss": 1.8115, "step": 4059 }, { "epoch": 0.31073013929282106, "grad_norm": 0.5792194215015601, "learning_rate": 8.458250763046806e-06, "loss": 1.3523, "step": 4060 }, { "epoch": 0.31080667380988825, "grad_norm": 1.127136304831437, "learning_rate": 8.457468518725395e-06, "loss": 1.865, "step": 4061 }, { "epoch": 0.31088320832695543, "grad_norm": 0.6248397631142484, "learning_rate": 8.45668612340977e-06, "loss": 1.4294, "step": 4062 }, { "epoch": 0.31095974284402267, "grad_norm": 0.5612652901077894, "learning_rate": 8.455903577148006e-06, "loss": 1.3745, "step": 4063 }, { "epoch": 0.31103627736108985, "grad_norm": 0.5310010873572468, "learning_rate": 8.455120879988185e-06, "loss": 1.13, "step": 4064 }, { "epoch": 0.31111281187815704, "grad_norm": 0.9110410573890367, "learning_rate": 8.454338031978398e-06, "loss": 1.543, "step": 4065 }, { "epoch": 0.3111893463952242, "grad_norm": 0.5355419129685622, "learning_rate": 8.453555033166744e-06, "loss": 1.3049, "step": 4066 }, { "epoch": 0.31126588091229146, "grad_norm": 0.5649576907109846, "learning_rate": 8.452771883601335e-06, "loss": 1.1915, "step": 4067 }, { "epoch": 0.31134241542935864, "grad_norm": 0.6402478669320976, "learning_rate": 8.451988583330288e-06, "loss": 1.4124, "step": 4068 }, { "epoch": 0.3114189499464258, "grad_norm": 0.5402361453202013, "learning_rate": 8.451205132401734e-06, "loss": 1.4536, "step": 4069 }, { "epoch": 0.311495484463493, "grad_norm": 0.6078975511273713, "learning_rate": 8.450421530863809e-06, "loss": 1.5579, "step": 4070 }, { "epoch": 0.31157201898056025, "grad_norm": 0.5337719623011968, "learning_rate": 8.449637778764657e-06, "loss": 1.2816, "step": 4071 }, { "epoch": 0.31164855349762743, "grad_norm": 0.5281344011518408, "learning_rate": 8.44885387615244e-06, "loss": 1.2277, "step": 4072 }, { "epoch": 0.3117250880146946, "grad_norm": 0.6325460384622215, "learning_rate": 8.448069823075318e-06, "loss": 1.4862, "step": 4073 }, { "epoch": 0.3118016225317618, "grad_norm": 0.5637421168499833, "learning_rate": 8.447285619581469e-06, "loss": 1.2822, "step": 4074 }, { "epoch": 0.31187815704882904, "grad_norm": 0.5949843453546655, "learning_rate": 8.446501265719076e-06, "loss": 1.2775, "step": 4075 }, { "epoch": 0.3119546915658962, "grad_norm": 0.6811616380589322, "learning_rate": 8.445716761536332e-06, "loss": 1.6664, "step": 4076 }, { "epoch": 0.3120312260829634, "grad_norm": 0.520329579468152, "learning_rate": 8.444932107081436e-06, "loss": 1.087, "step": 4077 }, { "epoch": 0.3121077606000306, "grad_norm": 0.6620430237428359, "learning_rate": 8.444147302402605e-06, "loss": 1.6008, "step": 4078 }, { "epoch": 0.31218429511709783, "grad_norm": 0.6436808744237438, "learning_rate": 8.443362347548056e-06, "loss": 1.3804, "step": 4079 }, { "epoch": 0.312260829634165, "grad_norm": 0.632775364025157, "learning_rate": 8.44257724256602e-06, "loss": 1.624, "step": 4080 }, { "epoch": 0.3123373641512322, "grad_norm": 0.5344801042979377, "learning_rate": 8.441791987504737e-06, "loss": 1.1046, "step": 4081 }, { "epoch": 0.3124138986682994, "grad_norm": 1.1000960625594414, "learning_rate": 8.441006582412454e-06, "loss": 1.7185, "step": 4082 }, { "epoch": 0.3124904331853666, "grad_norm": 0.6056548192671045, "learning_rate": 8.44022102733743e-06, "loss": 1.4947, "step": 4083 }, { "epoch": 0.3125669677024338, "grad_norm": 0.6167556023093361, "learning_rate": 8.43943532232793e-06, "loss": 1.5967, "step": 4084 }, { "epoch": 0.312643502219501, "grad_norm": 0.6391205520885863, "learning_rate": 8.438649467432232e-06, "loss": 1.4348, "step": 4085 }, { "epoch": 0.3127200367365682, "grad_norm": 0.5340583064462178, "learning_rate": 8.437863462698622e-06, "loss": 1.0278, "step": 4086 }, { "epoch": 0.3127965712536354, "grad_norm": 0.5758278018405222, "learning_rate": 8.43707730817539e-06, "loss": 1.4204, "step": 4087 }, { "epoch": 0.3128731057707026, "grad_norm": 0.6616533418813678, "learning_rate": 8.436291003910846e-06, "loss": 1.5702, "step": 4088 }, { "epoch": 0.3129496402877698, "grad_norm": 0.5775654507059858, "learning_rate": 8.435504549953297e-06, "loss": 1.3164, "step": 4089 }, { "epoch": 0.31302617480483697, "grad_norm": 0.6233271292672801, "learning_rate": 8.43471794635107e-06, "loss": 1.4593, "step": 4090 }, { "epoch": 0.3131027093219042, "grad_norm": 0.6764529945856361, "learning_rate": 8.433931193152491e-06, "loss": 1.2646, "step": 4091 }, { "epoch": 0.3131792438389714, "grad_norm": 0.7736346456992044, "learning_rate": 8.433144290405906e-06, "loss": 1.4957, "step": 4092 }, { "epoch": 0.31325577835603857, "grad_norm": 0.6458226413295997, "learning_rate": 8.43235723815966e-06, "loss": 1.7222, "step": 4093 }, { "epoch": 0.31333231287310576, "grad_norm": 0.517660941451792, "learning_rate": 8.431570036462117e-06, "loss": 1.1238, "step": 4094 }, { "epoch": 0.313408847390173, "grad_norm": 0.6428150157211162, "learning_rate": 8.430782685361639e-06, "loss": 1.6055, "step": 4095 }, { "epoch": 0.3134853819072402, "grad_norm": 0.6321474739339429, "learning_rate": 8.429995184906608e-06, "loss": 1.4302, "step": 4096 }, { "epoch": 0.31356191642430736, "grad_norm": 0.5335541870842938, "learning_rate": 8.429207535145407e-06, "loss": 1.3089, "step": 4097 }, { "epoch": 0.31363845094137455, "grad_norm": 0.6078811616745373, "learning_rate": 8.428419736126435e-06, "loss": 1.5337, "step": 4098 }, { "epoch": 0.31371498545844173, "grad_norm": 0.5089473702255799, "learning_rate": 8.427631787898093e-06, "loss": 1.1623, "step": 4099 }, { "epoch": 0.31379151997550897, "grad_norm": 0.5322333092513075, "learning_rate": 8.426843690508797e-06, "loss": 1.1508, "step": 4100 }, { "epoch": 0.31386805449257615, "grad_norm": 0.6459916695788396, "learning_rate": 8.42605544400697e-06, "loss": 1.6038, "step": 4101 }, { "epoch": 0.31394458900964334, "grad_norm": 0.7104195034301329, "learning_rate": 8.425267048441045e-06, "loss": 1.551, "step": 4102 }, { "epoch": 0.3140211235267105, "grad_norm": 0.583445847080793, "learning_rate": 8.42447850385946e-06, "loss": 1.359, "step": 4103 }, { "epoch": 0.31409765804377776, "grad_norm": 0.5982552440203048, "learning_rate": 8.423689810310671e-06, "loss": 1.4499, "step": 4104 }, { "epoch": 0.31417419256084494, "grad_norm": 0.7907200508988204, "learning_rate": 8.422900967843132e-06, "loss": 1.3439, "step": 4105 }, { "epoch": 0.31425072707791213, "grad_norm": 0.4933427647012354, "learning_rate": 8.422111976505317e-06, "loss": 0.9982, "step": 4106 }, { "epoch": 0.3143272615949793, "grad_norm": 0.5863380050731712, "learning_rate": 8.421322836345699e-06, "loss": 1.4487, "step": 4107 }, { "epoch": 0.31440379611204655, "grad_norm": 0.6633162789362437, "learning_rate": 8.42053354741277e-06, "loss": 1.52, "step": 4108 }, { "epoch": 0.31448033062911374, "grad_norm": 0.6163368043671288, "learning_rate": 8.41974410975502e-06, "loss": 1.5427, "step": 4109 }, { "epoch": 0.3145568651461809, "grad_norm": 1.2130873976029837, "learning_rate": 8.418954523420961e-06, "loss": 1.7676, "step": 4110 }, { "epoch": 0.3146333996632481, "grad_norm": 0.596472820330789, "learning_rate": 8.418164788459104e-06, "loss": 1.4104, "step": 4111 }, { "epoch": 0.31470993418031534, "grad_norm": 0.5111464143176867, "learning_rate": 8.417374904917975e-06, "loss": 1.2249, "step": 4112 }, { "epoch": 0.3147864686973825, "grad_norm": 0.609411324156267, "learning_rate": 8.416584872846103e-06, "loss": 1.448, "step": 4113 }, { "epoch": 0.3148630032144497, "grad_norm": 0.5580913108167275, "learning_rate": 8.415794692292034e-06, "loss": 1.3155, "step": 4114 }, { "epoch": 0.3149395377315169, "grad_norm": 0.5326060610334353, "learning_rate": 8.415004363304316e-06, "loss": 1.3273, "step": 4115 }, { "epoch": 0.31501607224858413, "grad_norm": 0.6349984580045207, "learning_rate": 8.414213885931511e-06, "loss": 1.5419, "step": 4116 }, { "epoch": 0.3150926067656513, "grad_norm": 0.5910992287863106, "learning_rate": 8.413423260222188e-06, "loss": 1.5171, "step": 4117 }, { "epoch": 0.3151691412827185, "grad_norm": 0.6982099379830822, "learning_rate": 8.412632486224924e-06, "loss": 1.4137, "step": 4118 }, { "epoch": 0.3152456757997857, "grad_norm": 0.7379179566864272, "learning_rate": 8.411841563988308e-06, "loss": 1.2192, "step": 4119 }, { "epoch": 0.3153222103168529, "grad_norm": 0.5441493568792587, "learning_rate": 8.411050493560938e-06, "loss": 1.3128, "step": 4120 }, { "epoch": 0.3153987448339201, "grad_norm": 0.5987004580655471, "learning_rate": 8.410259274991416e-06, "loss": 1.3231, "step": 4121 }, { "epoch": 0.3154752793509873, "grad_norm": 0.5890431083225508, "learning_rate": 8.40946790832836e-06, "loss": 1.3685, "step": 4122 }, { "epoch": 0.3155518138680545, "grad_norm": 0.5116990260781643, "learning_rate": 8.408676393620392e-06, "loss": 1.2504, "step": 4123 }, { "epoch": 0.3156283483851217, "grad_norm": 0.576445486129122, "learning_rate": 8.407884730916146e-06, "loss": 1.4473, "step": 4124 }, { "epoch": 0.3157048829021889, "grad_norm": 0.6497525662471632, "learning_rate": 8.407092920264264e-06, "loss": 1.4467, "step": 4125 }, { "epoch": 0.3157814174192561, "grad_norm": 0.5946575100693181, "learning_rate": 8.406300961713396e-06, "loss": 1.6042, "step": 4126 }, { "epoch": 0.31585795193632327, "grad_norm": 0.6022093929844893, "learning_rate": 8.405508855312206e-06, "loss": 1.3668, "step": 4127 }, { "epoch": 0.3159344864533905, "grad_norm": 0.5500082170362023, "learning_rate": 8.40471660110936e-06, "loss": 1.1891, "step": 4128 }, { "epoch": 0.3160110209704577, "grad_norm": 0.5760214580179271, "learning_rate": 8.403924199153536e-06, "loss": 1.4016, "step": 4129 }, { "epoch": 0.3160875554875249, "grad_norm": 0.7189881146375888, "learning_rate": 8.403131649493423e-06, "loss": 1.2865, "step": 4130 }, { "epoch": 0.31616409000459206, "grad_norm": 0.5011273831638098, "learning_rate": 8.402338952177718e-06, "loss": 1.0808, "step": 4131 }, { "epoch": 0.31624062452165924, "grad_norm": 0.6041628134008621, "learning_rate": 8.401546107255127e-06, "loss": 1.472, "step": 4132 }, { "epoch": 0.3163171590387265, "grad_norm": 0.6736909962100736, "learning_rate": 8.400753114774363e-06, "loss": 1.4727, "step": 4133 }, { "epoch": 0.31639369355579366, "grad_norm": 0.6741234240770831, "learning_rate": 8.39995997478415e-06, "loss": 1.5448, "step": 4134 }, { "epoch": 0.31647022807286085, "grad_norm": 0.5993634718989282, "learning_rate": 8.399166687333223e-06, "loss": 1.2715, "step": 4135 }, { "epoch": 0.31654676258992803, "grad_norm": 0.6530064926919424, "learning_rate": 8.398373252470322e-06, "loss": 1.4941, "step": 4136 }, { "epoch": 0.31662329710699527, "grad_norm": 0.8121997614984177, "learning_rate": 8.3975796702442e-06, "loss": 1.6311, "step": 4137 }, { "epoch": 0.31669983162406246, "grad_norm": 0.6043802003686882, "learning_rate": 8.396785940703613e-06, "loss": 1.523, "step": 4138 }, { "epoch": 0.31677636614112964, "grad_norm": 0.5232701514812639, "learning_rate": 8.395992063897333e-06, "loss": 1.3426, "step": 4139 }, { "epoch": 0.3168529006581968, "grad_norm": 0.5154279481353621, "learning_rate": 8.39519803987414e-06, "loss": 1.2727, "step": 4140 }, { "epoch": 0.31692943517526406, "grad_norm": 0.5394234544056474, "learning_rate": 8.394403868682818e-06, "loss": 1.2156, "step": 4141 }, { "epoch": 0.31700596969233125, "grad_norm": 0.5497390342743297, "learning_rate": 8.393609550372165e-06, "loss": 1.3613, "step": 4142 }, { "epoch": 0.31708250420939843, "grad_norm": 0.6924012887251683, "learning_rate": 8.392815084990986e-06, "loss": 1.3703, "step": 4143 }, { "epoch": 0.3171590387264656, "grad_norm": 0.6140053867425855, "learning_rate": 8.392020472588094e-06, "loss": 1.5084, "step": 4144 }, { "epoch": 0.31723557324353285, "grad_norm": 0.5321711405627776, "learning_rate": 8.391225713212313e-06, "loss": 1.3396, "step": 4145 }, { "epoch": 0.31731210776060004, "grad_norm": 0.7031063868054438, "learning_rate": 8.390430806912477e-06, "loss": 1.5701, "step": 4146 }, { "epoch": 0.3173886422776672, "grad_norm": 0.6292095789136846, "learning_rate": 8.389635753737426e-06, "loss": 1.3975, "step": 4147 }, { "epoch": 0.3174651767947344, "grad_norm": 0.6350373180829464, "learning_rate": 8.38884055373601e-06, "loss": 1.3356, "step": 4148 }, { "epoch": 0.31754171131180164, "grad_norm": 0.6966360258919293, "learning_rate": 8.388045206957091e-06, "loss": 1.908, "step": 4149 }, { "epoch": 0.31761824582886883, "grad_norm": 0.5306488578645346, "learning_rate": 8.387249713449532e-06, "loss": 1.1668, "step": 4150 }, { "epoch": 0.317694780345936, "grad_norm": 0.5986812868602587, "learning_rate": 8.386454073262217e-06, "loss": 1.3799, "step": 4151 }, { "epoch": 0.3177713148630032, "grad_norm": 0.5443859801285668, "learning_rate": 8.385658286444029e-06, "loss": 1.1515, "step": 4152 }, { "epoch": 0.31784784938007044, "grad_norm": 0.6513161669258196, "learning_rate": 8.384862353043864e-06, "loss": 1.7319, "step": 4153 }, { "epoch": 0.3179243838971376, "grad_norm": 0.5408672748616408, "learning_rate": 8.384066273110624e-06, "loss": 1.2911, "step": 4154 }, { "epoch": 0.3180009184142048, "grad_norm": 0.5702520338190038, "learning_rate": 8.38327004669323e-06, "loss": 1.4445, "step": 4155 }, { "epoch": 0.318077452931272, "grad_norm": 0.9551721757842563, "learning_rate": 8.382473673840598e-06, "loss": 1.6, "step": 4156 }, { "epoch": 0.3181539874483392, "grad_norm": 0.5218678502597109, "learning_rate": 8.38167715460166e-06, "loss": 1.1509, "step": 4157 }, { "epoch": 0.3182305219654064, "grad_norm": 0.5938746859667844, "learning_rate": 8.380880489025356e-06, "loss": 1.412, "step": 4158 }, { "epoch": 0.3183070564824736, "grad_norm": 0.6584102416854457, "learning_rate": 8.38008367716064e-06, "loss": 1.3171, "step": 4159 }, { "epoch": 0.3183835909995408, "grad_norm": 0.5782178451142439, "learning_rate": 8.379286719056468e-06, "loss": 1.4494, "step": 4160 }, { "epoch": 0.318460125516608, "grad_norm": 0.5477023070815358, "learning_rate": 8.378489614761808e-06, "loss": 1.3098, "step": 4161 }, { "epoch": 0.3185366600336752, "grad_norm": 0.7005638165573903, "learning_rate": 8.377692364325634e-06, "loss": 1.2838, "step": 4162 }, { "epoch": 0.3186131945507424, "grad_norm": 0.5247287917458294, "learning_rate": 8.376894967796933e-06, "loss": 1.2317, "step": 4163 }, { "epoch": 0.31868972906780957, "grad_norm": 0.5917039841336291, "learning_rate": 8.376097425224701e-06, "loss": 1.2972, "step": 4164 }, { "epoch": 0.31876626358487675, "grad_norm": 0.7213001150851766, "learning_rate": 8.375299736657939e-06, "loss": 1.7148, "step": 4165 }, { "epoch": 0.318842798101944, "grad_norm": 0.5511203633760505, "learning_rate": 8.374501902145661e-06, "loss": 1.2429, "step": 4166 }, { "epoch": 0.3189193326190112, "grad_norm": 0.9603373035896724, "learning_rate": 8.373703921736888e-06, "loss": 1.8241, "step": 4167 }, { "epoch": 0.31899586713607836, "grad_norm": 3.9457351220731915, "learning_rate": 8.372905795480648e-06, "loss": 1.5272, "step": 4168 }, { "epoch": 0.31907240165314554, "grad_norm": 0.5836148433000322, "learning_rate": 8.372107523425985e-06, "loss": 1.3654, "step": 4169 }, { "epoch": 0.3191489361702128, "grad_norm": 0.6353592549107294, "learning_rate": 8.371309105621943e-06, "loss": 1.532, "step": 4170 }, { "epoch": 0.31922547068727997, "grad_norm": 0.6439317585983046, "learning_rate": 8.37051054211758e-06, "loss": 1.6207, "step": 4171 }, { "epoch": 0.31930200520434715, "grad_norm": 0.5668987819342355, "learning_rate": 8.369711832961964e-06, "loss": 1.3129, "step": 4172 }, { "epoch": 0.31937853972141433, "grad_norm": 1.0141680892825298, "learning_rate": 8.36891297820417e-06, "loss": 1.6079, "step": 4173 }, { "epoch": 0.3194550742384816, "grad_norm": 0.6174298294030964, "learning_rate": 8.368113977893277e-06, "loss": 1.6868, "step": 4174 }, { "epoch": 0.31953160875554876, "grad_norm": 0.49817112690776044, "learning_rate": 8.367314832078384e-06, "loss": 1.2037, "step": 4175 }, { "epoch": 0.31960814327261594, "grad_norm": 0.5294618675929222, "learning_rate": 8.366515540808592e-06, "loss": 1.1593, "step": 4176 }, { "epoch": 0.3196846777896831, "grad_norm": 0.6867503060258577, "learning_rate": 8.36571610413301e-06, "loss": 1.5787, "step": 4177 }, { "epoch": 0.31976121230675036, "grad_norm": 0.5412783165007443, "learning_rate": 8.364916522100759e-06, "loss": 1.2971, "step": 4178 }, { "epoch": 0.31983774682381755, "grad_norm": 0.4739310202112224, "learning_rate": 8.364116794760966e-06, "loss": 0.8993, "step": 4179 }, { "epoch": 0.31991428134088473, "grad_norm": 0.6674340230933714, "learning_rate": 8.36331692216277e-06, "loss": 1.4141, "step": 4180 }, { "epoch": 0.3199908158579519, "grad_norm": 0.6962166799657926, "learning_rate": 8.362516904355316e-06, "loss": 1.4606, "step": 4181 }, { "epoch": 0.32006735037501916, "grad_norm": 0.576704382830591, "learning_rate": 8.361716741387764e-06, "loss": 1.3358, "step": 4182 }, { "epoch": 0.32014388489208634, "grad_norm": 0.6046240060312408, "learning_rate": 8.360916433309273e-06, "loss": 1.2963, "step": 4183 }, { "epoch": 0.3202204194091535, "grad_norm": 0.5659967479242937, "learning_rate": 8.36011598016902e-06, "loss": 1.2714, "step": 4184 }, { "epoch": 0.3202969539262207, "grad_norm": 0.6734953741083048, "learning_rate": 8.359315382016183e-06, "loss": 1.3295, "step": 4185 }, { "epoch": 0.32037348844328795, "grad_norm": 0.4923337233148323, "learning_rate": 8.358514638899959e-06, "loss": 1.0137, "step": 4186 }, { "epoch": 0.32045002296035513, "grad_norm": 0.6548319502707234, "learning_rate": 8.357713750869544e-06, "loss": 1.6515, "step": 4187 }, { "epoch": 0.3205265574774223, "grad_norm": 0.6101460983413971, "learning_rate": 8.356912717974146e-06, "loss": 1.4265, "step": 4188 }, { "epoch": 0.3206030919944895, "grad_norm": 0.5934808363324259, "learning_rate": 8.356111540262987e-06, "loss": 1.3238, "step": 4189 }, { "epoch": 0.32067962651155674, "grad_norm": 0.5972288303402568, "learning_rate": 8.355310217785291e-06, "loss": 1.3324, "step": 4190 }, { "epoch": 0.3207561610286239, "grad_norm": 0.6404104961573083, "learning_rate": 8.354508750590294e-06, "loss": 1.5033, "step": 4191 }, { "epoch": 0.3208326955456911, "grad_norm": 0.5947631405459705, "learning_rate": 8.353707138727241e-06, "loss": 1.5468, "step": 4192 }, { "epoch": 0.3209092300627583, "grad_norm": 0.5418598036797219, "learning_rate": 8.352905382245384e-06, "loss": 1.0335, "step": 4193 }, { "epoch": 0.3209857645798255, "grad_norm": 0.5671288720249104, "learning_rate": 8.352103481193989e-06, "loss": 1.2103, "step": 4194 }, { "epoch": 0.3210622990968927, "grad_norm": 0.4997396118957707, "learning_rate": 8.351301435622321e-06, "loss": 1.0938, "step": 4195 }, { "epoch": 0.3211388336139599, "grad_norm": 0.6702990290024922, "learning_rate": 8.350499245579667e-06, "loss": 1.4935, "step": 4196 }, { "epoch": 0.3212153681310271, "grad_norm": 0.6532701313406066, "learning_rate": 8.349696911115311e-06, "loss": 1.4588, "step": 4197 }, { "epoch": 0.32129190264809426, "grad_norm": 0.6487828651094762, "learning_rate": 8.348894432278554e-06, "loss": 1.8339, "step": 4198 }, { "epoch": 0.3213684371651615, "grad_norm": 0.621789280119998, "learning_rate": 8.3480918091187e-06, "loss": 1.5573, "step": 4199 }, { "epoch": 0.3214449716822287, "grad_norm": 0.6857755180647537, "learning_rate": 8.347289041685067e-06, "loss": 1.659, "step": 4200 }, { "epoch": 0.32152150619929587, "grad_norm": 0.5742408354446742, "learning_rate": 8.346486130026978e-06, "loss": 1.3987, "step": 4201 }, { "epoch": 0.32159804071636305, "grad_norm": 0.6394004269974318, "learning_rate": 8.345683074193765e-06, "loss": 1.6371, "step": 4202 }, { "epoch": 0.3216745752334303, "grad_norm": 0.544190481328548, "learning_rate": 8.344879874234771e-06, "loss": 1.2672, "step": 4203 }, { "epoch": 0.3217511097504975, "grad_norm": 0.6864445489386138, "learning_rate": 8.34407653019935e-06, "loss": 1.605, "step": 4204 }, { "epoch": 0.32182764426756466, "grad_norm": 0.7289013992608114, "learning_rate": 8.343273042136858e-06, "loss": 1.5596, "step": 4205 }, { "epoch": 0.32190417878463184, "grad_norm": 0.6515510128757853, "learning_rate": 8.342469410096667e-06, "loss": 1.6272, "step": 4206 }, { "epoch": 0.3219807133016991, "grad_norm": 0.6673943354280719, "learning_rate": 8.34166563412815e-06, "loss": 1.4529, "step": 4207 }, { "epoch": 0.32205724781876627, "grad_norm": 0.7828840996895141, "learning_rate": 8.340861714280699e-06, "loss": 1.7873, "step": 4208 }, { "epoch": 0.32213378233583345, "grad_norm": 0.6143846572776532, "learning_rate": 8.340057650603705e-06, "loss": 1.3575, "step": 4209 }, { "epoch": 0.32221031685290064, "grad_norm": 0.5764069782577028, "learning_rate": 8.339253443146574e-06, "loss": 1.3222, "step": 4210 }, { "epoch": 0.3222868513699679, "grad_norm": 0.591678351987051, "learning_rate": 8.338449091958718e-06, "loss": 1.3177, "step": 4211 }, { "epoch": 0.32236338588703506, "grad_norm": 0.5415072039708207, "learning_rate": 8.33764459708956e-06, "loss": 1.1226, "step": 4212 }, { "epoch": 0.32243992040410224, "grad_norm": 0.6830657055952223, "learning_rate": 8.336839958588528e-06, "loss": 1.3389, "step": 4213 }, { "epoch": 0.3225164549211694, "grad_norm": 0.5599767088473578, "learning_rate": 8.336035176505065e-06, "loss": 1.3813, "step": 4214 }, { "epoch": 0.32259298943823667, "grad_norm": 0.5998730109629707, "learning_rate": 8.335230250888616e-06, "loss": 1.432, "step": 4215 }, { "epoch": 0.32266952395530385, "grad_norm": 0.47547400286166047, "learning_rate": 8.334425181788642e-06, "loss": 1.0295, "step": 4216 }, { "epoch": 0.32274605847237103, "grad_norm": 0.6804412929970909, "learning_rate": 8.333619969254602e-06, "loss": 1.368, "step": 4217 }, { "epoch": 0.3228225929894382, "grad_norm": 0.9933161770569715, "learning_rate": 8.332814613335977e-06, "loss": 1.4987, "step": 4218 }, { "epoch": 0.32289912750650546, "grad_norm": 0.6783639553354174, "learning_rate": 8.332009114082248e-06, "loss": 1.801, "step": 4219 }, { "epoch": 0.32297566202357264, "grad_norm": 0.5722153478184933, "learning_rate": 8.33120347154291e-06, "loss": 1.3143, "step": 4220 }, { "epoch": 0.3230521965406398, "grad_norm": 0.7404961746204161, "learning_rate": 8.33039768576746e-06, "loss": 1.6008, "step": 4221 }, { "epoch": 0.323128731057707, "grad_norm": 0.5231690210077566, "learning_rate": 8.329591756805408e-06, "loss": 0.9685, "step": 4222 }, { "epoch": 0.32320526557477425, "grad_norm": 1.1671834115445374, "learning_rate": 8.328785684706277e-06, "loss": 1.3143, "step": 4223 }, { "epoch": 0.32328180009184143, "grad_norm": 0.5397750008003357, "learning_rate": 8.32797946951959e-06, "loss": 1.1838, "step": 4224 }, { "epoch": 0.3233583346089086, "grad_norm": 0.5129868539976223, "learning_rate": 8.327173111294886e-06, "loss": 1.1545, "step": 4225 }, { "epoch": 0.3234348691259758, "grad_norm": 0.6308368458784139, "learning_rate": 8.326366610081708e-06, "loss": 1.5054, "step": 4226 }, { "epoch": 0.32351140364304304, "grad_norm": 0.5686154230942518, "learning_rate": 8.325559965929613e-06, "loss": 1.2549, "step": 4227 }, { "epoch": 0.3235879381601102, "grad_norm": 0.5893338383687862, "learning_rate": 8.32475317888816e-06, "loss": 1.3147, "step": 4228 }, { "epoch": 0.3236644726771774, "grad_norm": 0.5393748070965411, "learning_rate": 8.323946249006922e-06, "loss": 1.2197, "step": 4229 }, { "epoch": 0.3237410071942446, "grad_norm": 0.8867681728567636, "learning_rate": 8.32313917633548e-06, "loss": 1.6415, "step": 4230 }, { "epoch": 0.32381754171131183, "grad_norm": 0.613141975823701, "learning_rate": 8.322331960923422e-06, "loss": 1.4161, "step": 4231 }, { "epoch": 0.323894076228379, "grad_norm": 0.9536441569267113, "learning_rate": 8.321524602820344e-06, "loss": 1.352, "step": 4232 }, { "epoch": 0.3239706107454462, "grad_norm": 0.6651825984848809, "learning_rate": 8.320717102075855e-06, "loss": 1.5098, "step": 4233 }, { "epoch": 0.3240471452625134, "grad_norm": 0.5944402119126915, "learning_rate": 8.319909458739572e-06, "loss": 1.4046, "step": 4234 }, { "epoch": 0.32412367977958056, "grad_norm": 0.5752003342078105, "learning_rate": 8.319101672861116e-06, "loss": 1.2258, "step": 4235 }, { "epoch": 0.3242002142966478, "grad_norm": 0.5212600060726377, "learning_rate": 8.31829374449012e-06, "loss": 1.2317, "step": 4236 }, { "epoch": 0.324276748813715, "grad_norm": 1.2630430357740718, "learning_rate": 8.317485673676225e-06, "loss": 1.2537, "step": 4237 }, { "epoch": 0.32435328333078217, "grad_norm": 0.6836140551363259, "learning_rate": 8.316677460469082e-06, "loss": 1.6318, "step": 4238 }, { "epoch": 0.32442981784784936, "grad_norm": 0.5114326765923132, "learning_rate": 8.315869104918352e-06, "loss": 1.1833, "step": 4239 }, { "epoch": 0.3245063523649166, "grad_norm": 0.5671957350438477, "learning_rate": 8.315060607073699e-06, "loss": 1.3914, "step": 4240 }, { "epoch": 0.3245828868819838, "grad_norm": 0.6295169314900633, "learning_rate": 8.314251966984803e-06, "loss": 1.7366, "step": 4241 }, { "epoch": 0.32465942139905096, "grad_norm": 0.5617171660396117, "learning_rate": 8.313443184701348e-06, "loss": 1.3341, "step": 4242 }, { "epoch": 0.32473595591611815, "grad_norm": 0.7989109857267909, "learning_rate": 8.312634260273027e-06, "loss": 1.7299, "step": 4243 }, { "epoch": 0.3248124904331854, "grad_norm": 0.5833818376513673, "learning_rate": 8.311825193749545e-06, "loss": 1.4734, "step": 4244 }, { "epoch": 0.32488902495025257, "grad_norm": 0.6711573465590046, "learning_rate": 8.31101598518061e-06, "loss": 1.3615, "step": 4245 }, { "epoch": 0.32496555946731975, "grad_norm": 1.6379776707329867, "learning_rate": 8.310206634615944e-06, "loss": 1.6631, "step": 4246 }, { "epoch": 0.32504209398438694, "grad_norm": 0.5813989340380703, "learning_rate": 8.309397142105278e-06, "loss": 1.5628, "step": 4247 }, { "epoch": 0.3251186285014542, "grad_norm": 0.9486925114347396, "learning_rate": 8.308587507698348e-06, "loss": 1.726, "step": 4248 }, { "epoch": 0.32519516301852136, "grad_norm": 0.7778722520079674, "learning_rate": 8.307777731444899e-06, "loss": 1.5731, "step": 4249 }, { "epoch": 0.32527169753558854, "grad_norm": 0.629871695442103, "learning_rate": 8.306967813394687e-06, "loss": 1.4308, "step": 4250 }, { "epoch": 0.32534823205265573, "grad_norm": 0.5803063382483872, "learning_rate": 8.306157753597477e-06, "loss": 1.3565, "step": 4251 }, { "epoch": 0.32542476656972297, "grad_norm": 0.5506617608852296, "learning_rate": 8.30534755210304e-06, "loss": 1.3733, "step": 4252 }, { "epoch": 0.32550130108679015, "grad_norm": 0.6924708875016918, "learning_rate": 8.304537208961159e-06, "loss": 1.3958, "step": 4253 }, { "epoch": 0.32557783560385734, "grad_norm": 0.6890010127298333, "learning_rate": 8.30372672422162e-06, "loss": 1.7822, "step": 4254 }, { "epoch": 0.3256543701209245, "grad_norm": 0.570509013656356, "learning_rate": 8.302916097934226e-06, "loss": 1.4317, "step": 4255 }, { "epoch": 0.32573090463799176, "grad_norm": 0.4945551586754541, "learning_rate": 8.302105330148784e-06, "loss": 1.0677, "step": 4256 }, { "epoch": 0.32580743915505894, "grad_norm": 0.6273501863923505, "learning_rate": 8.301294420915107e-06, "loss": 1.4236, "step": 4257 }, { "epoch": 0.3258839736721261, "grad_norm": 0.5606854415420652, "learning_rate": 8.30048337028302e-06, "loss": 1.2227, "step": 4258 }, { "epoch": 0.3259605081891933, "grad_norm": 0.7033020480716526, "learning_rate": 8.29967217830236e-06, "loss": 1.4671, "step": 4259 }, { "epoch": 0.32603704270626055, "grad_norm": 0.5652767323377944, "learning_rate": 8.298860845022964e-06, "loss": 1.4379, "step": 4260 }, { "epoch": 0.32611357722332773, "grad_norm": 0.6285309767754838, "learning_rate": 8.298049370494688e-06, "loss": 1.6034, "step": 4261 }, { "epoch": 0.3261901117403949, "grad_norm": 0.540216714034004, "learning_rate": 8.297237754767387e-06, "loss": 1.4209, "step": 4262 }, { "epoch": 0.3262666462574621, "grad_norm": 0.6325408122339482, "learning_rate": 8.29642599789093e-06, "loss": 1.5317, "step": 4263 }, { "epoch": 0.32634318077452934, "grad_norm": 0.6016577825723999, "learning_rate": 8.295614099915197e-06, "loss": 1.4453, "step": 4264 }, { "epoch": 0.3264197152915965, "grad_norm": 0.6167250058300993, "learning_rate": 8.29480206089007e-06, "loss": 1.452, "step": 4265 }, { "epoch": 0.3264962498086637, "grad_norm": 0.5426367910349271, "learning_rate": 8.293989880865444e-06, "loss": 1.1105, "step": 4266 }, { "epoch": 0.3265727843257309, "grad_norm": 0.7077491168625009, "learning_rate": 8.293177559891224e-06, "loss": 1.7723, "step": 4267 }, { "epoch": 0.3266493188427981, "grad_norm": 1.259519106773177, "learning_rate": 8.292365098017317e-06, "loss": 1.487, "step": 4268 }, { "epoch": 0.3267258533598653, "grad_norm": 0.486536469131935, "learning_rate": 8.291552495293647e-06, "loss": 1.0077, "step": 4269 }, { "epoch": 0.3268023878769325, "grad_norm": 0.6044748445280557, "learning_rate": 8.29073975177014e-06, "loss": 1.2587, "step": 4270 }, { "epoch": 0.3268789223939997, "grad_norm": 0.6793762240758123, "learning_rate": 8.289926867496736e-06, "loss": 1.9061, "step": 4271 }, { "epoch": 0.32695545691106687, "grad_norm": 0.5714174799201863, "learning_rate": 8.289113842523378e-06, "loss": 1.3938, "step": 4272 }, { "epoch": 0.3270319914281341, "grad_norm": 0.534975856820526, "learning_rate": 8.288300676900024e-06, "loss": 1.269, "step": 4273 }, { "epoch": 0.3271085259452013, "grad_norm": 0.5453429414613085, "learning_rate": 8.287487370676635e-06, "loss": 1.2561, "step": 4274 }, { "epoch": 0.3271850604622685, "grad_norm": 0.6127121958309552, "learning_rate": 8.286673923903184e-06, "loss": 1.5071, "step": 4275 }, { "epoch": 0.32726159497933566, "grad_norm": 0.6922379037367726, "learning_rate": 8.28586033662965e-06, "loss": 1.554, "step": 4276 }, { "epoch": 0.3273381294964029, "grad_norm": 0.6337011204520819, "learning_rate": 8.285046608906026e-06, "loss": 1.5794, "step": 4277 }, { "epoch": 0.3274146640134701, "grad_norm": 0.5322871740894904, "learning_rate": 8.284232740782305e-06, "loss": 1.1441, "step": 4278 }, { "epoch": 0.32749119853053726, "grad_norm": 0.6020126952936288, "learning_rate": 8.283418732308496e-06, "loss": 1.6006, "step": 4279 }, { "epoch": 0.32756773304760445, "grad_norm": 1.129967443001579, "learning_rate": 8.282604583534616e-06, "loss": 1.5759, "step": 4280 }, { "epoch": 0.3276442675646717, "grad_norm": 0.6018843005014187, "learning_rate": 8.281790294510684e-06, "loss": 1.3731, "step": 4281 }, { "epoch": 0.32772080208173887, "grad_norm": 1.5057692212093627, "learning_rate": 8.280975865286737e-06, "loss": 1.6347, "step": 4282 }, { "epoch": 0.32779733659880606, "grad_norm": 0.7961965328583569, "learning_rate": 8.280161295912814e-06, "loss": 1.531, "step": 4283 }, { "epoch": 0.32787387111587324, "grad_norm": 0.592640973931532, "learning_rate": 8.279346586438965e-06, "loss": 1.337, "step": 4284 }, { "epoch": 0.3279504056329405, "grad_norm": 0.5999616063651007, "learning_rate": 8.278531736915246e-06, "loss": 1.2477, "step": 4285 }, { "epoch": 0.32802694015000766, "grad_norm": 0.4905334301552148, "learning_rate": 8.277716747391726e-06, "loss": 1.0604, "step": 4286 }, { "epoch": 0.32810347466707485, "grad_norm": 0.5900526753382659, "learning_rate": 8.27690161791848e-06, "loss": 1.4141, "step": 4287 }, { "epoch": 0.32818000918414203, "grad_norm": 0.5158581226361281, "learning_rate": 8.276086348545593e-06, "loss": 1.0196, "step": 4288 }, { "epoch": 0.32825654370120927, "grad_norm": 0.8868191987694686, "learning_rate": 8.275270939323155e-06, "loss": 1.5142, "step": 4289 }, { "epoch": 0.32833307821827645, "grad_norm": 0.5705485976440031, "learning_rate": 8.27445539030127e-06, "loss": 1.3741, "step": 4290 }, { "epoch": 0.32840961273534364, "grad_norm": 0.6086964577548702, "learning_rate": 8.273639701530046e-06, "loss": 1.543, "step": 4291 }, { "epoch": 0.3284861472524108, "grad_norm": 0.8256151152649626, "learning_rate": 8.272823873059601e-06, "loss": 1.6319, "step": 4292 }, { "epoch": 0.32856268176947806, "grad_norm": 0.637271502179199, "learning_rate": 8.272007904940066e-06, "loss": 1.6234, "step": 4293 }, { "epoch": 0.32863921628654524, "grad_norm": 0.53123774665632, "learning_rate": 8.271191797221569e-06, "loss": 1.2628, "step": 4294 }, { "epoch": 0.3287157508036124, "grad_norm": 0.6906999528505738, "learning_rate": 8.27037554995426e-06, "loss": 1.7702, "step": 4295 }, { "epoch": 0.3287922853206796, "grad_norm": 1.2799642221236918, "learning_rate": 8.269559163188291e-06, "loss": 1.3794, "step": 4296 }, { "epoch": 0.32886881983774685, "grad_norm": 0.5420099583850038, "learning_rate": 8.268742636973822e-06, "loss": 1.3642, "step": 4297 }, { "epoch": 0.32894535435481403, "grad_norm": 0.5668342688708291, "learning_rate": 8.267925971361022e-06, "loss": 1.4328, "step": 4298 }, { "epoch": 0.3290218888718812, "grad_norm": 0.6439217658582455, "learning_rate": 8.267109166400072e-06, "loss": 1.5014, "step": 4299 }, { "epoch": 0.3290984233889484, "grad_norm": 0.61057655175208, "learning_rate": 8.266292222141157e-06, "loss": 1.6478, "step": 4300 }, { "epoch": 0.3291749579060156, "grad_norm": 2.5552431641557147, "learning_rate": 8.265475138634473e-06, "loss": 1.682, "step": 4301 }, { "epoch": 0.3292514924230828, "grad_norm": 0.5337304486648582, "learning_rate": 8.264657915930224e-06, "loss": 1.2039, "step": 4302 }, { "epoch": 0.32932802694015, "grad_norm": 0.6530814625280752, "learning_rate": 8.263840554078622e-06, "loss": 1.3151, "step": 4303 }, { "epoch": 0.3294045614572172, "grad_norm": 0.6659497947515464, "learning_rate": 8.263023053129889e-06, "loss": 1.5205, "step": 4304 }, { "epoch": 0.3294810959742844, "grad_norm": 0.9862525955856223, "learning_rate": 8.262205413134256e-06, "loss": 1.3698, "step": 4305 }, { "epoch": 0.3295576304913516, "grad_norm": 0.5901928974472606, "learning_rate": 8.261387634141958e-06, "loss": 1.5578, "step": 4306 }, { "epoch": 0.3296341650084188, "grad_norm": 0.7174776839516411, "learning_rate": 8.260569716203244e-06, "loss": 1.7554, "step": 4307 }, { "epoch": 0.329710699525486, "grad_norm": 0.5985849773812356, "learning_rate": 8.259751659368367e-06, "loss": 1.506, "step": 4308 }, { "epoch": 0.32978723404255317, "grad_norm": 0.6323602827038789, "learning_rate": 8.258933463687596e-06, "loss": 1.3211, "step": 4309 }, { "epoch": 0.3298637685596204, "grad_norm": 0.6199726751244453, "learning_rate": 8.258115129211199e-06, "loss": 1.4321, "step": 4310 }, { "epoch": 0.3299403030766876, "grad_norm": 0.6195285232027523, "learning_rate": 8.257296655989456e-06, "loss": 1.502, "step": 4311 }, { "epoch": 0.3300168375937548, "grad_norm": 0.682477962349129, "learning_rate": 8.25647804407266e-06, "loss": 1.7235, "step": 4312 }, { "epoch": 0.33009337211082196, "grad_norm": 0.6907985276380161, "learning_rate": 8.255659293511106e-06, "loss": 1.5728, "step": 4313 }, { "epoch": 0.3301699066278892, "grad_norm": 0.5979919442879992, "learning_rate": 8.254840404355105e-06, "loss": 1.4449, "step": 4314 }, { "epoch": 0.3302464411449564, "grad_norm": 0.7727064589431218, "learning_rate": 8.254021376654965e-06, "loss": 1.4488, "step": 4315 }, { "epoch": 0.33032297566202357, "grad_norm": 0.5996934522780252, "learning_rate": 8.253202210461017e-06, "loss": 1.5022, "step": 4316 }, { "epoch": 0.33039951017909075, "grad_norm": 0.5549294124204507, "learning_rate": 8.252382905823586e-06, "loss": 1.4123, "step": 4317 }, { "epoch": 0.330476044696158, "grad_norm": 0.6438712849161423, "learning_rate": 8.25156346279302e-06, "loss": 1.2835, "step": 4318 }, { "epoch": 0.3305525792132252, "grad_norm": 1.3548624594912018, "learning_rate": 8.250743881419662e-06, "loss": 1.6139, "step": 4319 }, { "epoch": 0.33062911373029236, "grad_norm": 0.5125219042108615, "learning_rate": 8.24992416175387e-06, "loss": 1.1823, "step": 4320 }, { "epoch": 0.33070564824735954, "grad_norm": 0.6170901817080113, "learning_rate": 8.249104303846014e-06, "loss": 1.5478, "step": 4321 }, { "epoch": 0.3307821827644268, "grad_norm": 0.6198598909802584, "learning_rate": 8.248284307746463e-06, "loss": 1.2427, "step": 4322 }, { "epoch": 0.33085871728149396, "grad_norm": 0.5124671814602421, "learning_rate": 8.247464173505606e-06, "loss": 1.0753, "step": 4323 }, { "epoch": 0.33093525179856115, "grad_norm": 1.1698564622403982, "learning_rate": 8.24664390117383e-06, "loss": 1.2922, "step": 4324 }, { "epoch": 0.33101178631562833, "grad_norm": 0.5820832721260873, "learning_rate": 8.245823490801535e-06, "loss": 1.3047, "step": 4325 }, { "epoch": 0.33108832083269557, "grad_norm": 0.6570782652435039, "learning_rate": 8.245002942439134e-06, "loss": 1.3728, "step": 4326 }, { "epoch": 0.33116485534976275, "grad_norm": 0.6058624191147653, "learning_rate": 8.24418225613704e-06, "loss": 1.616, "step": 4327 }, { "epoch": 0.33124138986682994, "grad_norm": 0.7976971565880733, "learning_rate": 8.243361431945678e-06, "loss": 1.5187, "step": 4328 }, { "epoch": 0.3313179243838971, "grad_norm": 1.2969833991552364, "learning_rate": 8.242540469915486e-06, "loss": 1.542, "step": 4329 }, { "epoch": 0.33139445890096436, "grad_norm": 0.6509410568483531, "learning_rate": 8.2417193700969e-06, "loss": 1.5295, "step": 4330 }, { "epoch": 0.33147099341803155, "grad_norm": 0.7665354984077956, "learning_rate": 8.240898132540377e-06, "loss": 1.8651, "step": 4331 }, { "epoch": 0.33154752793509873, "grad_norm": 0.6238945844570094, "learning_rate": 8.240076757296373e-06, "loss": 1.3726, "step": 4332 }, { "epoch": 0.3316240624521659, "grad_norm": 0.6171289236975873, "learning_rate": 8.239255244415356e-06, "loss": 1.4794, "step": 4333 }, { "epoch": 0.3317005969692331, "grad_norm": 0.5959556309884817, "learning_rate": 8.238433593947804e-06, "loss": 1.2958, "step": 4334 }, { "epoch": 0.33177713148630034, "grad_norm": 0.573352160011903, "learning_rate": 8.2376118059442e-06, "loss": 1.1864, "step": 4335 }, { "epoch": 0.3318536660033675, "grad_norm": 0.5934506994321345, "learning_rate": 8.236789880455036e-06, "loss": 1.4183, "step": 4336 }, { "epoch": 0.3319302005204347, "grad_norm": 0.7480764647501023, "learning_rate": 8.235967817530815e-06, "loss": 1.683, "step": 4337 }, { "epoch": 0.3320067350375019, "grad_norm": 0.6016408471677077, "learning_rate": 8.23514561722205e-06, "loss": 1.4905, "step": 4338 }, { "epoch": 0.3320832695545691, "grad_norm": 0.8374440449137671, "learning_rate": 8.234323279579253e-06, "loss": 1.528, "step": 4339 }, { "epoch": 0.3321598040716363, "grad_norm": 0.7012320322703066, "learning_rate": 8.233500804652957e-06, "loss": 1.3591, "step": 4340 }, { "epoch": 0.3322363385887035, "grad_norm": 0.6237188882643069, "learning_rate": 8.232678192493694e-06, "loss": 1.4209, "step": 4341 }, { "epoch": 0.3323128731057707, "grad_norm": 0.6072845463183759, "learning_rate": 8.231855443152007e-06, "loss": 1.5077, "step": 4342 }, { "epoch": 0.3323894076228379, "grad_norm": 0.5431896458188126, "learning_rate": 8.231032556678451e-06, "loss": 1.3004, "step": 4343 }, { "epoch": 0.3324659421399051, "grad_norm": 0.8222569036094803, "learning_rate": 8.230209533123585e-06, "loss": 1.4372, "step": 4344 }, { "epoch": 0.3325424766569723, "grad_norm": 0.5933166463006548, "learning_rate": 8.229386372537978e-06, "loss": 1.4553, "step": 4345 }, { "epoch": 0.33261901117403947, "grad_norm": 1.9371418623875933, "learning_rate": 8.228563074972207e-06, "loss": 1.6738, "step": 4346 }, { "epoch": 0.3326955456911067, "grad_norm": 0.5566335046872357, "learning_rate": 8.227739640476859e-06, "loss": 1.1865, "step": 4347 }, { "epoch": 0.3327720802081739, "grad_norm": 0.6319999222428709, "learning_rate": 8.226916069102528e-06, "loss": 1.5134, "step": 4348 }, { "epoch": 0.3328486147252411, "grad_norm": 0.6087120991296949, "learning_rate": 8.226092360899816e-06, "loss": 1.5012, "step": 4349 }, { "epoch": 0.33292514924230826, "grad_norm": 0.6448651731102086, "learning_rate": 8.225268515919335e-06, "loss": 1.6049, "step": 4350 }, { "epoch": 0.3330016837593755, "grad_norm": 0.6209362547076124, "learning_rate": 8.224444534211705e-06, "loss": 1.3628, "step": 4351 }, { "epoch": 0.3330782182764427, "grad_norm": 0.6197005126390293, "learning_rate": 8.223620415827554e-06, "loss": 1.6056, "step": 4352 }, { "epoch": 0.33315475279350987, "grad_norm": 0.5390972623656988, "learning_rate": 8.222796160817514e-06, "loss": 1.3204, "step": 4353 }, { "epoch": 0.33323128731057705, "grad_norm": 0.5947146765381962, "learning_rate": 8.221971769232235e-06, "loss": 1.4211, "step": 4354 }, { "epoch": 0.3333078218276443, "grad_norm": 0.5514475589305408, "learning_rate": 8.221147241122365e-06, "loss": 1.2951, "step": 4355 }, { "epoch": 0.3333843563447115, "grad_norm": 0.5810207701474237, "learning_rate": 8.220322576538571e-06, "loss": 1.43, "step": 4356 }, { "epoch": 0.33346089086177866, "grad_norm": 0.5605251922991494, "learning_rate": 8.219497775531523e-06, "loss": 1.4551, "step": 4357 }, { "epoch": 0.33353742537884584, "grad_norm": 0.5838737644736727, "learning_rate": 8.218672838151893e-06, "loss": 1.3189, "step": 4358 }, { "epoch": 0.3336139598959131, "grad_norm": 0.5927971285027286, "learning_rate": 8.217847764450374e-06, "loss": 1.4378, "step": 4359 }, { "epoch": 0.33369049441298027, "grad_norm": 0.6640565349569418, "learning_rate": 8.217022554477657e-06, "loss": 1.6022, "step": 4360 }, { "epoch": 0.33376702893004745, "grad_norm": 0.6425180390805952, "learning_rate": 8.216197208284444e-06, "loss": 1.4688, "step": 4361 }, { "epoch": 0.33384356344711463, "grad_norm": 0.8931149676164398, "learning_rate": 8.215371725921452e-06, "loss": 1.3888, "step": 4362 }, { "epoch": 0.3339200979641819, "grad_norm": 0.5196218814720426, "learning_rate": 8.214546107439399e-06, "loss": 1.1544, "step": 4363 }, { "epoch": 0.33399663248124906, "grad_norm": 0.5846902563185702, "learning_rate": 8.21372035288901e-06, "loss": 1.2099, "step": 4364 }, { "epoch": 0.33407316699831624, "grad_norm": 0.6158668347268245, "learning_rate": 8.212894462321027e-06, "loss": 1.4155, "step": 4365 }, { "epoch": 0.3341497015153834, "grad_norm": 0.6533865018846113, "learning_rate": 8.212068435786193e-06, "loss": 1.5399, "step": 4366 }, { "epoch": 0.3342262360324506, "grad_norm": 0.7295209538569893, "learning_rate": 8.211242273335262e-06, "loss": 1.8148, "step": 4367 }, { "epoch": 0.33430277054951785, "grad_norm": 0.5742426214418197, "learning_rate": 8.210415975018993e-06, "loss": 1.4604, "step": 4368 }, { "epoch": 0.33437930506658503, "grad_norm": 0.6678649949493365, "learning_rate": 8.20958954088816e-06, "loss": 1.653, "step": 4369 }, { "epoch": 0.3344558395836522, "grad_norm": 1.1719387291549375, "learning_rate": 8.20876297099354e-06, "loss": 1.3087, "step": 4370 }, { "epoch": 0.3345323741007194, "grad_norm": 1.0912634355703865, "learning_rate": 8.207936265385922e-06, "loss": 1.4089, "step": 4371 }, { "epoch": 0.33460890861778664, "grad_norm": 0.5470446264430574, "learning_rate": 8.207109424116096e-06, "loss": 1.3432, "step": 4372 }, { "epoch": 0.3346854431348538, "grad_norm": 0.5371302760061124, "learning_rate": 8.206282447234872e-06, "loss": 1.3288, "step": 4373 }, { "epoch": 0.334761977651921, "grad_norm": 0.5203419909737476, "learning_rate": 8.205455334793059e-06, "loss": 1.1631, "step": 4374 }, { "epoch": 0.3348385121689882, "grad_norm": 0.7213964520717525, "learning_rate": 8.204628086841475e-06, "loss": 1.292, "step": 4375 }, { "epoch": 0.33491504668605543, "grad_norm": 0.5743207198913424, "learning_rate": 8.203800703430952e-06, "loss": 1.2803, "step": 4376 }, { "epoch": 0.3349915812031226, "grad_norm": 0.6049907808637415, "learning_rate": 8.202973184612324e-06, "loss": 1.5028, "step": 4377 }, { "epoch": 0.3350681157201898, "grad_norm": 0.7843671074910272, "learning_rate": 8.20214553043644e-06, "loss": 1.5603, "step": 4378 }, { "epoch": 0.335144650237257, "grad_norm": 1.162173713981155, "learning_rate": 8.201317740954151e-06, "loss": 1.6003, "step": 4379 }, { "epoch": 0.3352211847543242, "grad_norm": 0.6711265012988133, "learning_rate": 8.200489816216319e-06, "loss": 1.4289, "step": 4380 }, { "epoch": 0.3352977192713914, "grad_norm": 0.4927853904909478, "learning_rate": 8.199661756273813e-06, "loss": 1.142, "step": 4381 }, { "epoch": 0.3353742537884586, "grad_norm": 0.5906494566993247, "learning_rate": 8.198833561177515e-06, "loss": 1.3852, "step": 4382 }, { "epoch": 0.33545078830552577, "grad_norm": 0.5137030891693369, "learning_rate": 8.19800523097831e-06, "loss": 1.094, "step": 4383 }, { "epoch": 0.335527322822593, "grad_norm": 0.5834735419282409, "learning_rate": 8.197176765727091e-06, "loss": 1.3545, "step": 4384 }, { "epoch": 0.3356038573396602, "grad_norm": 0.7881102159365074, "learning_rate": 8.196348165474765e-06, "loss": 1.3778, "step": 4385 }, { "epoch": 0.3356803918567274, "grad_norm": 0.5779700462318415, "learning_rate": 8.19551943027224e-06, "loss": 1.4879, "step": 4386 }, { "epoch": 0.33575692637379456, "grad_norm": 0.5648681003525122, "learning_rate": 8.19469056017044e-06, "loss": 1.3722, "step": 4387 }, { "epoch": 0.3358334608908618, "grad_norm": 0.765950562097947, "learning_rate": 8.19386155522029e-06, "loss": 1.7481, "step": 4388 }, { "epoch": 0.335909995407929, "grad_norm": 0.5623113361461216, "learning_rate": 8.193032415472727e-06, "loss": 1.2721, "step": 4389 }, { "epoch": 0.33598652992499617, "grad_norm": 0.6680706978826184, "learning_rate": 8.192203140978697e-06, "loss": 1.615, "step": 4390 }, { "epoch": 0.33606306444206335, "grad_norm": 0.5376498643116627, "learning_rate": 8.191373731789153e-06, "loss": 1.0282, "step": 4391 }, { "epoch": 0.3361395989591306, "grad_norm": 0.6401553346806206, "learning_rate": 8.190544187955057e-06, "loss": 1.8097, "step": 4392 }, { "epoch": 0.3362161334761978, "grad_norm": 2.034025642216135, "learning_rate": 8.189714509527376e-06, "loss": 1.4624, "step": 4393 }, { "epoch": 0.33629266799326496, "grad_norm": 0.627364704721108, "learning_rate": 8.188884696557089e-06, "loss": 1.4136, "step": 4394 }, { "epoch": 0.33636920251033214, "grad_norm": 0.5646541585048834, "learning_rate": 8.188054749095184e-06, "loss": 1.4035, "step": 4395 }, { "epoch": 0.3364457370273994, "grad_norm": 0.5131153155477981, "learning_rate": 8.187224667192653e-06, "loss": 1.2227, "step": 4396 }, { "epoch": 0.33652227154446657, "grad_norm": 0.6565041533961183, "learning_rate": 8.1863944509005e-06, "loss": 1.5901, "step": 4397 }, { "epoch": 0.33659880606153375, "grad_norm": 0.5763293410897393, "learning_rate": 8.185564100269737e-06, "loss": 1.0504, "step": 4398 }, { "epoch": 0.33667534057860093, "grad_norm": 0.5871605630376319, "learning_rate": 8.184733615351381e-06, "loss": 1.2695, "step": 4399 }, { "epoch": 0.3367518750956682, "grad_norm": 0.594113336195451, "learning_rate": 8.18390299619646e-06, "loss": 1.451, "step": 4400 }, { "epoch": 0.33682840961273536, "grad_norm": 0.5543767427976568, "learning_rate": 8.183072242856013e-06, "loss": 0.9849, "step": 4401 }, { "epoch": 0.33690494412980254, "grad_norm": 0.6784766812985572, "learning_rate": 8.18224135538108e-06, "loss": 1.5743, "step": 4402 }, { "epoch": 0.3369814786468697, "grad_norm": 0.5572543422586453, "learning_rate": 8.181410333822713e-06, "loss": 1.428, "step": 4403 }, { "epoch": 0.3370580131639369, "grad_norm": 0.5888623588280647, "learning_rate": 8.180579178231977e-06, "loss": 1.3872, "step": 4404 }, { "epoch": 0.33713454768100415, "grad_norm": 0.6296239548183807, "learning_rate": 8.179747888659935e-06, "loss": 1.4327, "step": 4405 }, { "epoch": 0.33721108219807133, "grad_norm": 0.6659697359826515, "learning_rate": 8.178916465157666e-06, "loss": 1.7051, "step": 4406 }, { "epoch": 0.3372876167151385, "grad_norm": 0.6433074334056607, "learning_rate": 8.178084907776258e-06, "loss": 1.4934, "step": 4407 }, { "epoch": 0.3373641512322057, "grad_norm": 0.5840171764045664, "learning_rate": 8.1772532165668e-06, "loss": 1.2316, "step": 4408 }, { "epoch": 0.33744068574927294, "grad_norm": 0.8638892341030334, "learning_rate": 8.176421391580397e-06, "loss": 1.5266, "step": 4409 }, { "epoch": 0.3375172202663401, "grad_norm": 0.6526082694305543, "learning_rate": 8.175589432868156e-06, "loss": 1.6012, "step": 4410 }, { "epoch": 0.3375937547834073, "grad_norm": 0.5373363586671441, "learning_rate": 8.174757340481198e-06, "loss": 1.2539, "step": 4411 }, { "epoch": 0.3376702893004745, "grad_norm": 0.6925990485708491, "learning_rate": 8.173925114470648e-06, "loss": 1.4233, "step": 4412 }, { "epoch": 0.33774682381754173, "grad_norm": 0.6896269023033319, "learning_rate": 8.173092754887637e-06, "loss": 1.706, "step": 4413 }, { "epoch": 0.3378233583346089, "grad_norm": 0.6385988443573225, "learning_rate": 8.172260261783315e-06, "loss": 1.4736, "step": 4414 }, { "epoch": 0.3378998928516761, "grad_norm": 0.6643101790247817, "learning_rate": 8.171427635208826e-06, "loss": 1.2974, "step": 4415 }, { "epoch": 0.3379764273687433, "grad_norm": 0.5666452399266534, "learning_rate": 8.170594875215332e-06, "loss": 1.1823, "step": 4416 }, { "epoch": 0.3380529618858105, "grad_norm": 0.5844866114370416, "learning_rate": 8.169761981854002e-06, "loss": 1.2613, "step": 4417 }, { "epoch": 0.3381294964028777, "grad_norm": 0.5164194382239743, "learning_rate": 8.168928955176007e-06, "loss": 1.1592, "step": 4418 }, { "epoch": 0.3382060309199449, "grad_norm": 0.5513555346215446, "learning_rate": 8.168095795232534e-06, "loss": 1.2681, "step": 4419 }, { "epoch": 0.3382825654370121, "grad_norm": 0.5930464617982115, "learning_rate": 8.167262502074775e-06, "loss": 1.3805, "step": 4420 }, { "epoch": 0.3383590999540793, "grad_norm": 0.9978653642783533, "learning_rate": 8.166429075753927e-06, "loss": 1.6399, "step": 4421 }, { "epoch": 0.3384356344711465, "grad_norm": 0.5988207754284474, "learning_rate": 8.165595516321201e-06, "loss": 1.4283, "step": 4422 }, { "epoch": 0.3385121689882137, "grad_norm": 0.9098162753136722, "learning_rate": 8.164761823827814e-06, "loss": 1.6184, "step": 4423 }, { "epoch": 0.33858870350528086, "grad_norm": 0.6352813958927358, "learning_rate": 8.163927998324987e-06, "loss": 1.3809, "step": 4424 }, { "epoch": 0.3386652380223481, "grad_norm": 0.6043646589766111, "learning_rate": 8.163094039863957e-06, "loss": 1.1357, "step": 4425 }, { "epoch": 0.3387417725394153, "grad_norm": 0.7509889718509689, "learning_rate": 8.162259948495962e-06, "loss": 1.465, "step": 4426 }, { "epoch": 0.33881830705648247, "grad_norm": 0.5189842934596526, "learning_rate": 8.161425724272252e-06, "loss": 1.2331, "step": 4427 }, { "epoch": 0.33889484157354965, "grad_norm": 0.665215252579417, "learning_rate": 8.160591367244084e-06, "loss": 1.4472, "step": 4428 }, { "epoch": 0.3389713760906169, "grad_norm": 0.6491589835719559, "learning_rate": 8.159756877462722e-06, "loss": 1.3676, "step": 4429 }, { "epoch": 0.3390479106076841, "grad_norm": 0.5030959787612526, "learning_rate": 8.15892225497944e-06, "loss": 0.981, "step": 4430 }, { "epoch": 0.33912444512475126, "grad_norm": 0.5352824780584503, "learning_rate": 8.158087499845525e-06, "loss": 1.1883, "step": 4431 }, { "epoch": 0.33920097964181845, "grad_norm": 0.5512165946245374, "learning_rate": 8.15725261211226e-06, "loss": 1.2881, "step": 4432 }, { "epoch": 0.3392775141588857, "grad_norm": 0.655146630978072, "learning_rate": 8.156417591830946e-06, "loss": 1.6227, "step": 4433 }, { "epoch": 0.33935404867595287, "grad_norm": 0.5945633790022643, "learning_rate": 8.155582439052889e-06, "loss": 1.4621, "step": 4434 }, { "epoch": 0.33943058319302005, "grad_norm": 0.633393640902841, "learning_rate": 8.154747153829401e-06, "loss": 1.5136, "step": 4435 }, { "epoch": 0.33950711771008724, "grad_norm": 0.5908497880519578, "learning_rate": 8.153911736211808e-06, "loss": 1.4956, "step": 4436 }, { "epoch": 0.3395836522271544, "grad_norm": 0.7155174816669867, "learning_rate": 8.153076186251441e-06, "loss": 1.5496, "step": 4437 }, { "epoch": 0.33966018674422166, "grad_norm": 0.6125079009087748, "learning_rate": 8.152240503999634e-06, "loss": 1.6804, "step": 4438 }, { "epoch": 0.33973672126128884, "grad_norm": 0.6101712710261138, "learning_rate": 8.151404689507735e-06, "loss": 1.5186, "step": 4439 }, { "epoch": 0.339813255778356, "grad_norm": 0.5247817563194901, "learning_rate": 8.150568742827104e-06, "loss": 1.2688, "step": 4440 }, { "epoch": 0.3398897902954232, "grad_norm": 0.5579423930508371, "learning_rate": 8.149732664009097e-06, "loss": 1.2398, "step": 4441 }, { "epoch": 0.33996632481249045, "grad_norm": 0.7471926544896551, "learning_rate": 8.148896453105093e-06, "loss": 1.4333, "step": 4442 }, { "epoch": 0.34004285932955763, "grad_norm": 0.6166294835858303, "learning_rate": 8.148060110166465e-06, "loss": 1.5202, "step": 4443 }, { "epoch": 0.3401193938466248, "grad_norm": 0.7129669214237113, "learning_rate": 8.1472236352446e-06, "loss": 1.7356, "step": 4444 }, { "epoch": 0.340195928363692, "grad_norm": 0.6117155184672542, "learning_rate": 8.1463870283909e-06, "loss": 1.4442, "step": 4445 }, { "epoch": 0.34027246288075924, "grad_norm": 0.5431237768311326, "learning_rate": 8.145550289656759e-06, "loss": 1.2164, "step": 4446 }, { "epoch": 0.3403489973978264, "grad_norm": 0.7374058895392661, "learning_rate": 8.1447134190936e-06, "loss": 1.7565, "step": 4447 }, { "epoch": 0.3404255319148936, "grad_norm": 0.80371991202284, "learning_rate": 8.143876416752834e-06, "loss": 1.5337, "step": 4448 }, { "epoch": 0.3405020664319608, "grad_norm": 0.6475739711236123, "learning_rate": 8.143039282685892e-06, "loss": 1.4619, "step": 4449 }, { "epoch": 0.34057860094902803, "grad_norm": 0.5183067840725806, "learning_rate": 8.14220201694421e-06, "loss": 1.3499, "step": 4450 }, { "epoch": 0.3406551354660952, "grad_norm": 0.6296001157439994, "learning_rate": 8.141364619579234e-06, "loss": 1.5789, "step": 4451 }, { "epoch": 0.3407316699831624, "grad_norm": 0.5919706768357217, "learning_rate": 8.140527090642412e-06, "loss": 1.5379, "step": 4452 }, { "epoch": 0.3408082045002296, "grad_norm": 0.4837926439330553, "learning_rate": 8.139689430185207e-06, "loss": 0.9799, "step": 4453 }, { "epoch": 0.3408847390172968, "grad_norm": 0.5773279081832307, "learning_rate": 8.138851638259087e-06, "loss": 1.448, "step": 4454 }, { "epoch": 0.340961273534364, "grad_norm": 0.508145745990556, "learning_rate": 8.13801371491553e-06, "loss": 0.9935, "step": 4455 }, { "epoch": 0.3410378080514312, "grad_norm": 0.6283614539169298, "learning_rate": 8.137175660206018e-06, "loss": 1.6366, "step": 4456 }, { "epoch": 0.3411143425684984, "grad_norm": 0.560897276454374, "learning_rate": 8.136337474182044e-06, "loss": 1.324, "step": 4457 }, { "epoch": 0.3411908770855656, "grad_norm": 0.5380949796339931, "learning_rate": 8.135499156895107e-06, "loss": 1.2432, "step": 4458 }, { "epoch": 0.3412674116026328, "grad_norm": 0.6698960969113958, "learning_rate": 8.134660708396723e-06, "loss": 1.7896, "step": 4459 }, { "epoch": 0.3413439461197, "grad_norm": 1.0214067246480951, "learning_rate": 8.1338221287384e-06, "loss": 1.6605, "step": 4460 }, { "epoch": 0.34142048063676717, "grad_norm": 0.5267469008569987, "learning_rate": 8.132983417971669e-06, "loss": 1.1322, "step": 4461 }, { "epoch": 0.3414970151538344, "grad_norm": 0.7270149794867184, "learning_rate": 8.132144576148057e-06, "loss": 1.572, "step": 4462 }, { "epoch": 0.3415735496709016, "grad_norm": 0.7362888702072908, "learning_rate": 8.13130560331911e-06, "loss": 1.4644, "step": 4463 }, { "epoch": 0.34165008418796877, "grad_norm": 0.5678951390968843, "learning_rate": 8.130466499536376e-06, "loss": 1.3247, "step": 4464 }, { "epoch": 0.34172661870503596, "grad_norm": 0.5992854224245051, "learning_rate": 8.12962726485141e-06, "loss": 1.4769, "step": 4465 }, { "epoch": 0.3418031532221032, "grad_norm": 0.5364364159914368, "learning_rate": 8.128787899315782e-06, "loss": 1.1927, "step": 4466 }, { "epoch": 0.3418796877391704, "grad_norm": 0.8397967971837043, "learning_rate": 8.127948402981058e-06, "loss": 1.7026, "step": 4467 }, { "epoch": 0.34195622225623756, "grad_norm": 0.5469755418347523, "learning_rate": 8.127108775898823e-06, "loss": 1.3064, "step": 4468 }, { "epoch": 0.34203275677330475, "grad_norm": 0.6384214194130744, "learning_rate": 8.126269018120668e-06, "loss": 1.5889, "step": 4469 }, { "epoch": 0.34210929129037193, "grad_norm": 0.6812990985867422, "learning_rate": 8.125429129698186e-06, "loss": 1.8295, "step": 4470 }, { "epoch": 0.34218582580743917, "grad_norm": 0.6138237692865421, "learning_rate": 8.124589110682984e-06, "loss": 1.5896, "step": 4471 }, { "epoch": 0.34226236032450635, "grad_norm": 0.5555517929955631, "learning_rate": 8.123748961126679e-06, "loss": 1.267, "step": 4472 }, { "epoch": 0.34233889484157354, "grad_norm": 0.5860800225365409, "learning_rate": 8.122908681080885e-06, "loss": 1.2472, "step": 4473 }, { "epoch": 0.3424154293586407, "grad_norm": 0.5892257896934122, "learning_rate": 8.122068270597239e-06, "loss": 1.4787, "step": 4474 }, { "epoch": 0.34249196387570796, "grad_norm": 0.5470313425280539, "learning_rate": 8.121227729727373e-06, "loss": 1.2798, "step": 4475 }, { "epoch": 0.34256849839277514, "grad_norm": 0.5351498190588271, "learning_rate": 8.120387058522933e-06, "loss": 1.2657, "step": 4476 }, { "epoch": 0.34264503290984233, "grad_norm": 0.9027479511798281, "learning_rate": 8.119546257035575e-06, "loss": 1.432, "step": 4477 }, { "epoch": 0.3427215674269095, "grad_norm": 0.5424415953262448, "learning_rate": 8.118705325316958e-06, "loss": 0.9786, "step": 4478 }, { "epoch": 0.34279810194397675, "grad_norm": 0.6368615225837659, "learning_rate": 8.117864263418753e-06, "loss": 1.611, "step": 4479 }, { "epoch": 0.34287463646104394, "grad_norm": 0.705040870322649, "learning_rate": 8.117023071392636e-06, "loss": 1.3242, "step": 4480 }, { "epoch": 0.3429511709781111, "grad_norm": 0.5887893373614583, "learning_rate": 8.116181749290292e-06, "loss": 1.2679, "step": 4481 }, { "epoch": 0.3430277054951783, "grad_norm": 0.5496019327613489, "learning_rate": 8.115340297163416e-06, "loss": 1.2012, "step": 4482 }, { "epoch": 0.34310424001224554, "grad_norm": 0.5846963740061948, "learning_rate": 8.114498715063708e-06, "loss": 1.4023, "step": 4483 }, { "epoch": 0.3431807745293127, "grad_norm": 0.6078701092306611, "learning_rate": 8.113657003042877e-06, "loss": 1.2339, "step": 4484 }, { "epoch": 0.3432573090463799, "grad_norm": 0.5583147161525499, "learning_rate": 8.112815161152644e-06, "loss": 1.395, "step": 4485 }, { "epoch": 0.3433338435634471, "grad_norm": 0.8772732469607559, "learning_rate": 8.111973189444728e-06, "loss": 1.5153, "step": 4486 }, { "epoch": 0.34341037808051433, "grad_norm": 0.5465661305630598, "learning_rate": 8.111131087970867e-06, "loss": 1.2539, "step": 4487 }, { "epoch": 0.3434869125975815, "grad_norm": 0.5751844990724819, "learning_rate": 8.110288856782803e-06, "loss": 1.3255, "step": 4488 }, { "epoch": 0.3435634471146487, "grad_norm": 0.5073162480515414, "learning_rate": 8.10944649593228e-06, "loss": 1.0272, "step": 4489 }, { "epoch": 0.3436399816317159, "grad_norm": 0.6301335841754334, "learning_rate": 8.108604005471059e-06, "loss": 1.4745, "step": 4490 }, { "epoch": 0.3437165161487831, "grad_norm": 0.6454121974708935, "learning_rate": 8.107761385450905e-06, "loss": 1.5623, "step": 4491 }, { "epoch": 0.3437930506658503, "grad_norm": 0.5562704050542709, "learning_rate": 8.106918635923591e-06, "loss": 1.3157, "step": 4492 }, { "epoch": 0.3438695851829175, "grad_norm": 0.6726631233792817, "learning_rate": 8.106075756940895e-06, "loss": 1.3856, "step": 4493 }, { "epoch": 0.3439461196999847, "grad_norm": 0.6448834078360467, "learning_rate": 8.10523274855461e-06, "loss": 1.5792, "step": 4494 }, { "epoch": 0.3440226542170519, "grad_norm": 0.5280574741808441, "learning_rate": 8.104389610816532e-06, "loss": 1.2324, "step": 4495 }, { "epoch": 0.3440991887341191, "grad_norm": 0.9303470640295494, "learning_rate": 8.103546343778466e-06, "loss": 1.4917, "step": 4496 }, { "epoch": 0.3441757232511863, "grad_norm": 0.638718767767862, "learning_rate": 8.102702947492222e-06, "loss": 1.6829, "step": 4497 }, { "epoch": 0.34425225776825347, "grad_norm": 0.6514285894057665, "learning_rate": 8.101859422009626e-06, "loss": 1.5065, "step": 4498 }, { "epoch": 0.3443287922853207, "grad_norm": 0.6009425023256864, "learning_rate": 8.101015767382502e-06, "loss": 1.1463, "step": 4499 }, { "epoch": 0.3444053268023879, "grad_norm": 0.6161437121750996, "learning_rate": 8.100171983662689e-06, "loss": 1.4126, "step": 4500 }, { "epoch": 0.3444818613194551, "grad_norm": 0.526193185598962, "learning_rate": 8.09932807090203e-06, "loss": 1.2222, "step": 4501 }, { "epoch": 0.34455839583652226, "grad_norm": 0.5956782262727885, "learning_rate": 8.098484029152379e-06, "loss": 1.4308, "step": 4502 }, { "epoch": 0.34463493035358944, "grad_norm": 0.5666083609640177, "learning_rate": 8.097639858465595e-06, "loss": 1.4931, "step": 4503 }, { "epoch": 0.3447114648706567, "grad_norm": 0.6318861316509755, "learning_rate": 8.096795558893548e-06, "loss": 1.6269, "step": 4504 }, { "epoch": 0.34478799938772386, "grad_norm": 0.5782846193588469, "learning_rate": 8.095951130488115e-06, "loss": 1.2925, "step": 4505 }, { "epoch": 0.34486453390479105, "grad_norm": 0.7154162314092397, "learning_rate": 8.095106573301177e-06, "loss": 1.7999, "step": 4506 }, { "epoch": 0.34494106842185823, "grad_norm": 0.5577622922786819, "learning_rate": 8.09426188738463e-06, "loss": 1.272, "step": 4507 }, { "epoch": 0.34501760293892547, "grad_norm": 0.6238862876825227, "learning_rate": 8.09341707279037e-06, "loss": 1.6095, "step": 4508 }, { "epoch": 0.34509413745599266, "grad_norm": 0.7257566914406982, "learning_rate": 8.092572129570308e-06, "loss": 1.5548, "step": 4509 }, { "epoch": 0.34517067197305984, "grad_norm": 0.5656968148090292, "learning_rate": 8.091727057776358e-06, "loss": 1.3573, "step": 4510 }, { "epoch": 0.345247206490127, "grad_norm": 0.8422128837024373, "learning_rate": 8.090881857460445e-06, "loss": 1.8268, "step": 4511 }, { "epoch": 0.34532374100719426, "grad_norm": 0.5141837192765628, "learning_rate": 8.0900365286745e-06, "loss": 1.3131, "step": 4512 }, { "epoch": 0.34540027552426145, "grad_norm": 0.6853942701177288, "learning_rate": 8.089191071470464e-06, "loss": 1.4603, "step": 4513 }, { "epoch": 0.34547681004132863, "grad_norm": 0.6322424145277846, "learning_rate": 8.08834548590028e-06, "loss": 1.6282, "step": 4514 }, { "epoch": 0.3455533445583958, "grad_norm": 0.6826812247046494, "learning_rate": 8.087499772015909e-06, "loss": 1.4959, "step": 4515 }, { "epoch": 0.34562987907546305, "grad_norm": 0.5679892715050473, "learning_rate": 8.086653929869312e-06, "loss": 1.243, "step": 4516 }, { "epoch": 0.34570641359253024, "grad_norm": 0.656601287313752, "learning_rate": 8.085807959512457e-06, "loss": 1.2372, "step": 4517 }, { "epoch": 0.3457829481095974, "grad_norm": 0.6756133358446494, "learning_rate": 8.084961860997327e-06, "loss": 1.4741, "step": 4518 }, { "epoch": 0.3458594826266646, "grad_norm": 0.7193021367693814, "learning_rate": 8.084115634375907e-06, "loss": 1.7252, "step": 4519 }, { "epoch": 0.34593601714373184, "grad_norm": 0.5985939496931549, "learning_rate": 8.083269279700192e-06, "loss": 1.3709, "step": 4520 }, { "epoch": 0.34601255166079903, "grad_norm": 0.769950188347995, "learning_rate": 8.082422797022185e-06, "loss": 1.5543, "step": 4521 }, { "epoch": 0.3460890861778662, "grad_norm": 0.5814360084714312, "learning_rate": 8.081576186393898e-06, "loss": 1.3716, "step": 4522 }, { "epoch": 0.3461656206949334, "grad_norm": 0.5609470744782302, "learning_rate": 8.080729447867346e-06, "loss": 1.2824, "step": 4523 }, { "epoch": 0.34624215521200064, "grad_norm": 0.776327398744409, "learning_rate": 8.079882581494555e-06, "loss": 1.5471, "step": 4524 }, { "epoch": 0.3463186897290678, "grad_norm": 0.5652321122464411, "learning_rate": 8.079035587327561e-06, "loss": 1.4223, "step": 4525 }, { "epoch": 0.346395224246135, "grad_norm": 0.6117811458984614, "learning_rate": 8.078188465418408e-06, "loss": 1.3738, "step": 4526 }, { "epoch": 0.3464717587632022, "grad_norm": 0.5104957742781843, "learning_rate": 8.077341215819142e-06, "loss": 1.1691, "step": 4527 }, { "epoch": 0.3465482932802694, "grad_norm": 0.5033586207166926, "learning_rate": 8.076493838581819e-06, "loss": 1.2321, "step": 4528 }, { "epoch": 0.3466248277973366, "grad_norm": 0.504815551257679, "learning_rate": 8.075646333758508e-06, "loss": 1.2309, "step": 4529 }, { "epoch": 0.3467013623144038, "grad_norm": 0.8430703105505211, "learning_rate": 8.074798701401282e-06, "loss": 1.3808, "step": 4530 }, { "epoch": 0.346777896831471, "grad_norm": 0.615177080558426, "learning_rate": 8.073950941562223e-06, "loss": 1.2279, "step": 4531 }, { "epoch": 0.3468544313485382, "grad_norm": 0.7001665063491935, "learning_rate": 8.073103054293417e-06, "loss": 1.4226, "step": 4532 }, { "epoch": 0.3469309658656054, "grad_norm": 0.5443329860665208, "learning_rate": 8.07225503964696e-06, "loss": 1.3355, "step": 4533 }, { "epoch": 0.3470075003826726, "grad_norm": 0.6004632145115552, "learning_rate": 8.07140689767496e-06, "loss": 1.3507, "step": 4534 }, { "epoch": 0.34708403489973977, "grad_norm": 0.5832988001926118, "learning_rate": 8.07055862842953e-06, "loss": 1.4688, "step": 4535 }, { "epoch": 0.34716056941680695, "grad_norm": 0.6429557466522096, "learning_rate": 8.069710231962784e-06, "loss": 1.4865, "step": 4536 }, { "epoch": 0.3472371039338742, "grad_norm": 0.5790844430014226, "learning_rate": 8.068861708326857e-06, "loss": 1.4321, "step": 4537 }, { "epoch": 0.3473136384509414, "grad_norm": 0.5657817522375487, "learning_rate": 8.068013057573882e-06, "loss": 1.3827, "step": 4538 }, { "epoch": 0.34739017296800856, "grad_norm": 0.5634443790884126, "learning_rate": 8.067164279756e-06, "loss": 1.4654, "step": 4539 }, { "epoch": 0.34746670748507574, "grad_norm": 0.6705117937282096, "learning_rate": 8.066315374925367e-06, "loss": 1.2853, "step": 4540 }, { "epoch": 0.347543242002143, "grad_norm": 0.520669568245515, "learning_rate": 8.06546634313414e-06, "loss": 1.0832, "step": 4541 }, { "epoch": 0.34761977651921017, "grad_norm": 1.0350088342658497, "learning_rate": 8.064617184434486e-06, "loss": 2.0553, "step": 4542 }, { "epoch": 0.34769631103627735, "grad_norm": 0.5666762734080693, "learning_rate": 8.063767898878579e-06, "loss": 1.5079, "step": 4543 }, { "epoch": 0.34777284555334453, "grad_norm": 0.5948819088112154, "learning_rate": 8.062918486518604e-06, "loss": 1.1554, "step": 4544 }, { "epoch": 0.3478493800704118, "grad_norm": 0.8926717068508943, "learning_rate": 8.06206894740675e-06, "loss": 1.4742, "step": 4545 }, { "epoch": 0.34792591458747896, "grad_norm": 0.6607359291755555, "learning_rate": 8.061219281595218e-06, "loss": 1.5858, "step": 4546 }, { "epoch": 0.34800244910454614, "grad_norm": 0.6893089476952888, "learning_rate": 8.06036948913621e-06, "loss": 1.6923, "step": 4547 }, { "epoch": 0.3480789836216133, "grad_norm": 1.1698930156711096, "learning_rate": 8.059519570081939e-06, "loss": 1.5523, "step": 4548 }, { "epoch": 0.34815551813868056, "grad_norm": 0.7206477816091845, "learning_rate": 8.05866952448463e-06, "loss": 1.6713, "step": 4549 }, { "epoch": 0.34823205265574775, "grad_norm": 0.5458980956319903, "learning_rate": 8.057819352396512e-06, "loss": 1.2146, "step": 4550 }, { "epoch": 0.34830858717281493, "grad_norm": 1.4183781098297552, "learning_rate": 8.056969053869821e-06, "loss": 1.5244, "step": 4551 }, { "epoch": 0.3483851216898821, "grad_norm": 0.6299957994834405, "learning_rate": 8.056118628956802e-06, "loss": 1.4906, "step": 4552 }, { "epoch": 0.34846165620694936, "grad_norm": 0.5450961677617601, "learning_rate": 8.055268077709708e-06, "loss": 1.375, "step": 4553 }, { "epoch": 0.34853819072401654, "grad_norm": 0.6005499293471487, "learning_rate": 8.054417400180799e-06, "loss": 1.4325, "step": 4554 }, { "epoch": 0.3486147252410837, "grad_norm": 0.6992137566847745, "learning_rate": 8.053566596422343e-06, "loss": 1.7287, "step": 4555 }, { "epoch": 0.3486912597581509, "grad_norm": 0.585902339750511, "learning_rate": 8.052715666486617e-06, "loss": 1.5357, "step": 4556 }, { "epoch": 0.34876779427521815, "grad_norm": 0.5588195653566573, "learning_rate": 8.051864610425904e-06, "loss": 1.3613, "step": 4557 }, { "epoch": 0.34884432879228533, "grad_norm": 0.5547988464473628, "learning_rate": 8.051013428292495e-06, "loss": 1.3892, "step": 4558 }, { "epoch": 0.3489208633093525, "grad_norm": 3.2215920905026456, "learning_rate": 8.05016212013869e-06, "loss": 1.549, "step": 4559 }, { "epoch": 0.3489973978264197, "grad_norm": 0.768274070545118, "learning_rate": 8.049310686016795e-06, "loss": 1.5294, "step": 4560 }, { "epoch": 0.34907393234348694, "grad_norm": 0.5512290687937142, "learning_rate": 8.048459125979124e-06, "loss": 1.2631, "step": 4561 }, { "epoch": 0.3491504668605541, "grad_norm": 0.6926319153939114, "learning_rate": 8.047607440078004e-06, "loss": 1.3605, "step": 4562 }, { "epoch": 0.3492270013776213, "grad_norm": 0.5612254408282774, "learning_rate": 8.046755628365758e-06, "loss": 1.4333, "step": 4563 }, { "epoch": 0.3493035358946885, "grad_norm": 0.8256351646720684, "learning_rate": 8.04590369089473e-06, "loss": 1.515, "step": 4564 }, { "epoch": 0.3493800704117557, "grad_norm": 0.5485237988888413, "learning_rate": 8.045051627717264e-06, "loss": 1.3476, "step": 4565 }, { "epoch": 0.3494566049288229, "grad_norm": 0.57099396136221, "learning_rate": 8.04419943888571e-06, "loss": 1.2822, "step": 4566 }, { "epoch": 0.3495331394458901, "grad_norm": 1.0109567997698288, "learning_rate": 8.043347124452432e-06, "loss": 1.4807, "step": 4567 }, { "epoch": 0.3496096739629573, "grad_norm": 0.6446874405381198, "learning_rate": 8.042494684469798e-06, "loss": 1.4139, "step": 4568 }, { "epoch": 0.3496862084800245, "grad_norm": 0.6329616984369493, "learning_rate": 8.041642118990185e-06, "loss": 1.4978, "step": 4569 }, { "epoch": 0.3497627429970917, "grad_norm": 0.6576536677006146, "learning_rate": 8.040789428065978e-06, "loss": 1.5816, "step": 4570 }, { "epoch": 0.3498392775141589, "grad_norm": 0.6591646216028326, "learning_rate": 8.039936611749567e-06, "loss": 1.6505, "step": 4571 }, { "epoch": 0.34991581203122607, "grad_norm": 0.7983299187984579, "learning_rate": 8.039083670093352e-06, "loss": 1.7533, "step": 4572 }, { "epoch": 0.34999234654829325, "grad_norm": 0.537945211028702, "learning_rate": 8.038230603149742e-06, "loss": 1.1963, "step": 4573 }, { "epoch": 0.3500688810653605, "grad_norm": 0.5914215225870461, "learning_rate": 8.037377410971149e-06, "loss": 1.4945, "step": 4574 }, { "epoch": 0.3501454155824277, "grad_norm": 0.60940624602412, "learning_rate": 8.036524093609996e-06, "loss": 1.5483, "step": 4575 }, { "epoch": 0.35022195009949486, "grad_norm": 0.6908076904606126, "learning_rate": 8.035670651118717e-06, "loss": 1.5962, "step": 4576 }, { "epoch": 0.35029848461656204, "grad_norm": 0.5263432921664273, "learning_rate": 8.034817083549747e-06, "loss": 1.2002, "step": 4577 }, { "epoch": 0.3503750191336293, "grad_norm": 0.8903736109696905, "learning_rate": 8.033963390955531e-06, "loss": 1.5474, "step": 4578 }, { "epoch": 0.35045155365069647, "grad_norm": 0.5824060899152493, "learning_rate": 8.033109573388527e-06, "loss": 1.2879, "step": 4579 }, { "epoch": 0.35052808816776365, "grad_norm": 0.5448793428960157, "learning_rate": 8.032255630901191e-06, "loss": 1.2906, "step": 4580 }, { "epoch": 0.35060462268483084, "grad_norm": 0.9604582172928762, "learning_rate": 8.031401563545994e-06, "loss": 1.6198, "step": 4581 }, { "epoch": 0.3506811572018981, "grad_norm": 0.5420459826216362, "learning_rate": 8.030547371375413e-06, "loss": 1.1527, "step": 4582 }, { "epoch": 0.35075769171896526, "grad_norm": 0.5951829494964826, "learning_rate": 8.02969305444193e-06, "loss": 1.4097, "step": 4583 }, { "epoch": 0.35083422623603244, "grad_norm": 0.7566368360839847, "learning_rate": 8.02883861279804e-06, "loss": 1.9638, "step": 4584 }, { "epoch": 0.3509107607530996, "grad_norm": 0.5149895271875178, "learning_rate": 8.027984046496238e-06, "loss": 1.1785, "step": 4585 }, { "epoch": 0.35098729527016687, "grad_norm": 0.7794112118729363, "learning_rate": 8.027129355589035e-06, "loss": 1.4675, "step": 4586 }, { "epoch": 0.35106382978723405, "grad_norm": 0.6575794541822798, "learning_rate": 8.026274540128945e-06, "loss": 1.4948, "step": 4587 }, { "epoch": 0.35114036430430123, "grad_norm": 0.6472367062300375, "learning_rate": 8.025419600168488e-06, "loss": 1.1706, "step": 4588 }, { "epoch": 0.3512168988213684, "grad_norm": 0.5992098746198853, "learning_rate": 8.024564535760197e-06, "loss": 1.5063, "step": 4589 }, { "epoch": 0.35129343333843566, "grad_norm": 0.6337084137603199, "learning_rate": 8.023709346956608e-06, "loss": 1.6022, "step": 4590 }, { "epoch": 0.35136996785550284, "grad_norm": 0.6013590904780636, "learning_rate": 8.022854033810269e-06, "loss": 1.4276, "step": 4591 }, { "epoch": 0.35144650237257, "grad_norm": 0.5936061677763983, "learning_rate": 8.02199859637373e-06, "loss": 1.4346, "step": 4592 }, { "epoch": 0.3515230368896372, "grad_norm": 0.6240313459395286, "learning_rate": 8.021143034699553e-06, "loss": 1.7518, "step": 4593 }, { "epoch": 0.35159957140670445, "grad_norm": 0.5162452967504222, "learning_rate": 8.020287348840304e-06, "loss": 1.222, "step": 4594 }, { "epoch": 0.35167610592377163, "grad_norm": 0.5581908675358259, "learning_rate": 8.019431538848562e-06, "loss": 1.3208, "step": 4595 }, { "epoch": 0.3517526404408388, "grad_norm": 0.5352010812295417, "learning_rate": 8.01857560477691e-06, "loss": 1.2331, "step": 4596 }, { "epoch": 0.351829174957906, "grad_norm": 0.5769024044243356, "learning_rate": 8.017719546677937e-06, "loss": 1.2868, "step": 4597 }, { "epoch": 0.35190570947497324, "grad_norm": 0.5501526381108875, "learning_rate": 8.016863364604244e-06, "loss": 1.2774, "step": 4598 }, { "epoch": 0.3519822439920404, "grad_norm": 0.5342935354013567, "learning_rate": 8.016007058608436e-06, "loss": 1.2691, "step": 4599 }, { "epoch": 0.3520587785091076, "grad_norm": 0.5224243960807163, "learning_rate": 8.015150628743128e-06, "loss": 1.2647, "step": 4600 }, { "epoch": 0.3521353130261748, "grad_norm": 0.5669518412976443, "learning_rate": 8.014294075060942e-06, "loss": 1.3417, "step": 4601 }, { "epoch": 0.35221184754324203, "grad_norm": 0.530032059328516, "learning_rate": 8.013437397614506e-06, "loss": 1.1759, "step": 4602 }, { "epoch": 0.3522883820603092, "grad_norm": 0.7550497635756291, "learning_rate": 8.012580596456456e-06, "loss": 1.892, "step": 4603 }, { "epoch": 0.3523649165773764, "grad_norm": 0.6329084340207933, "learning_rate": 8.01172367163944e-06, "loss": 1.5803, "step": 4604 }, { "epoch": 0.3524414510944436, "grad_norm": 0.5777974171162723, "learning_rate": 8.010866623216104e-06, "loss": 1.4894, "step": 4605 }, { "epoch": 0.35251798561151076, "grad_norm": 1.7575389964563217, "learning_rate": 8.010009451239116e-06, "loss": 1.5268, "step": 4606 }, { "epoch": 0.352594520128578, "grad_norm": 0.5794432529455594, "learning_rate": 8.009152155761136e-06, "loss": 1.3016, "step": 4607 }, { "epoch": 0.3526710546456452, "grad_norm": 0.5577973204833402, "learning_rate": 8.008294736834841e-06, "loss": 1.3143, "step": 4608 }, { "epoch": 0.35274758916271237, "grad_norm": 0.6248234542345835, "learning_rate": 8.007437194512916e-06, "loss": 1.4199, "step": 4609 }, { "epoch": 0.35282412367977956, "grad_norm": 0.827607509537118, "learning_rate": 8.006579528848046e-06, "loss": 1.7637, "step": 4610 }, { "epoch": 0.3529006581968468, "grad_norm": 0.5829526957639766, "learning_rate": 8.005721739892932e-06, "loss": 1.4987, "step": 4611 }, { "epoch": 0.352977192713914, "grad_norm": 0.60035229288542, "learning_rate": 8.004863827700278e-06, "loss": 1.5059, "step": 4612 }, { "epoch": 0.35305372723098116, "grad_norm": 0.6406193289603882, "learning_rate": 8.004005792322798e-06, "loss": 1.6779, "step": 4613 }, { "epoch": 0.35313026174804835, "grad_norm": 0.838249504082437, "learning_rate": 8.003147633813211e-06, "loss": 1.5633, "step": 4614 }, { "epoch": 0.3532067962651156, "grad_norm": 0.549064245665553, "learning_rate": 8.002289352224244e-06, "loss": 1.2838, "step": 4615 }, { "epoch": 0.35328333078218277, "grad_norm": 0.5420714376834709, "learning_rate": 8.001430947608635e-06, "loss": 1.1224, "step": 4616 }, { "epoch": 0.35335986529924995, "grad_norm": 0.718745722755514, "learning_rate": 8.000572420019126e-06, "loss": 1.4564, "step": 4617 }, { "epoch": 0.35343639981631714, "grad_norm": 0.5002782445675256, "learning_rate": 7.999713769508465e-06, "loss": 1.1223, "step": 4618 }, { "epoch": 0.3535129343333844, "grad_norm": 0.5690949500533329, "learning_rate": 7.998854996129411e-06, "loss": 1.3146, "step": 4619 }, { "epoch": 0.35358946885045156, "grad_norm": 0.6227672438200587, "learning_rate": 7.997996099934734e-06, "loss": 1.6223, "step": 4620 }, { "epoch": 0.35366600336751874, "grad_norm": 0.5810900038728576, "learning_rate": 7.997137080977201e-06, "loss": 1.3738, "step": 4621 }, { "epoch": 0.35374253788458593, "grad_norm": 0.7113256885872973, "learning_rate": 7.996277939309598e-06, "loss": 1.6966, "step": 4622 }, { "epoch": 0.35381907240165317, "grad_norm": 0.5349322339912221, "learning_rate": 7.99541867498471e-06, "loss": 1.2259, "step": 4623 }, { "epoch": 0.35389560691872035, "grad_norm": 0.6013039227903557, "learning_rate": 7.994559288055332e-06, "loss": 1.2933, "step": 4624 }, { "epoch": 0.35397214143578754, "grad_norm": 0.5713868548841374, "learning_rate": 7.99369977857427e-06, "loss": 1.2952, "step": 4625 }, { "epoch": 0.3540486759528547, "grad_norm": 0.6339640002404102, "learning_rate": 7.992840146594332e-06, "loss": 1.5517, "step": 4626 }, { "epoch": 0.35412521046992196, "grad_norm": 0.5559540549509414, "learning_rate": 7.99198039216834e-06, "loss": 1.4264, "step": 4627 }, { "epoch": 0.35420174498698914, "grad_norm": 0.5762592694459124, "learning_rate": 7.991120515349116e-06, "loss": 1.3158, "step": 4628 }, { "epoch": 0.3542782795040563, "grad_norm": 0.7540532028055488, "learning_rate": 7.990260516189496e-06, "loss": 1.734, "step": 4629 }, { "epoch": 0.3543548140211235, "grad_norm": 0.7795235763963249, "learning_rate": 7.98940039474232e-06, "loss": 1.4542, "step": 4630 }, { "epoch": 0.35443134853819075, "grad_norm": 0.5180997394247602, "learning_rate": 7.988540151060436e-06, "loss": 1.1383, "step": 4631 }, { "epoch": 0.35450788305525793, "grad_norm": 0.5806507780627546, "learning_rate": 7.987679785196702e-06, "loss": 1.4141, "step": 4632 }, { "epoch": 0.3545844175723251, "grad_norm": 0.6206764454849468, "learning_rate": 7.98681929720398e-06, "loss": 1.7547, "step": 4633 }, { "epoch": 0.3546609520893923, "grad_norm": 0.5900210960765991, "learning_rate": 7.985958687135139e-06, "loss": 1.4265, "step": 4634 }, { "epoch": 0.35473748660645954, "grad_norm": 0.6537294930816651, "learning_rate": 7.985097955043063e-06, "loss": 1.127, "step": 4635 }, { "epoch": 0.3548140211235267, "grad_norm": 0.6342859463210953, "learning_rate": 7.98423710098063e-06, "loss": 1.4426, "step": 4636 }, { "epoch": 0.3548905556405939, "grad_norm": 0.6427489656992456, "learning_rate": 7.98337612500074e-06, "loss": 1.533, "step": 4637 }, { "epoch": 0.3549670901576611, "grad_norm": 0.5568106531182047, "learning_rate": 7.982515027156291e-06, "loss": 1.3201, "step": 4638 }, { "epoch": 0.3550436246747283, "grad_norm": 0.6402745893100323, "learning_rate": 7.981653807500193e-06, "loss": 1.4449, "step": 4639 }, { "epoch": 0.3551201591917955, "grad_norm": 0.5876230248771879, "learning_rate": 7.98079246608536e-06, "loss": 1.3813, "step": 4640 }, { "epoch": 0.3551966937088627, "grad_norm": 0.5978007624620717, "learning_rate": 7.979931002964716e-06, "loss": 1.4011, "step": 4641 }, { "epoch": 0.3552732282259299, "grad_norm": 0.7099254316155325, "learning_rate": 7.979069418191194e-06, "loss": 1.4255, "step": 4642 }, { "epoch": 0.35534976274299707, "grad_norm": 0.5816090303670626, "learning_rate": 7.97820771181773e-06, "loss": 1.3704, "step": 4643 }, { "epoch": 0.3554262972600643, "grad_norm": 0.4986661498165043, "learning_rate": 7.97734588389727e-06, "loss": 1.1148, "step": 4644 }, { "epoch": 0.3555028317771315, "grad_norm": 0.6801518280092974, "learning_rate": 7.976483934482769e-06, "loss": 1.7422, "step": 4645 }, { "epoch": 0.3555793662941987, "grad_norm": 0.6047952721075208, "learning_rate": 7.975621863627186e-06, "loss": 1.5096, "step": 4646 }, { "epoch": 0.35565590081126586, "grad_norm": 0.5924680080172364, "learning_rate": 7.97475967138349e-06, "loss": 1.5451, "step": 4647 }, { "epoch": 0.3557324353283331, "grad_norm": 0.5471318270299353, "learning_rate": 7.973897357804656e-06, "loss": 1.3012, "step": 4648 }, { "epoch": 0.3558089698454003, "grad_norm": 0.6095812412788049, "learning_rate": 7.973034922943667e-06, "loss": 1.6095, "step": 4649 }, { "epoch": 0.35588550436246746, "grad_norm": 0.5773545464797901, "learning_rate": 7.972172366853513e-06, "loss": 1.3616, "step": 4650 }, { "epoch": 0.35596203887953465, "grad_norm": 0.6379433061338718, "learning_rate": 7.971309689587194e-06, "loss": 1.4834, "step": 4651 }, { "epoch": 0.3560385733966019, "grad_norm": 0.5924832692851764, "learning_rate": 7.970446891197716e-06, "loss": 1.3051, "step": 4652 }, { "epoch": 0.35611510791366907, "grad_norm": 0.49855481832295767, "learning_rate": 7.96958397173809e-06, "loss": 1.0924, "step": 4653 }, { "epoch": 0.35619164243073625, "grad_norm": 0.5673682745499832, "learning_rate": 7.968720931261336e-06, "loss": 1.2405, "step": 4654 }, { "epoch": 0.35626817694780344, "grad_norm": 0.6364066365299277, "learning_rate": 7.967857769820481e-06, "loss": 1.6864, "step": 4655 }, { "epoch": 0.3563447114648707, "grad_norm": 0.6418486549629238, "learning_rate": 7.966994487468566e-06, "loss": 1.6177, "step": 4656 }, { "epoch": 0.35642124598193786, "grad_norm": 0.43990973023440166, "learning_rate": 7.966131084258627e-06, "loss": 0.8673, "step": 4657 }, { "epoch": 0.35649778049900505, "grad_norm": 0.5721133266404038, "learning_rate": 7.965267560243718e-06, "loss": 1.4302, "step": 4658 }, { "epoch": 0.35657431501607223, "grad_norm": 0.739944289989043, "learning_rate": 7.964403915476893e-06, "loss": 1.6036, "step": 4659 }, { "epoch": 0.35665084953313947, "grad_norm": 0.5408157867846753, "learning_rate": 7.96354015001122e-06, "loss": 1.2956, "step": 4660 }, { "epoch": 0.35672738405020665, "grad_norm": 1.8816723075750692, "learning_rate": 7.962676263899771e-06, "loss": 1.9923, "step": 4661 }, { "epoch": 0.35680391856727384, "grad_norm": 0.7165438844409369, "learning_rate": 7.961812257195622e-06, "loss": 1.2961, "step": 4662 }, { "epoch": 0.356880453084341, "grad_norm": 0.6596093168320639, "learning_rate": 7.960948129951866e-06, "loss": 1.8626, "step": 4663 }, { "epoch": 0.35695698760140826, "grad_norm": 0.5545665169979221, "learning_rate": 7.960083882221595e-06, "loss": 1.2917, "step": 4664 }, { "epoch": 0.35703352211847544, "grad_norm": 0.7302848340514452, "learning_rate": 7.95921951405791e-06, "loss": 1.75, "step": 4665 }, { "epoch": 0.3571100566355426, "grad_norm": 0.585220773037137, "learning_rate": 7.95835502551392e-06, "loss": 1.4315, "step": 4666 }, { "epoch": 0.3571865911526098, "grad_norm": 0.5331517304860539, "learning_rate": 7.957490416642746e-06, "loss": 1.2607, "step": 4667 }, { "epoch": 0.35726312566967705, "grad_norm": 0.578887837220605, "learning_rate": 7.956625687497506e-06, "loss": 1.36, "step": 4668 }, { "epoch": 0.35733966018674423, "grad_norm": 0.627222670125089, "learning_rate": 7.955760838131337e-06, "loss": 1.4598, "step": 4669 }, { "epoch": 0.3574161947038114, "grad_norm": 0.5322941280567228, "learning_rate": 7.954895868597373e-06, "loss": 1.224, "step": 4670 }, { "epoch": 0.3574927292208786, "grad_norm": 0.5343898582868641, "learning_rate": 7.954030778948763e-06, "loss": 1.1931, "step": 4671 }, { "epoch": 0.3575692637379458, "grad_norm": 0.5977785436395953, "learning_rate": 7.953165569238663e-06, "loss": 1.2846, "step": 4672 }, { "epoch": 0.357645798255013, "grad_norm": 0.7349328278824584, "learning_rate": 7.952300239520228e-06, "loss": 1.4867, "step": 4673 }, { "epoch": 0.3577223327720802, "grad_norm": 1.5765838281305806, "learning_rate": 7.951434789846634e-06, "loss": 1.2805, "step": 4674 }, { "epoch": 0.3577988672891474, "grad_norm": 0.6309381519121894, "learning_rate": 7.95056922027105e-06, "loss": 1.3655, "step": 4675 }, { "epoch": 0.3578754018062146, "grad_norm": 0.5788199704925234, "learning_rate": 7.949703530846664e-06, "loss": 1.3392, "step": 4676 }, { "epoch": 0.3579519363232818, "grad_norm": 0.6755612510777124, "learning_rate": 7.948837721626662e-06, "loss": 1.7252, "step": 4677 }, { "epoch": 0.358028470840349, "grad_norm": 0.6034349779341365, "learning_rate": 7.947971792664247e-06, "loss": 1.3545, "step": 4678 }, { "epoch": 0.3581050053574162, "grad_norm": 0.6092303832999421, "learning_rate": 7.94710574401262e-06, "loss": 1.472, "step": 4679 }, { "epoch": 0.35818153987448337, "grad_norm": 0.6644200725238735, "learning_rate": 7.946239575724996e-06, "loss": 1.5644, "step": 4680 }, { "epoch": 0.3582580743915506, "grad_norm": 0.5921774302907515, "learning_rate": 7.945373287854596e-06, "loss": 1.2394, "step": 4681 }, { "epoch": 0.3583346089086178, "grad_norm": 0.5387414095416699, "learning_rate": 7.944506880454643e-06, "loss": 1.1594, "step": 4682 }, { "epoch": 0.358411143425685, "grad_norm": 0.6158464089111838, "learning_rate": 7.943640353578375e-06, "loss": 1.4107, "step": 4683 }, { "epoch": 0.35848767794275216, "grad_norm": 0.7379658021131874, "learning_rate": 7.942773707279034e-06, "loss": 1.4323, "step": 4684 }, { "epoch": 0.3585642124598194, "grad_norm": 0.5877706537944694, "learning_rate": 7.941906941609867e-06, "loss": 1.5071, "step": 4685 }, { "epoch": 0.3586407469768866, "grad_norm": 0.511747532697888, "learning_rate": 7.941040056624132e-06, "loss": 1.1651, "step": 4686 }, { "epoch": 0.35871728149395377, "grad_norm": 0.5716382814939137, "learning_rate": 7.940173052375097e-06, "loss": 1.3049, "step": 4687 }, { "epoch": 0.35879381601102095, "grad_norm": 0.5921873612410011, "learning_rate": 7.939305928916026e-06, "loss": 1.2605, "step": 4688 }, { "epoch": 0.3588703505280882, "grad_norm": 0.5493210153846991, "learning_rate": 7.938438686300205e-06, "loss": 1.3615, "step": 4689 }, { "epoch": 0.3589468850451554, "grad_norm": 0.64774161351238, "learning_rate": 7.937571324580912e-06, "loss": 1.3188, "step": 4690 }, { "epoch": 0.35902341956222256, "grad_norm": 0.5711613609469628, "learning_rate": 7.936703843811447e-06, "loss": 1.4784, "step": 4691 }, { "epoch": 0.35909995407928974, "grad_norm": 0.679660986019893, "learning_rate": 7.935836244045108e-06, "loss": 1.6497, "step": 4692 }, { "epoch": 0.359176488596357, "grad_norm": 0.43820947640260194, "learning_rate": 7.934968525335203e-06, "loss": 0.8022, "step": 4693 }, { "epoch": 0.35925302311342416, "grad_norm": 1.0189348919407186, "learning_rate": 7.934100687735046e-06, "loss": 1.4047, "step": 4694 }, { "epoch": 0.35932955763049135, "grad_norm": 0.6913120188154415, "learning_rate": 7.933232731297963e-06, "loss": 1.5526, "step": 4695 }, { "epoch": 0.35940609214755853, "grad_norm": 0.554165246507593, "learning_rate": 7.932364656077281e-06, "loss": 1.2587, "step": 4696 }, { "epoch": 0.35948262666462577, "grad_norm": 0.5667170831889984, "learning_rate": 7.931496462126339e-06, "loss": 1.4004, "step": 4697 }, { "epoch": 0.35955916118169295, "grad_norm": 0.6021410398524648, "learning_rate": 7.930628149498477e-06, "loss": 1.5811, "step": 4698 }, { "epoch": 0.35963569569876014, "grad_norm": 0.5748894281722425, "learning_rate": 7.929759718247052e-06, "loss": 1.2602, "step": 4699 }, { "epoch": 0.3597122302158273, "grad_norm": 0.6255549949125282, "learning_rate": 7.928891168425422e-06, "loss": 1.4905, "step": 4700 }, { "epoch": 0.35978876473289456, "grad_norm": 0.4858244206039789, "learning_rate": 7.92802250008695e-06, "loss": 1.0632, "step": 4701 }, { "epoch": 0.35986529924996175, "grad_norm": 0.5462716068466189, "learning_rate": 7.927153713285014e-06, "loss": 1.1796, "step": 4702 }, { "epoch": 0.35994183376702893, "grad_norm": 0.8482210026132302, "learning_rate": 7.926284808072993e-06, "loss": 1.5222, "step": 4703 }, { "epoch": 0.3600183682840961, "grad_norm": 0.5745210918566814, "learning_rate": 7.925415784504273e-06, "loss": 1.3709, "step": 4704 }, { "epoch": 0.36009490280116335, "grad_norm": 0.7972884559537244, "learning_rate": 7.924546642632253e-06, "loss": 1.4808, "step": 4705 }, { "epoch": 0.36017143731823054, "grad_norm": 0.6738385040577565, "learning_rate": 7.923677382510333e-06, "loss": 1.708, "step": 4706 }, { "epoch": 0.3602479718352977, "grad_norm": 0.5451181673254454, "learning_rate": 7.922808004191924e-06, "loss": 1.3336, "step": 4707 }, { "epoch": 0.3603245063523649, "grad_norm": 0.567920272935716, "learning_rate": 7.921938507730443e-06, "loss": 1.3148, "step": 4708 }, { "epoch": 0.3604010408694321, "grad_norm": 0.6849367623237527, "learning_rate": 7.921068893179314e-06, "loss": 1.7075, "step": 4709 }, { "epoch": 0.3604775753864993, "grad_norm": 1.1088521712911832, "learning_rate": 7.920199160591971e-06, "loss": 1.4929, "step": 4710 }, { "epoch": 0.3605541099035665, "grad_norm": 0.5999307388441445, "learning_rate": 7.919329310021851e-06, "loss": 1.1894, "step": 4711 }, { "epoch": 0.3606306444206337, "grad_norm": 0.5712509822274449, "learning_rate": 7.9184593415224e-06, "loss": 1.3376, "step": 4712 }, { "epoch": 0.3607071789377009, "grad_norm": 0.6913300495517649, "learning_rate": 7.917589255147071e-06, "loss": 1.4763, "step": 4713 }, { "epoch": 0.3607837134547681, "grad_norm": 0.5703540795675461, "learning_rate": 7.916719050949327e-06, "loss": 1.3095, "step": 4714 }, { "epoch": 0.3608602479718353, "grad_norm": 0.5376980533122022, "learning_rate": 7.915848728982634e-06, "loss": 1.1349, "step": 4715 }, { "epoch": 0.3609367824889025, "grad_norm": 0.6037056845651733, "learning_rate": 7.914978289300469e-06, "loss": 1.3734, "step": 4716 }, { "epoch": 0.36101331700596967, "grad_norm": 0.6592879185794586, "learning_rate": 7.914107731956312e-06, "loss": 1.7232, "step": 4717 }, { "epoch": 0.3610898515230369, "grad_norm": 0.6127420086162553, "learning_rate": 7.913237057003652e-06, "loss": 1.5311, "step": 4718 }, { "epoch": 0.3611663860401041, "grad_norm": 0.5252273617172278, "learning_rate": 7.912366264495991e-06, "loss": 1.1692, "step": 4719 }, { "epoch": 0.3612429205571713, "grad_norm": 0.681339109511771, "learning_rate": 7.911495354486827e-06, "loss": 1.5718, "step": 4720 }, { "epoch": 0.36131945507423846, "grad_norm": 0.5754843582692638, "learning_rate": 7.910624327029677e-06, "loss": 1.4108, "step": 4721 }, { "epoch": 0.3613959895913057, "grad_norm": 0.45695073690225285, "learning_rate": 7.909753182178055e-06, "loss": 0.8147, "step": 4722 }, { "epoch": 0.3614725241083729, "grad_norm": 0.5981046898600475, "learning_rate": 7.908881919985489e-06, "loss": 1.3965, "step": 4723 }, { "epoch": 0.36154905862544007, "grad_norm": 0.5344074576734087, "learning_rate": 7.908010540505512e-06, "loss": 1.3257, "step": 4724 }, { "epoch": 0.36162559314250725, "grad_norm": 0.5536032846554516, "learning_rate": 7.907139043791662e-06, "loss": 1.3002, "step": 4725 }, { "epoch": 0.3617021276595745, "grad_norm": 0.7077137528433624, "learning_rate": 7.90626742989749e-06, "loss": 1.6228, "step": 4726 }, { "epoch": 0.3617786621766417, "grad_norm": 0.677500478437719, "learning_rate": 7.905395698876546e-06, "loss": 1.5336, "step": 4727 }, { "epoch": 0.36185519669370886, "grad_norm": 0.7411305399482371, "learning_rate": 7.904523850782394e-06, "loss": 1.6321, "step": 4728 }, { "epoch": 0.36193173121077604, "grad_norm": 0.5973371390490583, "learning_rate": 7.903651885668605e-06, "loss": 1.4017, "step": 4729 }, { "epoch": 0.3620082657278433, "grad_norm": 0.610910437287761, "learning_rate": 7.902779803588754e-06, "loss": 1.5941, "step": 4730 }, { "epoch": 0.36208480024491047, "grad_norm": 0.7217993513161303, "learning_rate": 7.901907604596421e-06, "loss": 1.5544, "step": 4731 }, { "epoch": 0.36216133476197765, "grad_norm": 0.7953936117046878, "learning_rate": 7.901035288745202e-06, "loss": 1.5903, "step": 4732 }, { "epoch": 0.36223786927904483, "grad_norm": 0.6339755777615569, "learning_rate": 7.900162856088689e-06, "loss": 1.6649, "step": 4733 }, { "epoch": 0.3623144037961121, "grad_norm": 0.6213107242716628, "learning_rate": 7.899290306680491e-06, "loss": 1.5157, "step": 4734 }, { "epoch": 0.36239093831317926, "grad_norm": 0.6527668454754555, "learning_rate": 7.898417640574218e-06, "loss": 1.7464, "step": 4735 }, { "epoch": 0.36246747283024644, "grad_norm": 0.7829208581501801, "learning_rate": 7.897544857823492e-06, "loss": 1.7534, "step": 4736 }, { "epoch": 0.3625440073473136, "grad_norm": 0.5799305142636949, "learning_rate": 7.896671958481937e-06, "loss": 1.0331, "step": 4737 }, { "epoch": 0.36262054186438086, "grad_norm": 0.8217348996679632, "learning_rate": 7.895798942603186e-06, "loss": 1.8092, "step": 4738 }, { "epoch": 0.36269707638144805, "grad_norm": 0.5357964273058518, "learning_rate": 7.89492581024088e-06, "loss": 1.3656, "step": 4739 }, { "epoch": 0.36277361089851523, "grad_norm": 0.6876043030933894, "learning_rate": 7.894052561448667e-06, "loss": 1.6673, "step": 4740 }, { "epoch": 0.3628501454155824, "grad_norm": 0.568888386402701, "learning_rate": 7.893179196280206e-06, "loss": 1.4805, "step": 4741 }, { "epoch": 0.3629266799326496, "grad_norm": 0.4818241154869549, "learning_rate": 7.892305714789152e-06, "loss": 1.0706, "step": 4742 }, { "epoch": 0.36300321444971684, "grad_norm": 0.5486781161927415, "learning_rate": 7.891432117029178e-06, "loss": 1.2306, "step": 4743 }, { "epoch": 0.363079748966784, "grad_norm": 0.7063567141933463, "learning_rate": 7.890558403053962e-06, "loss": 1.4595, "step": 4744 }, { "epoch": 0.3631562834838512, "grad_norm": 0.6181430780357701, "learning_rate": 7.889684572917186e-06, "loss": 1.2706, "step": 4745 }, { "epoch": 0.3632328180009184, "grad_norm": 0.7201174014134847, "learning_rate": 7.88881062667254e-06, "loss": 1.6718, "step": 4746 }, { "epoch": 0.36330935251798563, "grad_norm": 0.6565653503398605, "learning_rate": 7.887936564373722e-06, "loss": 1.5547, "step": 4747 }, { "epoch": 0.3633858870350528, "grad_norm": 1.0448148667435215, "learning_rate": 7.887062386074441e-06, "loss": 1.4237, "step": 4748 }, { "epoch": 0.36346242155212, "grad_norm": 0.7342541894070632, "learning_rate": 7.886188091828403e-06, "loss": 1.5268, "step": 4749 }, { "epoch": 0.3635389560691872, "grad_norm": 0.6010445747218485, "learning_rate": 7.88531368168933e-06, "loss": 1.4368, "step": 4750 }, { "epoch": 0.3636154905862544, "grad_norm": 0.5745301599885998, "learning_rate": 7.88443915571095e-06, "loss": 1.3123, "step": 4751 }, { "epoch": 0.3636920251033216, "grad_norm": 0.6164994511402557, "learning_rate": 7.883564513946994e-06, "loss": 1.5323, "step": 4752 }, { "epoch": 0.3637685596203888, "grad_norm": 1.0014501654394454, "learning_rate": 7.882689756451201e-06, "loss": 1.6051, "step": 4753 }, { "epoch": 0.36384509413745597, "grad_norm": 0.5830846402408307, "learning_rate": 7.881814883277324e-06, "loss": 1.4411, "step": 4754 }, { "epoch": 0.3639216286545232, "grad_norm": 0.5542348930832005, "learning_rate": 7.880939894479114e-06, "loss": 1.1597, "step": 4755 }, { "epoch": 0.3639981631715904, "grad_norm": 0.6137979571975414, "learning_rate": 7.880064790110335e-06, "loss": 1.4194, "step": 4756 }, { "epoch": 0.3640746976886576, "grad_norm": 0.5338672843393113, "learning_rate": 7.879189570224754e-06, "loss": 1.2073, "step": 4757 }, { "epoch": 0.36415123220572476, "grad_norm": 0.5293817823149427, "learning_rate": 7.878314234876148e-06, "loss": 1.1913, "step": 4758 }, { "epoch": 0.364227766722792, "grad_norm": 0.6214012794882665, "learning_rate": 7.8774387841183e-06, "loss": 1.5008, "step": 4759 }, { "epoch": 0.3643043012398592, "grad_norm": 0.7002265296622503, "learning_rate": 7.876563218005004e-06, "loss": 1.396, "step": 4760 }, { "epoch": 0.36438083575692637, "grad_norm": 0.5691521356768767, "learning_rate": 7.87568753659005e-06, "loss": 1.2896, "step": 4761 }, { "epoch": 0.36445737027399355, "grad_norm": 0.5954177532850893, "learning_rate": 7.874811739927249e-06, "loss": 1.3546, "step": 4762 }, { "epoch": 0.3645339047910608, "grad_norm": 0.47332190105542715, "learning_rate": 7.873935828070409e-06, "loss": 0.9929, "step": 4763 }, { "epoch": 0.364610439308128, "grad_norm": 0.5048861800839671, "learning_rate": 7.873059801073348e-06, "loss": 1.2488, "step": 4764 }, { "epoch": 0.36468697382519516, "grad_norm": 0.9010231759396862, "learning_rate": 7.872183658989896e-06, "loss": 1.5101, "step": 4765 }, { "epoch": 0.36476350834226234, "grad_norm": 0.5749432503321135, "learning_rate": 7.871307401873882e-06, "loss": 1.4138, "step": 4766 }, { "epoch": 0.3648400428593296, "grad_norm": 0.5415979760263646, "learning_rate": 7.870431029779146e-06, "loss": 1.1216, "step": 4767 }, { "epoch": 0.36491657737639677, "grad_norm": 0.5089618526368864, "learning_rate": 7.869554542759538e-06, "loss": 1.116, "step": 4768 }, { "epoch": 0.36499311189346395, "grad_norm": 0.7494152956501201, "learning_rate": 7.86867794086891e-06, "loss": 1.4383, "step": 4769 }, { "epoch": 0.36506964641053113, "grad_norm": 0.6215907016552624, "learning_rate": 7.867801224161124e-06, "loss": 1.599, "step": 4770 }, { "epoch": 0.3651461809275984, "grad_norm": 0.5685597587284603, "learning_rate": 7.866924392690046e-06, "loss": 1.3995, "step": 4771 }, { "epoch": 0.36522271544466556, "grad_norm": 1.040339807337726, "learning_rate": 7.86604744650955e-06, "loss": 1.5178, "step": 4772 }, { "epoch": 0.36529924996173274, "grad_norm": 0.6261725314396652, "learning_rate": 7.865170385673523e-06, "loss": 1.6192, "step": 4773 }, { "epoch": 0.3653757844787999, "grad_norm": 0.5409339143829848, "learning_rate": 7.86429321023585e-06, "loss": 1.1096, "step": 4774 }, { "epoch": 0.3654523189958671, "grad_norm": 0.5956024444719499, "learning_rate": 7.86341592025043e-06, "loss": 1.3119, "step": 4775 }, { "epoch": 0.36552885351293435, "grad_norm": 0.5694873924961701, "learning_rate": 7.862538515771166e-06, "loss": 1.2752, "step": 4776 }, { "epoch": 0.36560538803000153, "grad_norm": 0.7366483304112917, "learning_rate": 7.861660996851968e-06, "loss": 1.4772, "step": 4777 }, { "epoch": 0.3656819225470687, "grad_norm": 0.6333190207396899, "learning_rate": 7.86078336354675e-06, "loss": 1.4656, "step": 4778 }, { "epoch": 0.3657584570641359, "grad_norm": 0.6437074429088906, "learning_rate": 7.859905615909443e-06, "loss": 1.3665, "step": 4779 }, { "epoch": 0.36583499158120314, "grad_norm": 0.7254738699359372, "learning_rate": 7.859027753993974e-06, "loss": 1.817, "step": 4780 }, { "epoch": 0.3659115260982703, "grad_norm": 0.5717917256270649, "learning_rate": 7.858149777854282e-06, "loss": 1.387, "step": 4781 }, { "epoch": 0.3659880606153375, "grad_norm": 0.6453196498231429, "learning_rate": 7.857271687544311e-06, "loss": 1.6395, "step": 4782 }, { "epoch": 0.3660645951324047, "grad_norm": 0.563037960634818, "learning_rate": 7.856393483118017e-06, "loss": 1.3735, "step": 4783 }, { "epoch": 0.36614112964947193, "grad_norm": 0.9398191356759183, "learning_rate": 7.855515164629358e-06, "loss": 1.755, "step": 4784 }, { "epoch": 0.3662176641665391, "grad_norm": 0.6146875168171672, "learning_rate": 7.8546367321323e-06, "loss": 1.5419, "step": 4785 }, { "epoch": 0.3662941986836063, "grad_norm": 0.5639984424350772, "learning_rate": 7.853758185680816e-06, "loss": 1.4261, "step": 4786 }, { "epoch": 0.3663707332006735, "grad_norm": 0.5748705167431233, "learning_rate": 7.852879525328888e-06, "loss": 1.3569, "step": 4787 }, { "epoch": 0.3664472677177407, "grad_norm": 0.6119308676968653, "learning_rate": 7.852000751130503e-06, "loss": 1.5342, "step": 4788 }, { "epoch": 0.3665238022348079, "grad_norm": 0.5943775467583597, "learning_rate": 7.851121863139657e-06, "loss": 1.2538, "step": 4789 }, { "epoch": 0.3666003367518751, "grad_norm": 0.8675108234770882, "learning_rate": 7.850242861410349e-06, "loss": 1.3376, "step": 4790 }, { "epoch": 0.3666768712689423, "grad_norm": 1.2438713728639323, "learning_rate": 7.849363745996587e-06, "loss": 1.3125, "step": 4791 }, { "epoch": 0.3667534057860095, "grad_norm": 0.5759627688758004, "learning_rate": 7.848484516952389e-06, "loss": 1.2483, "step": 4792 }, { "epoch": 0.3668299403030767, "grad_norm": 0.9647703339583987, "learning_rate": 7.847605174331777e-06, "loss": 2.0317, "step": 4793 }, { "epoch": 0.3669064748201439, "grad_norm": 0.5602535165863944, "learning_rate": 7.84672571818878e-06, "loss": 1.2484, "step": 4794 }, { "epoch": 0.36698300933721106, "grad_norm": 0.6677985906411783, "learning_rate": 7.845846148577432e-06, "loss": 1.6841, "step": 4795 }, { "epoch": 0.3670595438542783, "grad_norm": 0.5802692124419174, "learning_rate": 7.84496646555178e-06, "loss": 1.4472, "step": 4796 }, { "epoch": 0.3671360783713455, "grad_norm": 0.5763462815982433, "learning_rate": 7.844086669165874e-06, "loss": 1.4094, "step": 4797 }, { "epoch": 0.36721261288841267, "grad_norm": 0.5926770409028673, "learning_rate": 7.843206759473769e-06, "loss": 1.2963, "step": 4798 }, { "epoch": 0.36728914740547985, "grad_norm": 0.5522157629898856, "learning_rate": 7.842326736529531e-06, "loss": 1.3572, "step": 4799 }, { "epoch": 0.3673656819225471, "grad_norm": 0.5640881582003294, "learning_rate": 7.84144660038723e-06, "loss": 1.4118, "step": 4800 }, { "epoch": 0.3674422164396143, "grad_norm": 0.5436681387066917, "learning_rate": 7.840566351100946e-06, "loss": 1.3039, "step": 4801 }, { "epoch": 0.36751875095668146, "grad_norm": 0.6083425809966077, "learning_rate": 7.83968598872476e-06, "loss": 1.3003, "step": 4802 }, { "epoch": 0.36759528547374865, "grad_norm": 0.9473156845714571, "learning_rate": 7.838805513312771e-06, "loss": 1.565, "step": 4803 }, { "epoch": 0.3676718199908159, "grad_norm": 0.7847466952134838, "learning_rate": 7.837924924919074e-06, "loss": 1.4835, "step": 4804 }, { "epoch": 0.36774835450788307, "grad_norm": 0.5608753129750014, "learning_rate": 7.837044223597774e-06, "loss": 1.4219, "step": 4805 }, { "epoch": 0.36782488902495025, "grad_norm": 0.6174092807788324, "learning_rate": 7.836163409402987e-06, "loss": 1.4749, "step": 4806 }, { "epoch": 0.36790142354201744, "grad_norm": 0.5712419357570374, "learning_rate": 7.83528248238883e-06, "loss": 1.3174, "step": 4807 }, { "epoch": 0.3679779580590846, "grad_norm": 0.6671855903745258, "learning_rate": 7.83440144260943e-06, "loss": 1.6382, "step": 4808 }, { "epoch": 0.36805449257615186, "grad_norm": 0.538560836803926, "learning_rate": 7.833520290118922e-06, "loss": 1.2958, "step": 4809 }, { "epoch": 0.36813102709321904, "grad_norm": 0.8828974650844033, "learning_rate": 7.832639024971446e-06, "loss": 1.5664, "step": 4810 }, { "epoch": 0.3682075616102862, "grad_norm": 0.6600707544699272, "learning_rate": 7.83175764722115e-06, "loss": 1.4437, "step": 4811 }, { "epoch": 0.3682840961273534, "grad_norm": 0.5511054189894925, "learning_rate": 7.830876156922189e-06, "loss": 1.4117, "step": 4812 }, { "epoch": 0.36836063064442065, "grad_norm": 0.5762499142662469, "learning_rate": 7.829994554128722e-06, "loss": 1.4212, "step": 4813 }, { "epoch": 0.36843716516148783, "grad_norm": 0.5877819376630283, "learning_rate": 7.82911283889492e-06, "loss": 1.3102, "step": 4814 }, { "epoch": 0.368513699678555, "grad_norm": 0.8277692904193189, "learning_rate": 7.828231011274956e-06, "loss": 1.4948, "step": 4815 }, { "epoch": 0.3685902341956222, "grad_norm": 0.6717277674498638, "learning_rate": 7.827349071323015e-06, "loss": 1.6901, "step": 4816 }, { "epoch": 0.36866676871268944, "grad_norm": 0.724537958277017, "learning_rate": 7.826467019093283e-06, "loss": 1.642, "step": 4817 }, { "epoch": 0.3687433032297566, "grad_norm": 0.5884415827676867, "learning_rate": 7.825584854639958e-06, "loss": 1.4089, "step": 4818 }, { "epoch": 0.3688198377468238, "grad_norm": 0.61090780343541, "learning_rate": 7.824702578017242e-06, "loss": 1.4673, "step": 4819 }, { "epoch": 0.368896372263891, "grad_norm": 0.6064579220445522, "learning_rate": 7.823820189279345e-06, "loss": 1.388, "step": 4820 }, { "epoch": 0.36897290678095823, "grad_norm": 0.6180475195297751, "learning_rate": 7.822937688480482e-06, "loss": 1.4131, "step": 4821 }, { "epoch": 0.3690494412980254, "grad_norm": 0.5610193282781842, "learning_rate": 7.82205507567488e-06, "loss": 1.2241, "step": 4822 }, { "epoch": 0.3691259758150926, "grad_norm": 0.5401264576343578, "learning_rate": 7.821172350916765e-06, "loss": 1.2952, "step": 4823 }, { "epoch": 0.3692025103321598, "grad_norm": 0.5921858748411617, "learning_rate": 7.820289514260379e-06, "loss": 1.407, "step": 4824 }, { "epoch": 0.369279044849227, "grad_norm": 0.6060730340156997, "learning_rate": 7.819406565759961e-06, "loss": 1.3993, "step": 4825 }, { "epoch": 0.3693555793662942, "grad_norm": 0.5734490595606613, "learning_rate": 7.818523505469767e-06, "loss": 1.3899, "step": 4826 }, { "epoch": 0.3694321138833614, "grad_norm": 0.535536616790757, "learning_rate": 7.817640333444053e-06, "loss": 1.3151, "step": 4827 }, { "epoch": 0.3695086484004286, "grad_norm": 0.5558523029731974, "learning_rate": 7.81675704973708e-06, "loss": 1.0769, "step": 4828 }, { "epoch": 0.3695851829174958, "grad_norm": 0.606244827156823, "learning_rate": 7.815873654403125e-06, "loss": 1.4315, "step": 4829 }, { "epoch": 0.369661717434563, "grad_norm": 0.7379332285500483, "learning_rate": 7.814990147496463e-06, "loss": 1.7024, "step": 4830 }, { "epoch": 0.3697382519516302, "grad_norm": 0.49499153279908575, "learning_rate": 7.814106529071382e-06, "loss": 1.176, "step": 4831 }, { "epoch": 0.36981478646869737, "grad_norm": 0.6854554202249805, "learning_rate": 7.81322279918217e-06, "loss": 1.6673, "step": 4832 }, { "epoch": 0.3698913209857646, "grad_norm": 0.6245594952106666, "learning_rate": 7.812338957883131e-06, "loss": 1.5738, "step": 4833 }, { "epoch": 0.3699678555028318, "grad_norm": 0.6316511540081402, "learning_rate": 7.81145500522857e-06, "loss": 1.2223, "step": 4834 }, { "epoch": 0.37004439001989897, "grad_norm": 0.6207386050976916, "learning_rate": 7.810570941272795e-06, "loss": 1.4857, "step": 4835 }, { "epoch": 0.37012092453696616, "grad_norm": 0.6279348515013272, "learning_rate": 7.80968676607013e-06, "loss": 1.483, "step": 4836 }, { "epoch": 0.3701974590540334, "grad_norm": 0.5419013615841852, "learning_rate": 7.808802479674901e-06, "loss": 1.2018, "step": 4837 }, { "epoch": 0.3702739935711006, "grad_norm": 0.6648282907280701, "learning_rate": 7.80791808214144e-06, "loss": 1.6201, "step": 4838 }, { "epoch": 0.37035052808816776, "grad_norm": 1.238535550958747, "learning_rate": 7.807033573524088e-06, "loss": 1.5274, "step": 4839 }, { "epoch": 0.37042706260523495, "grad_norm": 0.6414112273145837, "learning_rate": 7.806148953877191e-06, "loss": 1.6579, "step": 4840 }, { "epoch": 0.37050359712230213, "grad_norm": 0.5537398993896905, "learning_rate": 7.805264223255101e-06, "loss": 1.3534, "step": 4841 }, { "epoch": 0.37058013163936937, "grad_norm": 0.643891781839087, "learning_rate": 7.804379381712183e-06, "loss": 1.3481, "step": 4842 }, { "epoch": 0.37065666615643655, "grad_norm": 0.6128749568347845, "learning_rate": 7.803494429302799e-06, "loss": 1.5446, "step": 4843 }, { "epoch": 0.37073320067350374, "grad_norm": 1.115177025507727, "learning_rate": 7.802609366081328e-06, "loss": 1.4779, "step": 4844 }, { "epoch": 0.3708097351905709, "grad_norm": 0.6542064823904158, "learning_rate": 7.801724192102149e-06, "loss": 1.4689, "step": 4845 }, { "epoch": 0.37088626970763816, "grad_norm": 0.6197596954520012, "learning_rate": 7.800838907419646e-06, "loss": 1.5818, "step": 4846 }, { "epoch": 0.37096280422470534, "grad_norm": 4.03309129666847, "learning_rate": 7.799953512088222e-06, "loss": 1.9754, "step": 4847 }, { "epoch": 0.37103933874177253, "grad_norm": 0.7929761963788585, "learning_rate": 7.799068006162268e-06, "loss": 1.5891, "step": 4848 }, { "epoch": 0.3711158732588397, "grad_norm": 0.46660720712088144, "learning_rate": 7.798182389696203e-06, "loss": 1.0147, "step": 4849 }, { "epoch": 0.37119240777590695, "grad_norm": 0.7395311643097331, "learning_rate": 7.797296662744434e-06, "loss": 1.8246, "step": 4850 }, { "epoch": 0.37126894229297414, "grad_norm": 0.7216616518778117, "learning_rate": 7.796410825361385e-06, "loss": 1.2861, "step": 4851 }, { "epoch": 0.3713454768100413, "grad_norm": 0.5634069816083797, "learning_rate": 7.795524877601483e-06, "loss": 1.1364, "step": 4852 }, { "epoch": 0.3714220113271085, "grad_norm": 0.6501406197950111, "learning_rate": 7.794638819519168e-06, "loss": 1.5193, "step": 4853 }, { "epoch": 0.37149854584417574, "grad_norm": 0.5096289209706536, "learning_rate": 7.793752651168877e-06, "loss": 1.2862, "step": 4854 }, { "epoch": 0.3715750803612429, "grad_norm": 0.6380631932130887, "learning_rate": 7.792866372605062e-06, "loss": 1.5882, "step": 4855 }, { "epoch": 0.3716516148783101, "grad_norm": 0.6059633986878217, "learning_rate": 7.791979983882176e-06, "loss": 1.4427, "step": 4856 }, { "epoch": 0.3717281493953773, "grad_norm": 0.7627316838718433, "learning_rate": 7.791093485054684e-06, "loss": 1.1606, "step": 4857 }, { "epoch": 0.37180468391244453, "grad_norm": 0.6692227764365382, "learning_rate": 7.790206876177054e-06, "loss": 1.4206, "step": 4858 }, { "epoch": 0.3718812184295117, "grad_norm": 0.6031229618931273, "learning_rate": 7.78932015730376e-06, "loss": 1.2103, "step": 4859 }, { "epoch": 0.3719577529465789, "grad_norm": 1.3313076759575104, "learning_rate": 7.788433328489289e-06, "loss": 1.3905, "step": 4860 }, { "epoch": 0.3720342874636461, "grad_norm": 0.5803377312706306, "learning_rate": 7.787546389788127e-06, "loss": 1.4605, "step": 4861 }, { "epoch": 0.3721108219807133, "grad_norm": 0.5222224413355298, "learning_rate": 7.78665934125477e-06, "loss": 0.9877, "step": 4862 }, { "epoch": 0.3721873564977805, "grad_norm": 0.6109741924802707, "learning_rate": 7.785772182943724e-06, "loss": 1.3715, "step": 4863 }, { "epoch": 0.3722638910148477, "grad_norm": 0.5997501019469547, "learning_rate": 7.784884914909496e-06, "loss": 1.4592, "step": 4864 }, { "epoch": 0.3723404255319149, "grad_norm": 0.612901099078711, "learning_rate": 7.783997537206602e-06, "loss": 1.2738, "step": 4865 }, { "epoch": 0.3724169600489821, "grad_norm": 0.5863312897458758, "learning_rate": 7.783110049889565e-06, "loss": 1.5681, "step": 4866 }, { "epoch": 0.3724934945660493, "grad_norm": 0.723273744238535, "learning_rate": 7.782222453012919e-06, "loss": 1.5483, "step": 4867 }, { "epoch": 0.3725700290831165, "grad_norm": 0.5397181460378533, "learning_rate": 7.781334746631195e-06, "loss": 1.1582, "step": 4868 }, { "epoch": 0.37264656360018367, "grad_norm": 0.6399038615020081, "learning_rate": 7.78044693079894e-06, "loss": 1.5521, "step": 4869 }, { "epoch": 0.3727230981172509, "grad_norm": 1.5735761991014103, "learning_rate": 7.779559005570706e-06, "loss": 1.4256, "step": 4870 }, { "epoch": 0.3727996326343181, "grad_norm": 0.5650843678385749, "learning_rate": 7.778670971001044e-06, "loss": 1.208, "step": 4871 }, { "epoch": 0.3728761671513853, "grad_norm": 0.9103779242629839, "learning_rate": 7.777782827144522e-06, "loss": 2.0423, "step": 4872 }, { "epoch": 0.37295270166845246, "grad_norm": 0.5837002669200506, "learning_rate": 7.776894574055706e-06, "loss": 1.2971, "step": 4873 }, { "epoch": 0.3730292361855197, "grad_norm": 0.5616681854170924, "learning_rate": 7.776006211789175e-06, "loss": 1.2284, "step": 4874 }, { "epoch": 0.3731057707025869, "grad_norm": 0.5673801011632706, "learning_rate": 7.775117740399517e-06, "loss": 1.3505, "step": 4875 }, { "epoch": 0.37318230521965406, "grad_norm": 0.5584073663386799, "learning_rate": 7.774229159941315e-06, "loss": 1.2567, "step": 4876 }, { "epoch": 0.37325883973672125, "grad_norm": 1.1370237867636386, "learning_rate": 7.77334047046917e-06, "loss": 1.6857, "step": 4877 }, { "epoch": 0.37333537425378843, "grad_norm": 0.5388815817259665, "learning_rate": 7.772451672037687e-06, "loss": 1.1936, "step": 4878 }, { "epoch": 0.37341190877085567, "grad_norm": 0.5994962266555232, "learning_rate": 7.771562764701471e-06, "loss": 1.3071, "step": 4879 }, { "epoch": 0.37348844328792286, "grad_norm": 0.661343008819144, "learning_rate": 7.770673748515146e-06, "loss": 1.6012, "step": 4880 }, { "epoch": 0.37356497780499004, "grad_norm": 0.8587835472814479, "learning_rate": 7.769784623533332e-06, "loss": 1.6534, "step": 4881 }, { "epoch": 0.3736415123220572, "grad_norm": 0.62750505425789, "learning_rate": 7.768895389810662e-06, "loss": 1.4004, "step": 4882 }, { "epoch": 0.37371804683912446, "grad_norm": 0.575462165069656, "learning_rate": 7.768006047401768e-06, "loss": 1.4223, "step": 4883 }, { "epoch": 0.37379458135619165, "grad_norm": 1.119365069565556, "learning_rate": 7.767116596361298e-06, "loss": 1.6633, "step": 4884 }, { "epoch": 0.37387111587325883, "grad_norm": 0.6021312601797917, "learning_rate": 7.766227036743902e-06, "loss": 1.3744, "step": 4885 }, { "epoch": 0.373947650390326, "grad_norm": 0.6363296017677926, "learning_rate": 7.765337368604237e-06, "loss": 1.5757, "step": 4886 }, { "epoch": 0.37402418490739325, "grad_norm": 2.588969697251701, "learning_rate": 7.764447591996966e-06, "loss": 1.3879, "step": 4887 }, { "epoch": 0.37410071942446044, "grad_norm": 0.6116826539004633, "learning_rate": 7.763557706976761e-06, "loss": 1.2033, "step": 4888 }, { "epoch": 0.3741772539415276, "grad_norm": 0.5743741740186509, "learning_rate": 7.762667713598298e-06, "loss": 1.3465, "step": 4889 }, { "epoch": 0.3742537884585948, "grad_norm": 0.5620559864354033, "learning_rate": 7.761777611916262e-06, "loss": 1.4148, "step": 4890 }, { "epoch": 0.37433032297566204, "grad_norm": 0.5709649030800459, "learning_rate": 7.760887401985342e-06, "loss": 1.3045, "step": 4891 }, { "epoch": 0.37440685749272923, "grad_norm": 0.7469783530084907, "learning_rate": 7.759997083860236e-06, "loss": 1.4188, "step": 4892 }, { "epoch": 0.3744833920097964, "grad_norm": 0.6618170078831629, "learning_rate": 7.759106657595648e-06, "loss": 1.7831, "step": 4893 }, { "epoch": 0.3745599265268636, "grad_norm": 0.53316593876029, "learning_rate": 7.758216123246289e-06, "loss": 1.307, "step": 4894 }, { "epoch": 0.37463646104393084, "grad_norm": 0.7958914085892328, "learning_rate": 7.757325480866873e-06, "loss": 1.3589, "step": 4895 }, { "epoch": 0.374712995560998, "grad_norm": 0.6198955155533079, "learning_rate": 7.75643473051213e-06, "loss": 1.5614, "step": 4896 }, { "epoch": 0.3747895300780652, "grad_norm": 0.5716283669477151, "learning_rate": 7.755543872236783e-06, "loss": 1.3854, "step": 4897 }, { "epoch": 0.3748660645951324, "grad_norm": 0.5243061288580441, "learning_rate": 7.754652906095573e-06, "loss": 1.4484, "step": 4898 }, { "epoch": 0.3749425991121996, "grad_norm": 0.6727461947338975, "learning_rate": 7.753761832143243e-06, "loss": 1.6623, "step": 4899 }, { "epoch": 0.3750191336292668, "grad_norm": 0.633466832656928, "learning_rate": 7.752870650434545e-06, "loss": 1.5171, "step": 4900 }, { "epoch": 0.375095668146334, "grad_norm": 0.5509302438060202, "learning_rate": 7.751979361024232e-06, "loss": 1.0396, "step": 4901 }, { "epoch": 0.3751722026634012, "grad_norm": 0.7036185925628426, "learning_rate": 7.751087963967071e-06, "loss": 1.6439, "step": 4902 }, { "epoch": 0.3752487371804684, "grad_norm": 0.6077488340709192, "learning_rate": 7.750196459317832e-06, "loss": 1.3489, "step": 4903 }, { "epoch": 0.3753252716975356, "grad_norm": 0.7406786683982107, "learning_rate": 7.749304847131288e-06, "loss": 1.4061, "step": 4904 }, { "epoch": 0.3754018062146028, "grad_norm": 0.5877552396859503, "learning_rate": 7.748413127462225e-06, "loss": 1.2913, "step": 4905 }, { "epoch": 0.37547834073166997, "grad_norm": 0.7821024056018516, "learning_rate": 7.747521300365434e-06, "loss": 1.2684, "step": 4906 }, { "epoch": 0.3755548752487372, "grad_norm": 0.6268203144517814, "learning_rate": 7.74662936589571e-06, "loss": 1.5135, "step": 4907 }, { "epoch": 0.3756314097658044, "grad_norm": 0.7958123293976533, "learning_rate": 7.745737324107854e-06, "loss": 1.6864, "step": 4908 }, { "epoch": 0.3757079442828716, "grad_norm": 0.49486747110590756, "learning_rate": 7.74484517505668e-06, "loss": 1.1018, "step": 4909 }, { "epoch": 0.37578447879993876, "grad_norm": 0.6160135611551575, "learning_rate": 7.743952918797003e-06, "loss": 1.3908, "step": 4910 }, { "epoch": 0.37586101331700594, "grad_norm": 0.6858550539679918, "learning_rate": 7.743060555383645e-06, "loss": 1.2896, "step": 4911 }, { "epoch": 0.3759375478340732, "grad_norm": 0.9678406260936653, "learning_rate": 7.742168084871433e-06, "loss": 1.1685, "step": 4912 }, { "epoch": 0.37601408235114037, "grad_norm": 0.6321208519240451, "learning_rate": 7.74127550731521e-06, "loss": 1.2294, "step": 4913 }, { "epoch": 0.37609061686820755, "grad_norm": 0.5719799719552272, "learning_rate": 7.74038282276981e-06, "loss": 1.4503, "step": 4914 }, { "epoch": 0.37616715138527473, "grad_norm": 0.6055738002927734, "learning_rate": 7.739490031290092e-06, "loss": 1.3044, "step": 4915 }, { "epoch": 0.376243685902342, "grad_norm": 1.0594054904170376, "learning_rate": 7.738597132930904e-06, "loss": 1.2682, "step": 4916 }, { "epoch": 0.37632022041940916, "grad_norm": 0.5106387202602971, "learning_rate": 7.737704127747108e-06, "loss": 1.1818, "step": 4917 }, { "epoch": 0.37639675493647634, "grad_norm": 0.5175869830790341, "learning_rate": 7.736811015793578e-06, "loss": 1.1216, "step": 4918 }, { "epoch": 0.3764732894535435, "grad_norm": 0.6019548159917161, "learning_rate": 7.735917797125186e-06, "loss": 1.4109, "step": 4919 }, { "epoch": 0.37654982397061076, "grad_norm": 0.5851040741251161, "learning_rate": 7.735024471796817e-06, "loss": 1.3612, "step": 4920 }, { "epoch": 0.37662635848767795, "grad_norm": 0.5677158853629914, "learning_rate": 7.734131039863357e-06, "loss": 1.2673, "step": 4921 }, { "epoch": 0.37670289300474513, "grad_norm": 0.56832147170831, "learning_rate": 7.733237501379701e-06, "loss": 1.2617, "step": 4922 }, { "epoch": 0.3767794275218123, "grad_norm": 0.6956993540426127, "learning_rate": 7.732343856400752e-06, "loss": 1.7196, "step": 4923 }, { "epoch": 0.37685596203887956, "grad_norm": 0.5632366913826666, "learning_rate": 7.731450104981418e-06, "loss": 1.3384, "step": 4924 }, { "epoch": 0.37693249655594674, "grad_norm": 0.5959589828283223, "learning_rate": 7.730556247176613e-06, "loss": 1.3538, "step": 4925 }, { "epoch": 0.3770090310730139, "grad_norm": 0.5486549610210459, "learning_rate": 7.729662283041258e-06, "loss": 1.1217, "step": 4926 }, { "epoch": 0.3770855655900811, "grad_norm": 0.6528143487075639, "learning_rate": 7.728768212630283e-06, "loss": 1.5898, "step": 4927 }, { "epoch": 0.37716210010714835, "grad_norm": 0.6432685568709319, "learning_rate": 7.727874035998622e-06, "loss": 1.6222, "step": 4928 }, { "epoch": 0.37723863462421553, "grad_norm": 0.5695565337548538, "learning_rate": 7.726979753201212e-06, "loss": 1.2595, "step": 4929 }, { "epoch": 0.3773151691412827, "grad_norm": 0.6053843683458686, "learning_rate": 7.726085364293007e-06, "loss": 1.4209, "step": 4930 }, { "epoch": 0.3773917036583499, "grad_norm": 0.5806027565252061, "learning_rate": 7.725190869328953e-06, "loss": 1.2568, "step": 4931 }, { "epoch": 0.37746823817541714, "grad_norm": 0.5630371909077694, "learning_rate": 7.724296268364018e-06, "loss": 1.3431, "step": 4932 }, { "epoch": 0.3775447726924843, "grad_norm": 0.5336205431476875, "learning_rate": 7.723401561453164e-06, "loss": 1.1804, "step": 4933 }, { "epoch": 0.3776213072095515, "grad_norm": 0.527242910735322, "learning_rate": 7.722506748651365e-06, "loss": 1.1804, "step": 4934 }, { "epoch": 0.3776978417266187, "grad_norm": 0.6098422562852279, "learning_rate": 7.721611830013605e-06, "loss": 1.1909, "step": 4935 }, { "epoch": 0.3777743762436859, "grad_norm": 0.5575619380649628, "learning_rate": 7.720716805594865e-06, "loss": 1.4048, "step": 4936 }, { "epoch": 0.3778509107607531, "grad_norm": 0.598815822306624, "learning_rate": 7.719821675450141e-06, "loss": 1.3821, "step": 4937 }, { "epoch": 0.3779274452778203, "grad_norm": 0.5834569526400695, "learning_rate": 7.718926439634433e-06, "loss": 1.5107, "step": 4938 }, { "epoch": 0.3780039797948875, "grad_norm": 0.5742500577950698, "learning_rate": 7.718031098202746e-06, "loss": 1.3647, "step": 4939 }, { "epoch": 0.3780805143119547, "grad_norm": 0.6076934925644377, "learning_rate": 7.717135651210092e-06, "loss": 1.453, "step": 4940 }, { "epoch": 0.3781570488290219, "grad_norm": 0.7871650758102527, "learning_rate": 7.71624009871149e-06, "loss": 1.574, "step": 4941 }, { "epoch": 0.3782335833460891, "grad_norm": 0.5644082665196857, "learning_rate": 7.715344440761965e-06, "loss": 1.454, "step": 4942 }, { "epoch": 0.37831011786315627, "grad_norm": 0.5468429183682476, "learning_rate": 7.714448677416551e-06, "loss": 1.1737, "step": 4943 }, { "epoch": 0.37838665238022345, "grad_norm": 0.6441587306943667, "learning_rate": 7.713552808730284e-06, "loss": 1.559, "step": 4944 }, { "epoch": 0.3784631868972907, "grad_norm": 0.5003784554080554, "learning_rate": 7.712656834758211e-06, "loss": 1.1732, "step": 4945 }, { "epoch": 0.3785397214143579, "grad_norm": 0.6169969712551361, "learning_rate": 7.711760755555381e-06, "loss": 1.4563, "step": 4946 }, { "epoch": 0.37861625593142506, "grad_norm": 0.6273499592092371, "learning_rate": 7.710864571176855e-06, "loss": 1.4294, "step": 4947 }, { "epoch": 0.37869279044849224, "grad_norm": 0.5711924328061054, "learning_rate": 7.709968281677693e-06, "loss": 1.4418, "step": 4948 }, { "epoch": 0.3787693249655595, "grad_norm": 0.5659484161088577, "learning_rate": 7.709071887112967e-06, "loss": 1.2446, "step": 4949 }, { "epoch": 0.37884585948262667, "grad_norm": 0.5227667034543074, "learning_rate": 7.708175387537758e-06, "loss": 1.2066, "step": 4950 }, { "epoch": 0.37892239399969385, "grad_norm": 1.037499932646326, "learning_rate": 7.707278783007146e-06, "loss": 1.6417, "step": 4951 }, { "epoch": 0.37899892851676104, "grad_norm": 0.582392979774567, "learning_rate": 7.706382073576221e-06, "loss": 1.5304, "step": 4952 }, { "epoch": 0.3790754630338283, "grad_norm": 0.5262730463184845, "learning_rate": 7.70548525930008e-06, "loss": 1.1633, "step": 4953 }, { "epoch": 0.37915199755089546, "grad_norm": 0.596516892074566, "learning_rate": 7.704588340233829e-06, "loss": 1.5411, "step": 4954 }, { "epoch": 0.37922853206796264, "grad_norm": 0.5668443048710603, "learning_rate": 7.703691316432572e-06, "loss": 1.4071, "step": 4955 }, { "epoch": 0.3793050665850298, "grad_norm": 0.5597466365199734, "learning_rate": 7.702794187951428e-06, "loss": 1.3079, "step": 4956 }, { "epoch": 0.37938160110209707, "grad_norm": 0.8034361117170699, "learning_rate": 7.701896954845517e-06, "loss": 1.4318, "step": 4957 }, { "epoch": 0.37945813561916425, "grad_norm": 0.7086845869385402, "learning_rate": 7.700999617169973e-06, "loss": 1.6878, "step": 4958 }, { "epoch": 0.37953467013623143, "grad_norm": 0.6514402889560378, "learning_rate": 7.700102174979924e-06, "loss": 1.6076, "step": 4959 }, { "epoch": 0.3796112046532986, "grad_norm": 1.0530218125506865, "learning_rate": 7.699204628330516e-06, "loss": 1.6824, "step": 4960 }, { "epoch": 0.37968773917036586, "grad_norm": 0.5316307946766323, "learning_rate": 7.698306977276897e-06, "loss": 1.3664, "step": 4961 }, { "epoch": 0.37976427368743304, "grad_norm": 0.8465414407267023, "learning_rate": 7.69740922187422e-06, "loss": 1.4456, "step": 4962 }, { "epoch": 0.3798408082045002, "grad_norm": 0.5296615687683024, "learning_rate": 7.696511362177648e-06, "loss": 1.2208, "step": 4963 }, { "epoch": 0.3799173427215674, "grad_norm": 1.1716188879468659, "learning_rate": 7.695613398242344e-06, "loss": 1.5008, "step": 4964 }, { "epoch": 0.37999387723863465, "grad_norm": 0.6399647274480657, "learning_rate": 7.694715330123485e-06, "loss": 1.4814, "step": 4965 }, { "epoch": 0.38007041175570183, "grad_norm": 0.8434691516373235, "learning_rate": 7.693817157876249e-06, "loss": 1.7042, "step": 4966 }, { "epoch": 0.380146946272769, "grad_norm": 0.6798777722038319, "learning_rate": 7.692918881555821e-06, "loss": 1.4744, "step": 4967 }, { "epoch": 0.3802234807898362, "grad_norm": 0.6286416254753125, "learning_rate": 7.6920205012174e-06, "loss": 1.4038, "step": 4968 }, { "epoch": 0.38030001530690344, "grad_norm": 0.5010949907970175, "learning_rate": 7.69112201691618e-06, "loss": 1.1217, "step": 4969 }, { "epoch": 0.3803765498239706, "grad_norm": 0.5837395362996393, "learning_rate": 7.690223428707367e-06, "loss": 1.2115, "step": 4970 }, { "epoch": 0.3804530843410378, "grad_norm": 0.6363707752285326, "learning_rate": 7.689324736646174e-06, "loss": 1.5148, "step": 4971 }, { "epoch": 0.380529618858105, "grad_norm": 0.5899777559556061, "learning_rate": 7.68842594078782e-06, "loss": 1.4105, "step": 4972 }, { "epoch": 0.38060615337517223, "grad_norm": 0.6703402693153653, "learning_rate": 7.687527041187529e-06, "loss": 1.7273, "step": 4973 }, { "epoch": 0.3806826878922394, "grad_norm": 0.6044862963679543, "learning_rate": 7.68662803790053e-06, "loss": 1.4869, "step": 4974 }, { "epoch": 0.3807592224093066, "grad_norm": 0.5433175962104037, "learning_rate": 7.685728930982062e-06, "loss": 1.2872, "step": 4975 }, { "epoch": 0.3808357569263738, "grad_norm": 0.750431304500732, "learning_rate": 7.68482972048737e-06, "loss": 1.7413, "step": 4976 }, { "epoch": 0.38091229144344096, "grad_norm": 1.2954054296567423, "learning_rate": 7.683930406471704e-06, "loss": 1.5847, "step": 4977 }, { "epoch": 0.3809888259605082, "grad_norm": 0.5584245441984937, "learning_rate": 7.68303098899032e-06, "loss": 1.3077, "step": 4978 }, { "epoch": 0.3810653604775754, "grad_norm": 0.6914478370156727, "learning_rate": 7.68213146809848e-06, "loss": 1.1687, "step": 4979 }, { "epoch": 0.38114189499464257, "grad_norm": 0.5198905346435145, "learning_rate": 7.681231843851453e-06, "loss": 1.2616, "step": 4980 }, { "epoch": 0.38121842951170976, "grad_norm": 0.5500136019507083, "learning_rate": 7.680332116304518e-06, "loss": 1.2888, "step": 4981 }, { "epoch": 0.381294964028777, "grad_norm": 0.8908573487857249, "learning_rate": 7.679432285512952e-06, "loss": 1.7936, "step": 4982 }, { "epoch": 0.3813714985458442, "grad_norm": 0.91448431611132, "learning_rate": 7.678532351532048e-06, "loss": 1.9884, "step": 4983 }, { "epoch": 0.38144803306291136, "grad_norm": 0.5308680841938406, "learning_rate": 7.677632314417099e-06, "loss": 0.9838, "step": 4984 }, { "epoch": 0.38152456757997855, "grad_norm": 0.7972783543157956, "learning_rate": 7.676732174223403e-06, "loss": 1.5654, "step": 4985 }, { "epoch": 0.3816011020970458, "grad_norm": 0.6350330321381704, "learning_rate": 7.67583193100627e-06, "loss": 1.3515, "step": 4986 }, { "epoch": 0.38167763661411297, "grad_norm": 0.6387116493108785, "learning_rate": 7.674931584821013e-06, "loss": 1.4595, "step": 4987 }, { "epoch": 0.38175417113118015, "grad_norm": 0.5953850142304599, "learning_rate": 7.674031135722955e-06, "loss": 1.4125, "step": 4988 }, { "epoch": 0.38183070564824734, "grad_norm": 0.7494206098269555, "learning_rate": 7.673130583767418e-06, "loss": 1.5823, "step": 4989 }, { "epoch": 0.3819072401653146, "grad_norm": 0.5703415171050178, "learning_rate": 7.672229929009736e-06, "loss": 1.4778, "step": 4990 }, { "epoch": 0.38198377468238176, "grad_norm": 0.626722112245391, "learning_rate": 7.671329171505247e-06, "loss": 1.5739, "step": 4991 }, { "epoch": 0.38206030919944894, "grad_norm": 0.5873781755627923, "learning_rate": 7.670428311309301e-06, "loss": 1.4192, "step": 4992 }, { "epoch": 0.38213684371651613, "grad_norm": 0.5005089046044028, "learning_rate": 7.669527348477243e-06, "loss": 1.0959, "step": 4993 }, { "epoch": 0.38221337823358337, "grad_norm": 0.6256509277424293, "learning_rate": 7.668626283064434e-06, "loss": 1.3141, "step": 4994 }, { "epoch": 0.38228991275065055, "grad_norm": 0.5698090111070346, "learning_rate": 7.66772511512624e-06, "loss": 1.3127, "step": 4995 }, { "epoch": 0.38236644726771774, "grad_norm": 2.2830059772904, "learning_rate": 7.666823844718027e-06, "loss": 1.8368, "step": 4996 }, { "epoch": 0.3824429817847849, "grad_norm": 0.6682117029023593, "learning_rate": 7.665922471895175e-06, "loss": 1.4109, "step": 4997 }, { "epoch": 0.38251951630185216, "grad_norm": 1.1242368403637317, "learning_rate": 7.665020996713066e-06, "loss": 1.6186, "step": 4998 }, { "epoch": 0.38259605081891934, "grad_norm": 0.5861882615890258, "learning_rate": 7.66411941922709e-06, "loss": 1.3146, "step": 4999 }, { "epoch": 0.3826725853359865, "grad_norm": 0.6003555825742756, "learning_rate": 7.663217739492641e-06, "loss": 1.4744, "step": 5000 }, { "epoch": 0.3826725853359865, "eval_loss": 1.4269700050354004, "eval_runtime": 648.6616, "eval_samples_per_second": 8.482, "eval_steps_per_second": 1.061, "step": 5000 }, { "epoch": 0.3827491198530537, "grad_norm": 0.7247933315148549, "learning_rate": 7.662315957565123e-06, "loss": 1.7803, "step": 5001 }, { "epoch": 0.38282565437012095, "grad_norm": 0.7006992420788688, "learning_rate": 7.661414073499941e-06, "loss": 1.6405, "step": 5002 }, { "epoch": 0.38290218888718813, "grad_norm": 0.45934005580442855, "learning_rate": 7.660512087352516e-06, "loss": 0.9627, "step": 5003 }, { "epoch": 0.3829787234042553, "grad_norm": 0.7287951795465065, "learning_rate": 7.65960999917826e-06, "loss": 1.5536, "step": 5004 }, { "epoch": 0.3830552579213225, "grad_norm": 0.6145667827564746, "learning_rate": 7.658707809032607e-06, "loss": 1.5004, "step": 5005 }, { "epoch": 0.38313179243838974, "grad_norm": 0.5608578142892952, "learning_rate": 7.657805516970988e-06, "loss": 1.3941, "step": 5006 }, { "epoch": 0.3832083269554569, "grad_norm": 0.5318024744085449, "learning_rate": 7.656903123048842e-06, "loss": 1.2599, "step": 5007 }, { "epoch": 0.3832848614725241, "grad_norm": 0.7578608951881622, "learning_rate": 7.656000627321612e-06, "loss": 1.4472, "step": 5008 }, { "epoch": 0.3833613959895913, "grad_norm": 0.5870190559664206, "learning_rate": 7.655098029844755e-06, "loss": 1.4169, "step": 5009 }, { "epoch": 0.3834379305066585, "grad_norm": 0.6204238249301542, "learning_rate": 7.654195330673728e-06, "loss": 1.4682, "step": 5010 }, { "epoch": 0.3835144650237257, "grad_norm": 0.6530854992053557, "learning_rate": 7.653292529863994e-06, "loss": 1.6162, "step": 5011 }, { "epoch": 0.3835909995407929, "grad_norm": 0.6416445247799858, "learning_rate": 7.652389627471024e-06, "loss": 1.751, "step": 5012 }, { "epoch": 0.3836675340578601, "grad_norm": 0.6107301809835058, "learning_rate": 7.651486623550297e-06, "loss": 1.6065, "step": 5013 }, { "epoch": 0.38374406857492727, "grad_norm": 0.6527888736976354, "learning_rate": 7.650583518157292e-06, "loss": 1.5294, "step": 5014 }, { "epoch": 0.3838206030919945, "grad_norm": 0.5425068274658765, "learning_rate": 7.649680311347503e-06, "loss": 1.2797, "step": 5015 }, { "epoch": 0.3838971376090617, "grad_norm": 0.608419660375568, "learning_rate": 7.648777003176425e-06, "loss": 1.4731, "step": 5016 }, { "epoch": 0.3839736721261289, "grad_norm": 0.5935956418739633, "learning_rate": 7.647873593699558e-06, "loss": 1.3404, "step": 5017 }, { "epoch": 0.38405020664319606, "grad_norm": 0.7475670108182539, "learning_rate": 7.646970082972413e-06, "loss": 1.3346, "step": 5018 }, { "epoch": 0.3841267411602633, "grad_norm": 0.5176456231308401, "learning_rate": 7.6460664710505e-06, "loss": 1.037, "step": 5019 }, { "epoch": 0.3842032756773305, "grad_norm": 0.5785614143161221, "learning_rate": 7.645162757989344e-06, "loss": 1.4381, "step": 5020 }, { "epoch": 0.38427981019439766, "grad_norm": 0.592028142793141, "learning_rate": 7.64425894384447e-06, "loss": 1.4477, "step": 5021 }, { "epoch": 0.38435634471146485, "grad_norm": 0.6273986637469342, "learning_rate": 7.64335502867141e-06, "loss": 1.5807, "step": 5022 }, { "epoch": 0.3844328792285321, "grad_norm": 0.5796821651708938, "learning_rate": 7.642451012525704e-06, "loss": 1.4101, "step": 5023 }, { "epoch": 0.38450941374559927, "grad_norm": 0.5548922251040354, "learning_rate": 7.641546895462898e-06, "loss": 1.4251, "step": 5024 }, { "epoch": 0.38458594826266645, "grad_norm": 1.6011428544910822, "learning_rate": 7.640642677538545e-06, "loss": 1.4813, "step": 5025 }, { "epoch": 0.38466248277973364, "grad_norm": 0.5866504839665764, "learning_rate": 7.639738358808201e-06, "loss": 1.2148, "step": 5026 }, { "epoch": 0.3847390172968009, "grad_norm": 0.490509487309416, "learning_rate": 7.63883393932743e-06, "loss": 1.0923, "step": 5027 }, { "epoch": 0.38481555181386806, "grad_norm": 0.4772041349628946, "learning_rate": 7.637929419151802e-06, "loss": 0.9665, "step": 5028 }, { "epoch": 0.38489208633093525, "grad_norm": 0.6106260866458456, "learning_rate": 7.637024798336895e-06, "loss": 1.3467, "step": 5029 }, { "epoch": 0.38496862084800243, "grad_norm": 0.5878202456270065, "learning_rate": 7.636120076938289e-06, "loss": 1.3471, "step": 5030 }, { "epoch": 0.38504515536506967, "grad_norm": 0.6151964202274854, "learning_rate": 7.635215255011575e-06, "loss": 1.5233, "step": 5031 }, { "epoch": 0.38512168988213685, "grad_norm": 0.5116105477186516, "learning_rate": 7.634310332612349e-06, "loss": 1.2498, "step": 5032 }, { "epoch": 0.38519822439920404, "grad_norm": 0.6835402222721243, "learning_rate": 7.633405309796208e-06, "loss": 1.689, "step": 5033 }, { "epoch": 0.3852747589162712, "grad_norm": 0.6158417808321074, "learning_rate": 7.632500186618765e-06, "loss": 1.6178, "step": 5034 }, { "epoch": 0.38535129343333846, "grad_norm": 0.6110330919286672, "learning_rate": 7.631594963135626e-06, "loss": 1.4763, "step": 5035 }, { "epoch": 0.38542782795040564, "grad_norm": 0.649722710854674, "learning_rate": 7.630689639402418e-06, "loss": 1.6994, "step": 5036 }, { "epoch": 0.3855043624674728, "grad_norm": 0.5156066342889135, "learning_rate": 7.629784215474762e-06, "loss": 1.2454, "step": 5037 }, { "epoch": 0.38558089698454, "grad_norm": 0.5345738493765858, "learning_rate": 7.628878691408293e-06, "loss": 1.4118, "step": 5038 }, { "epoch": 0.38565743150160725, "grad_norm": 0.5849409839587703, "learning_rate": 7.627973067258649e-06, "loss": 1.4869, "step": 5039 }, { "epoch": 0.38573396601867443, "grad_norm": 1.3145915009811806, "learning_rate": 7.62706734308147e-06, "loss": 1.4731, "step": 5040 }, { "epoch": 0.3858105005357416, "grad_norm": 0.6520939824390601, "learning_rate": 7.6261615189324125e-06, "loss": 1.4662, "step": 5041 }, { "epoch": 0.3858870350528088, "grad_norm": 0.9928869985698432, "learning_rate": 7.625255594867128e-06, "loss": 1.4422, "step": 5042 }, { "epoch": 0.38596356956987604, "grad_norm": 0.5929649525687383, "learning_rate": 7.624349570941281e-06, "loss": 1.36, "step": 5043 }, { "epoch": 0.3860401040869432, "grad_norm": 0.5898922065793087, "learning_rate": 7.623443447210541e-06, "loss": 1.4773, "step": 5044 }, { "epoch": 0.3861166386040104, "grad_norm": 0.59505454672134, "learning_rate": 7.622537223730583e-06, "loss": 1.2753, "step": 5045 }, { "epoch": 0.3861931731210776, "grad_norm": 0.6009734736693428, "learning_rate": 7.621630900557087e-06, "loss": 1.561, "step": 5046 }, { "epoch": 0.3862697076381448, "grad_norm": 0.6969650260261115, "learning_rate": 7.62072447774574e-06, "loss": 1.6244, "step": 5047 }, { "epoch": 0.386346242155212, "grad_norm": 0.5925602153034206, "learning_rate": 7.619817955352236e-06, "loss": 1.4663, "step": 5048 }, { "epoch": 0.3864227766722792, "grad_norm": 0.6577181401967659, "learning_rate": 7.618911333432276e-06, "loss": 1.4979, "step": 5049 }, { "epoch": 0.3864993111893464, "grad_norm": 0.5537340242016666, "learning_rate": 7.618004612041564e-06, "loss": 1.2205, "step": 5050 }, { "epoch": 0.38657584570641357, "grad_norm": 0.6009797008476194, "learning_rate": 7.617097791235811e-06, "loss": 1.5445, "step": 5051 }, { "epoch": 0.3866523802234808, "grad_norm": 0.5200251965039023, "learning_rate": 7.616190871070735e-06, "loss": 1.091, "step": 5052 }, { "epoch": 0.386728914740548, "grad_norm": 0.5966761451863043, "learning_rate": 7.615283851602062e-06, "loss": 1.5655, "step": 5053 }, { "epoch": 0.3868054492576152, "grad_norm": 0.664741196079674, "learning_rate": 7.614376732885518e-06, "loss": 1.7812, "step": 5054 }, { "epoch": 0.38688198377468236, "grad_norm": 0.6070654801889597, "learning_rate": 7.6134695149768435e-06, "loss": 1.3968, "step": 5055 }, { "epoch": 0.3869585182917496, "grad_norm": 0.539493500885738, "learning_rate": 7.6125621979317775e-06, "loss": 1.1886, "step": 5056 }, { "epoch": 0.3870350528088168, "grad_norm": 0.5940032349383654, "learning_rate": 7.611654781806069e-06, "loss": 1.487, "step": 5057 }, { "epoch": 0.38711158732588397, "grad_norm": 1.6653223750766613, "learning_rate": 7.610747266655474e-06, "loss": 1.8424, "step": 5058 }, { "epoch": 0.38718812184295115, "grad_norm": 0.47247671011881653, "learning_rate": 7.609839652535752e-06, "loss": 1.0034, "step": 5059 }, { "epoch": 0.3872646563600184, "grad_norm": 0.5311680314975635, "learning_rate": 7.608931939502668e-06, "loss": 1.2131, "step": 5060 }, { "epoch": 0.3873411908770856, "grad_norm": 0.6143832682917902, "learning_rate": 7.608024127611996e-06, "loss": 1.425, "step": 5061 }, { "epoch": 0.38741772539415276, "grad_norm": 1.0245759917705506, "learning_rate": 7.607116216919516e-06, "loss": 1.5116, "step": 5062 }, { "epoch": 0.38749425991121994, "grad_norm": 0.5862655127949926, "learning_rate": 7.6062082074810105e-06, "loss": 1.3548, "step": 5063 }, { "epoch": 0.3875707944282872, "grad_norm": 2.118313563095749, "learning_rate": 7.605300099352271e-06, "loss": 1.6781, "step": 5064 }, { "epoch": 0.38764732894535436, "grad_norm": 0.5268590134928409, "learning_rate": 7.604391892589092e-06, "loss": 1.0482, "step": 5065 }, { "epoch": 0.38772386346242155, "grad_norm": 0.6319225919544444, "learning_rate": 7.603483587247282e-06, "loss": 1.5373, "step": 5066 }, { "epoch": 0.38780039797948873, "grad_norm": 0.5679294661100708, "learning_rate": 7.6025751833826464e-06, "loss": 1.2526, "step": 5067 }, { "epoch": 0.38787693249655597, "grad_norm": 0.5765303040291889, "learning_rate": 7.601666681051e-06, "loss": 1.3246, "step": 5068 }, { "epoch": 0.38795346701362315, "grad_norm": 0.5564034604538566, "learning_rate": 7.600758080308165e-06, "loss": 1.2832, "step": 5069 }, { "epoch": 0.38803000153069034, "grad_norm": 0.5575143475544713, "learning_rate": 7.599849381209968e-06, "loss": 1.2481, "step": 5070 }, { "epoch": 0.3881065360477575, "grad_norm": 0.7205114723572332, "learning_rate": 7.5989405838122425e-06, "loss": 1.549, "step": 5071 }, { "epoch": 0.38818307056482476, "grad_norm": 0.6380554155440604, "learning_rate": 7.5980316881708284e-06, "loss": 1.4845, "step": 5072 }, { "epoch": 0.38825960508189195, "grad_norm": 0.6142246440669831, "learning_rate": 7.597122694341569e-06, "loss": 1.65, "step": 5073 }, { "epoch": 0.38833613959895913, "grad_norm": 0.637518972896849, "learning_rate": 7.596213602380318e-06, "loss": 1.5437, "step": 5074 }, { "epoch": 0.3884126741160263, "grad_norm": 0.5252662292416448, "learning_rate": 7.595304412342931e-06, "loss": 1.2378, "step": 5075 }, { "epoch": 0.38848920863309355, "grad_norm": 0.5491503144629166, "learning_rate": 7.594395124285271e-06, "loss": 1.2789, "step": 5076 }, { "epoch": 0.38856574315016074, "grad_norm": 0.6358925329233673, "learning_rate": 7.593485738263209e-06, "loss": 1.6278, "step": 5077 }, { "epoch": 0.3886422776672279, "grad_norm": 0.5902279888146933, "learning_rate": 7.59257625433262e-06, "loss": 1.4246, "step": 5078 }, { "epoch": 0.3887188121842951, "grad_norm": 0.7113692088794936, "learning_rate": 7.591666672549384e-06, "loss": 1.7215, "step": 5079 }, { "epoch": 0.3887953467013623, "grad_norm": 1.9627533453107722, "learning_rate": 7.5907569929693895e-06, "loss": 1.2779, "step": 5080 }, { "epoch": 0.3888718812184295, "grad_norm": 0.5337558244719597, "learning_rate": 7.58984721564853e-06, "loss": 1.3558, "step": 5081 }, { "epoch": 0.3889484157354967, "grad_norm": 0.5842230499326373, "learning_rate": 7.588937340642705e-06, "loss": 1.2472, "step": 5082 }, { "epoch": 0.3890249502525639, "grad_norm": 0.5387793281330429, "learning_rate": 7.588027368007821e-06, "loss": 1.3204, "step": 5083 }, { "epoch": 0.3891014847696311, "grad_norm": 0.6825556334773013, "learning_rate": 7.5871172977997864e-06, "loss": 1.5154, "step": 5084 }, { "epoch": 0.3891780192866983, "grad_norm": 0.5786208783666053, "learning_rate": 7.5862071300745225e-06, "loss": 1.4498, "step": 5085 }, { "epoch": 0.3892545538037655, "grad_norm": 0.5666818803881666, "learning_rate": 7.5852968648879485e-06, "loss": 1.3859, "step": 5086 }, { "epoch": 0.3893310883208327, "grad_norm": 0.5200096050100961, "learning_rate": 7.584386502295996e-06, "loss": 1.2015, "step": 5087 }, { "epoch": 0.38940762283789987, "grad_norm": 0.6154296911558924, "learning_rate": 7.5834760423546015e-06, "loss": 1.4837, "step": 5088 }, { "epoch": 0.3894841573549671, "grad_norm": 0.5468171548604548, "learning_rate": 7.582565485119705e-06, "loss": 1.2464, "step": 5089 }, { "epoch": 0.3895606918720343, "grad_norm": 0.6305588259149352, "learning_rate": 7.581654830647255e-06, "loss": 1.6092, "step": 5090 }, { "epoch": 0.3896372263891015, "grad_norm": 0.49930926867331205, "learning_rate": 7.580744078993203e-06, "loss": 1.0621, "step": 5091 }, { "epoch": 0.38971376090616866, "grad_norm": 0.898056024710098, "learning_rate": 7.579833230213509e-06, "loss": 1.5008, "step": 5092 }, { "epoch": 0.3897902954232359, "grad_norm": 0.683613749674743, "learning_rate": 7.578922284364139e-06, "loss": 1.3183, "step": 5093 }, { "epoch": 0.3898668299403031, "grad_norm": 0.6501601079935496, "learning_rate": 7.578011241501063e-06, "loss": 1.5847, "step": 5094 }, { "epoch": 0.38994336445737027, "grad_norm": 0.6042737257893048, "learning_rate": 7.577100101680261e-06, "loss": 1.5199, "step": 5095 }, { "epoch": 0.39001989897443745, "grad_norm": 0.5354776861992566, "learning_rate": 7.576188864957712e-06, "loss": 1.236, "step": 5096 }, { "epoch": 0.3900964334915047, "grad_norm": 0.7428032495446654, "learning_rate": 7.575277531389408e-06, "loss": 1.3864, "step": 5097 }, { "epoch": 0.3901729680085719, "grad_norm": 0.5807309173417702, "learning_rate": 7.574366101031343e-06, "loss": 1.3162, "step": 5098 }, { "epoch": 0.39024950252563906, "grad_norm": 0.5410561225518555, "learning_rate": 7.573454573939519e-06, "loss": 1.1416, "step": 5099 }, { "epoch": 0.39032603704270624, "grad_norm": 0.5972094116682676, "learning_rate": 7.572542950169942e-06, "loss": 1.4696, "step": 5100 }, { "epoch": 0.3904025715597735, "grad_norm": 0.6199947554218145, "learning_rate": 7.571631229778624e-06, "loss": 1.5994, "step": 5101 }, { "epoch": 0.39047910607684067, "grad_norm": 0.6799748347578766, "learning_rate": 7.5707194128215854e-06, "loss": 1.3305, "step": 5102 }, { "epoch": 0.39055564059390785, "grad_norm": 0.6775309139622255, "learning_rate": 7.569807499354852e-06, "loss": 1.4289, "step": 5103 }, { "epoch": 0.39063217511097503, "grad_norm": 0.5818683804264407, "learning_rate": 7.568895489434452e-06, "loss": 1.2652, "step": 5104 }, { "epoch": 0.39070870962804227, "grad_norm": 0.8136559201421352, "learning_rate": 7.567983383116423e-06, "loss": 1.7138, "step": 5105 }, { "epoch": 0.39078524414510946, "grad_norm": 0.6668785932145221, "learning_rate": 7.567071180456807e-06, "loss": 1.47, "step": 5106 }, { "epoch": 0.39086177866217664, "grad_norm": 0.5164609401380637, "learning_rate": 7.566158881511655e-06, "loss": 1.1665, "step": 5107 }, { "epoch": 0.3909383131792438, "grad_norm": 0.8154970558932058, "learning_rate": 7.565246486337018e-06, "loss": 1.5507, "step": 5108 }, { "epoch": 0.39101484769631106, "grad_norm": 0.6442786164552667, "learning_rate": 7.5643339949889576e-06, "loss": 1.5879, "step": 5109 }, { "epoch": 0.39109138221337825, "grad_norm": 0.6022949730377121, "learning_rate": 7.56342140752354e-06, "loss": 1.4989, "step": 5110 }, { "epoch": 0.39116791673044543, "grad_norm": 0.6312350150985613, "learning_rate": 7.562508723996838e-06, "loss": 1.7524, "step": 5111 }, { "epoch": 0.3912444512475126, "grad_norm": 1.0523276166907198, "learning_rate": 7.561595944464928e-06, "loss": 1.2486, "step": 5112 }, { "epoch": 0.3913209857645798, "grad_norm": 0.5751276726236811, "learning_rate": 7.560683068983895e-06, "loss": 1.418, "step": 5113 }, { "epoch": 0.39139752028164704, "grad_norm": 0.5680955177384148, "learning_rate": 7.55977009760983e-06, "loss": 1.3896, "step": 5114 }, { "epoch": 0.3914740547987142, "grad_norm": 0.5682737200364477, "learning_rate": 7.558857030398825e-06, "loss": 1.3098, "step": 5115 }, { "epoch": 0.3915505893157814, "grad_norm": 0.5064636585162333, "learning_rate": 7.557943867406986e-06, "loss": 1.052, "step": 5116 }, { "epoch": 0.3916271238328486, "grad_norm": 0.6467228387810956, "learning_rate": 7.557030608690418e-06, "loss": 1.5827, "step": 5117 }, { "epoch": 0.39170365834991583, "grad_norm": 0.5619397797558415, "learning_rate": 7.556117254305236e-06, "loss": 1.3917, "step": 5118 }, { "epoch": 0.391780192866983, "grad_norm": 0.5577387019601809, "learning_rate": 7.555203804307556e-06, "loss": 1.3729, "step": 5119 }, { "epoch": 0.3918567273840502, "grad_norm": 0.49255960999519727, "learning_rate": 7.554290258753505e-06, "loss": 1.1195, "step": 5120 }, { "epoch": 0.3919332619011174, "grad_norm": 0.7454537763068574, "learning_rate": 7.553376617699215e-06, "loss": 1.3682, "step": 5121 }, { "epoch": 0.3920097964181846, "grad_norm": 0.5582821091756105, "learning_rate": 7.552462881200822e-06, "loss": 1.3251, "step": 5122 }, { "epoch": 0.3920863309352518, "grad_norm": 0.6365081400861256, "learning_rate": 7.551549049314467e-06, "loss": 1.6833, "step": 5123 }, { "epoch": 0.392162865452319, "grad_norm": 0.6094312048468059, "learning_rate": 7.550635122096303e-06, "loss": 1.5996, "step": 5124 }, { "epoch": 0.39223939996938617, "grad_norm": 0.48419378434085636, "learning_rate": 7.54972109960248e-06, "loss": 1.2087, "step": 5125 }, { "epoch": 0.3923159344864534, "grad_norm": 0.5457867305801934, "learning_rate": 7.548806981889159e-06, "loss": 1.3229, "step": 5126 }, { "epoch": 0.3923924690035206, "grad_norm": 1.3500830753673227, "learning_rate": 7.547892769012508e-06, "loss": 1.4126, "step": 5127 }, { "epoch": 0.3924690035205878, "grad_norm": 0.7142063738568327, "learning_rate": 7.546978461028698e-06, "loss": 1.7264, "step": 5128 }, { "epoch": 0.39254553803765496, "grad_norm": 0.6117322622727084, "learning_rate": 7.546064057993906e-06, "loss": 1.5429, "step": 5129 }, { "epoch": 0.3926220725547222, "grad_norm": 0.7346682139663668, "learning_rate": 7.545149559964317e-06, "loss": 1.6913, "step": 5130 }, { "epoch": 0.3926986070717894, "grad_norm": 0.5618741837662906, "learning_rate": 7.54423496699612e-06, "loss": 1.402, "step": 5131 }, { "epoch": 0.39277514158885657, "grad_norm": 0.5886478214407994, "learning_rate": 7.543320279145511e-06, "loss": 1.5055, "step": 5132 }, { "epoch": 0.39285167610592375, "grad_norm": 0.6198978164829503, "learning_rate": 7.542405496468689e-06, "loss": 1.4542, "step": 5133 }, { "epoch": 0.392928210622991, "grad_norm": 0.5811592721879085, "learning_rate": 7.541490619021863e-06, "loss": 1.3613, "step": 5134 }, { "epoch": 0.3930047451400582, "grad_norm": 0.5928900728177241, "learning_rate": 7.5405756468612455e-06, "loss": 1.3448, "step": 5135 }, { "epoch": 0.39308127965712536, "grad_norm": 0.6059467248576449, "learning_rate": 7.539660580043054e-06, "loss": 1.5048, "step": 5136 }, { "epoch": 0.39315781417419254, "grad_norm": 0.5346382568187864, "learning_rate": 7.538745418623513e-06, "loss": 1.283, "step": 5137 }, { "epoch": 0.3932343486912598, "grad_norm": 0.7244031328098093, "learning_rate": 7.537830162658856e-06, "loss": 1.5693, "step": 5138 }, { "epoch": 0.39331088320832697, "grad_norm": 0.6203184419602509, "learning_rate": 7.536914812205315e-06, "loss": 1.4595, "step": 5139 }, { "epoch": 0.39338741772539415, "grad_norm": 0.6299248678383526, "learning_rate": 7.535999367319134e-06, "loss": 1.3208, "step": 5140 }, { "epoch": 0.39346395224246133, "grad_norm": 0.6385557873610952, "learning_rate": 7.535083828056561e-06, "loss": 1.506, "step": 5141 }, { "epoch": 0.3935404867595286, "grad_norm": 0.6185366861119054, "learning_rate": 7.5341681944738466e-06, "loss": 1.5274, "step": 5142 }, { "epoch": 0.39361702127659576, "grad_norm": 0.5074038546806459, "learning_rate": 7.533252466627253e-06, "loss": 1.2428, "step": 5143 }, { "epoch": 0.39369355579366294, "grad_norm": 0.6332152170981411, "learning_rate": 7.532336644573044e-06, "loss": 1.2087, "step": 5144 }, { "epoch": 0.3937700903107301, "grad_norm": 0.7047985837068381, "learning_rate": 7.53142072836749e-06, "loss": 1.4704, "step": 5145 }, { "epoch": 0.3938466248277973, "grad_norm": 0.5605016320589065, "learning_rate": 7.53050471806687e-06, "loss": 1.4404, "step": 5146 }, { "epoch": 0.39392315934486455, "grad_norm": 0.6648003898181759, "learning_rate": 7.529588613727464e-06, "loss": 1.4901, "step": 5147 }, { "epoch": 0.39399969386193173, "grad_norm": 0.540822300880013, "learning_rate": 7.52867241540556e-06, "loss": 1.1925, "step": 5148 }, { "epoch": 0.3940762283789989, "grad_norm": 0.6155809480654401, "learning_rate": 7.5277561231574545e-06, "loss": 1.4131, "step": 5149 }, { "epoch": 0.3941527628960661, "grad_norm": 0.5097581585955833, "learning_rate": 7.526839737039446e-06, "loss": 1.2578, "step": 5150 }, { "epoch": 0.39422929741313334, "grad_norm": 0.6902324126775038, "learning_rate": 7.525923257107838e-06, "loss": 1.4329, "step": 5151 }, { "epoch": 0.3943058319302005, "grad_norm": 0.5197174479096759, "learning_rate": 7.525006683418944e-06, "loss": 1.2176, "step": 5152 }, { "epoch": 0.3943823664472677, "grad_norm": 0.6716098376516428, "learning_rate": 7.524090016029081e-06, "loss": 1.3272, "step": 5153 }, { "epoch": 0.3944589009643349, "grad_norm": 0.8636363008177597, "learning_rate": 7.523173254994572e-06, "loss": 1.4503, "step": 5154 }, { "epoch": 0.39453543548140213, "grad_norm": 0.5605849230008647, "learning_rate": 7.5222564003717435e-06, "loss": 1.2637, "step": 5155 }, { "epoch": 0.3946119699984693, "grad_norm": 0.5624498256444803, "learning_rate": 7.52133945221693e-06, "loss": 1.2491, "step": 5156 }, { "epoch": 0.3946885045155365, "grad_norm": 0.5778864334417804, "learning_rate": 7.520422410586474e-06, "loss": 1.5667, "step": 5157 }, { "epoch": 0.3947650390326037, "grad_norm": 0.585923934886404, "learning_rate": 7.51950527553672e-06, "loss": 1.5187, "step": 5158 }, { "epoch": 0.3948415735496709, "grad_norm": 0.7196187613031573, "learning_rate": 7.518588047124019e-06, "loss": 1.8835, "step": 5159 }, { "epoch": 0.3949181080667381, "grad_norm": 0.6476073674616486, "learning_rate": 7.51767072540473e-06, "loss": 1.4323, "step": 5160 }, { "epoch": 0.3949946425838053, "grad_norm": 0.5576629618652529, "learning_rate": 7.516753310435213e-06, "loss": 1.4579, "step": 5161 }, { "epoch": 0.3950711771008725, "grad_norm": 0.5556708568481744, "learning_rate": 7.515835802271839e-06, "loss": 1.1766, "step": 5162 }, { "epoch": 0.3951477116179397, "grad_norm": 0.49651024263951166, "learning_rate": 7.514918200970982e-06, "loss": 1.2684, "step": 5163 }, { "epoch": 0.3952242461350069, "grad_norm": 0.6186419528749252, "learning_rate": 7.514000506589022e-06, "loss": 1.5024, "step": 5164 }, { "epoch": 0.3953007806520741, "grad_norm": 0.5424814017797484, "learning_rate": 7.513082719182345e-06, "loss": 1.2854, "step": 5165 }, { "epoch": 0.39537731516914126, "grad_norm": 0.5773940273306847, "learning_rate": 7.512164838807342e-06, "loss": 1.2992, "step": 5166 }, { "epoch": 0.3954538496862085, "grad_norm": 0.5894179303145295, "learning_rate": 7.5112468655204105e-06, "loss": 1.4265, "step": 5167 }, { "epoch": 0.3955303842032757, "grad_norm": 0.5894294187147575, "learning_rate": 7.510328799377954e-06, "loss": 1.314, "step": 5168 }, { "epoch": 0.39560691872034287, "grad_norm": 0.5693462755441382, "learning_rate": 7.509410640436382e-06, "loss": 1.4461, "step": 5169 }, { "epoch": 0.39568345323741005, "grad_norm": 0.5732783216863206, "learning_rate": 7.508492388752107e-06, "loss": 1.203, "step": 5170 }, { "epoch": 0.3957599877544773, "grad_norm": 0.5526176095859034, "learning_rate": 7.5075740443815504e-06, "loss": 1.3555, "step": 5171 }, { "epoch": 0.3958365222715445, "grad_norm": 0.8123676681664025, "learning_rate": 7.506655607381136e-06, "loss": 1.4825, "step": 5172 }, { "epoch": 0.39591305678861166, "grad_norm": 0.5953451580829285, "learning_rate": 7.505737077807299e-06, "loss": 1.4049, "step": 5173 }, { "epoch": 0.39598959130567885, "grad_norm": 1.1639603523671886, "learning_rate": 7.504818455716475e-06, "loss": 1.7918, "step": 5174 }, { "epoch": 0.3960661258227461, "grad_norm": 0.5560758883501296, "learning_rate": 7.503899741165106e-06, "loss": 1.4187, "step": 5175 }, { "epoch": 0.39614266033981327, "grad_norm": 0.5755367588833732, "learning_rate": 7.502980934209641e-06, "loss": 1.406, "step": 5176 }, { "epoch": 0.39621919485688045, "grad_norm": 0.5150551272202326, "learning_rate": 7.502062034906535e-06, "loss": 1.1214, "step": 5177 }, { "epoch": 0.39629572937394764, "grad_norm": 0.7478678083067355, "learning_rate": 7.5011430433122464e-06, "loss": 1.306, "step": 5178 }, { "epoch": 0.3963722638910149, "grad_norm": 0.5616143500486417, "learning_rate": 7.500223959483242e-06, "loss": 1.4131, "step": 5179 }, { "epoch": 0.39644879840808206, "grad_norm": 0.541801571156767, "learning_rate": 7.499304783475992e-06, "loss": 1.1238, "step": 5180 }, { "epoch": 0.39652533292514924, "grad_norm": 0.5319625301308778, "learning_rate": 7.498385515346975e-06, "loss": 1.2509, "step": 5181 }, { "epoch": 0.3966018674422164, "grad_norm": 0.571297189416191, "learning_rate": 7.4974661551526726e-06, "loss": 1.4183, "step": 5182 }, { "epoch": 0.3966784019592836, "grad_norm": 0.5747327983442706, "learning_rate": 7.496546702949571e-06, "loss": 1.4241, "step": 5183 }, { "epoch": 0.39675493647635085, "grad_norm": 0.6071242650244242, "learning_rate": 7.495627158794169e-06, "loss": 1.4231, "step": 5184 }, { "epoch": 0.39683147099341803, "grad_norm": 0.6344156306945328, "learning_rate": 7.494707522742962e-06, "loss": 1.3569, "step": 5185 }, { "epoch": 0.3969080055104852, "grad_norm": 0.6564240338139362, "learning_rate": 7.493787794852457e-06, "loss": 1.4158, "step": 5186 }, { "epoch": 0.3969845400275524, "grad_norm": 0.5884357696111474, "learning_rate": 7.492867975179165e-06, "loss": 1.3183, "step": 5187 }, { "epoch": 0.39706107454461964, "grad_norm": 0.6055885969617052, "learning_rate": 7.4919480637795996e-06, "loss": 1.5673, "step": 5188 }, { "epoch": 0.3971376090616868, "grad_norm": 0.7458739937672083, "learning_rate": 7.491028060710286e-06, "loss": 1.5649, "step": 5189 }, { "epoch": 0.397214143578754, "grad_norm": 0.6021409365772744, "learning_rate": 7.490107966027751e-06, "loss": 1.317, "step": 5190 }, { "epoch": 0.3972906780958212, "grad_norm": 0.7667977158906845, "learning_rate": 7.4891877797885284e-06, "loss": 1.7602, "step": 5191 }, { "epoch": 0.39736721261288843, "grad_norm": 0.7798534175108295, "learning_rate": 7.488267502049156e-06, "loss": 1.6427, "step": 5192 }, { "epoch": 0.3974437471299556, "grad_norm": 0.6060506847862844, "learning_rate": 7.487347132866181e-06, "loss": 1.552, "step": 5193 }, { "epoch": 0.3975202816470228, "grad_norm": 0.8369863225109885, "learning_rate": 7.48642667229615e-06, "loss": 1.4469, "step": 5194 }, { "epoch": 0.39759681616409, "grad_norm": 0.6854032830642935, "learning_rate": 7.485506120395621e-06, "loss": 1.7667, "step": 5195 }, { "epoch": 0.3976733506811572, "grad_norm": 0.5254091346725793, "learning_rate": 7.484585477221156e-06, "loss": 1.0908, "step": 5196 }, { "epoch": 0.3977498851982244, "grad_norm": 0.7306341374128577, "learning_rate": 7.483664742829321e-06, "loss": 1.445, "step": 5197 }, { "epoch": 0.3978264197152916, "grad_norm": 0.6055629027265492, "learning_rate": 7.482743917276688e-06, "loss": 1.4263, "step": 5198 }, { "epoch": 0.3979029542323588, "grad_norm": 0.5376180814320941, "learning_rate": 7.481823000619838e-06, "loss": 1.1872, "step": 5199 }, { "epoch": 0.397979488749426, "grad_norm": 0.5276960925299228, "learning_rate": 7.480901992915349e-06, "loss": 1.1093, "step": 5200 }, { "epoch": 0.3980560232664932, "grad_norm": 0.6000597794187786, "learning_rate": 7.479980894219819e-06, "loss": 1.6988, "step": 5201 }, { "epoch": 0.3981325577835604, "grad_norm": 0.8174061991769968, "learning_rate": 7.479059704589835e-06, "loss": 1.5704, "step": 5202 }, { "epoch": 0.39820909230062757, "grad_norm": 0.5264725372560862, "learning_rate": 7.478138424082005e-06, "loss": 1.1222, "step": 5203 }, { "epoch": 0.3982856268176948, "grad_norm": 0.6714568881375873, "learning_rate": 7.477217052752929e-06, "loss": 1.5751, "step": 5204 }, { "epoch": 0.398362161334762, "grad_norm": 0.5335879890023433, "learning_rate": 7.4762955906592205e-06, "loss": 1.1041, "step": 5205 }, { "epoch": 0.39843869585182917, "grad_norm": 0.5357112443806484, "learning_rate": 7.475374037857499e-06, "loss": 1.3474, "step": 5206 }, { "epoch": 0.39851523036889636, "grad_norm": 0.5668965828509286, "learning_rate": 7.474452394404387e-06, "loss": 1.3321, "step": 5207 }, { "epoch": 0.3985917648859636, "grad_norm": 0.5457285975267089, "learning_rate": 7.473530660356511e-06, "loss": 1.2608, "step": 5208 }, { "epoch": 0.3986682994030308, "grad_norm": 0.5367237273623211, "learning_rate": 7.472608835770506e-06, "loss": 1.3064, "step": 5209 }, { "epoch": 0.39874483392009796, "grad_norm": 0.5499935020183154, "learning_rate": 7.471686920703012e-06, "loss": 1.058, "step": 5210 }, { "epoch": 0.39882136843716515, "grad_norm": 0.6462594629119166, "learning_rate": 7.4707649152106745e-06, "loss": 1.5542, "step": 5211 }, { "epoch": 0.3988979029542324, "grad_norm": 0.5536988746233812, "learning_rate": 7.4698428193501436e-06, "loss": 1.2668, "step": 5212 }, { "epoch": 0.39897443747129957, "grad_norm": 0.4950776997708746, "learning_rate": 7.468920633178075e-06, "loss": 1.0929, "step": 5213 }, { "epoch": 0.39905097198836675, "grad_norm": 0.9747913664597917, "learning_rate": 7.467998356751132e-06, "loss": 1.2169, "step": 5214 }, { "epoch": 0.39912750650543394, "grad_norm": 0.5556995941731921, "learning_rate": 7.467075990125981e-06, "loss": 1.3863, "step": 5215 }, { "epoch": 0.3992040410225011, "grad_norm": 0.876077750347422, "learning_rate": 7.4661535333592975e-06, "loss": 1.4111, "step": 5216 }, { "epoch": 0.39928057553956836, "grad_norm": 0.6154430109583807, "learning_rate": 7.465230986507755e-06, "loss": 1.5782, "step": 5217 }, { "epoch": 0.39935711005663554, "grad_norm": 0.8383652441326869, "learning_rate": 7.464308349628043e-06, "loss": 1.7326, "step": 5218 }, { "epoch": 0.39943364457370273, "grad_norm": 0.8871415395073865, "learning_rate": 7.463385622776848e-06, "loss": 1.6929, "step": 5219 }, { "epoch": 0.3995101790907699, "grad_norm": 0.5699037389544717, "learning_rate": 7.462462806010866e-06, "loss": 1.3434, "step": 5220 }, { "epoch": 0.39958671360783715, "grad_norm": 0.8160407736926139, "learning_rate": 7.461539899386797e-06, "loss": 1.3375, "step": 5221 }, { "epoch": 0.39966324812490434, "grad_norm": 0.627193564076971, "learning_rate": 7.460616902961346e-06, "loss": 1.6282, "step": 5222 }, { "epoch": 0.3997397826419715, "grad_norm": 0.6084959397399512, "learning_rate": 7.459693816791228e-06, "loss": 1.4294, "step": 5223 }, { "epoch": 0.3998163171590387, "grad_norm": 0.7276183119925478, "learning_rate": 7.4587706409331566e-06, "loss": 1.9681, "step": 5224 }, { "epoch": 0.39989285167610594, "grad_norm": 0.5942876883968881, "learning_rate": 7.4578473754438575e-06, "loss": 1.4842, "step": 5225 }, { "epoch": 0.3999693861931731, "grad_norm": 0.5206209667503925, "learning_rate": 7.456924020380056e-06, "loss": 1.2424, "step": 5226 }, { "epoch": 0.4000459207102403, "grad_norm": 0.6163558879065703, "learning_rate": 7.456000575798487e-06, "loss": 1.4065, "step": 5227 }, { "epoch": 0.4001224552273075, "grad_norm": 0.5072436322235392, "learning_rate": 7.455077041755891e-06, "loss": 1.094, "step": 5228 }, { "epoch": 0.40019898974437473, "grad_norm": 0.9226779148492957, "learning_rate": 7.454153418309014e-06, "loss": 1.7154, "step": 5229 }, { "epoch": 0.4002755242614419, "grad_norm": 0.6231705502994322, "learning_rate": 7.453229705514601e-06, "loss": 1.6103, "step": 5230 }, { "epoch": 0.4003520587785091, "grad_norm": 0.5693977304474198, "learning_rate": 7.452305903429412e-06, "loss": 1.347, "step": 5231 }, { "epoch": 0.4004285932955763, "grad_norm": 0.4789210798036222, "learning_rate": 7.451382012110206e-06, "loss": 0.991, "step": 5232 }, { "epoch": 0.4005051278126435, "grad_norm": 1.0178304286444668, "learning_rate": 7.45045803161375e-06, "loss": 1.2222, "step": 5233 }, { "epoch": 0.4005816623297107, "grad_norm": 0.640085058540141, "learning_rate": 7.449533961996817e-06, "loss": 1.507, "step": 5234 }, { "epoch": 0.4006581968467779, "grad_norm": 0.5867213833501299, "learning_rate": 7.448609803316183e-06, "loss": 1.384, "step": 5235 }, { "epoch": 0.4007347313638451, "grad_norm": 0.7455218245486039, "learning_rate": 7.447685555628633e-06, "loss": 1.284, "step": 5236 }, { "epoch": 0.4008112658809123, "grad_norm": 0.5798511078183828, "learning_rate": 7.446761218990956e-06, "loss": 1.307, "step": 5237 }, { "epoch": 0.4008878003979795, "grad_norm": 0.5241795658898802, "learning_rate": 7.445836793459943e-06, "loss": 1.1093, "step": 5238 }, { "epoch": 0.4009643349150467, "grad_norm": 0.6081866867049528, "learning_rate": 7.444912279092397e-06, "loss": 1.4216, "step": 5239 }, { "epoch": 0.40104086943211387, "grad_norm": 0.8616961993542104, "learning_rate": 7.443987675945119e-06, "loss": 1.4312, "step": 5240 }, { "epoch": 0.4011174039491811, "grad_norm": 0.5559599906044437, "learning_rate": 7.443062984074922e-06, "loss": 1.3481, "step": 5241 }, { "epoch": 0.4011939384662483, "grad_norm": 0.8290233413395354, "learning_rate": 7.442138203538622e-06, "loss": 1.6807, "step": 5242 }, { "epoch": 0.4012704729833155, "grad_norm": 0.6519997495035407, "learning_rate": 7.4412133343930405e-06, "loss": 1.712, "step": 5243 }, { "epoch": 0.40134700750038266, "grad_norm": 0.6069701992247909, "learning_rate": 7.440288376695001e-06, "loss": 1.4808, "step": 5244 }, { "epoch": 0.4014235420174499, "grad_norm": 0.810206183350296, "learning_rate": 7.439363330501339e-06, "loss": 1.4073, "step": 5245 }, { "epoch": 0.4015000765345171, "grad_norm": 0.5869826659064109, "learning_rate": 7.438438195868891e-06, "loss": 1.3423, "step": 5246 }, { "epoch": 0.40157661105158426, "grad_norm": 0.7213398153715685, "learning_rate": 7.4375129728545e-06, "loss": 1.4119, "step": 5247 }, { "epoch": 0.40165314556865145, "grad_norm": 0.5649090122989764, "learning_rate": 7.436587661515015e-06, "loss": 1.4875, "step": 5248 }, { "epoch": 0.40172968008571863, "grad_norm": 0.6623975894152836, "learning_rate": 7.435662261907289e-06, "loss": 1.7025, "step": 5249 }, { "epoch": 0.40180621460278587, "grad_norm": 0.5624371567868429, "learning_rate": 7.434736774088184e-06, "loss": 1.2397, "step": 5250 }, { "epoch": 0.40188274911985306, "grad_norm": 0.6156075099040845, "learning_rate": 7.4338111981145596e-06, "loss": 1.3907, "step": 5251 }, { "epoch": 0.40195928363692024, "grad_norm": 0.5671344607463972, "learning_rate": 7.43288553404329e-06, "loss": 1.4284, "step": 5252 }, { "epoch": 0.4020358181539874, "grad_norm": 0.7722568782526356, "learning_rate": 7.431959781931251e-06, "loss": 1.6872, "step": 5253 }, { "epoch": 0.40211235267105466, "grad_norm": 0.6550631601742322, "learning_rate": 7.431033941835321e-06, "loss": 1.4985, "step": 5254 }, { "epoch": 0.40218888718812185, "grad_norm": 0.7217454174035961, "learning_rate": 7.430108013812387e-06, "loss": 1.641, "step": 5255 }, { "epoch": 0.40226542170518903, "grad_norm": 0.7278920361775824, "learning_rate": 7.429181997919342e-06, "loss": 1.7056, "step": 5256 }, { "epoch": 0.4023419562222562, "grad_norm": 0.4762445414368096, "learning_rate": 7.428255894213082e-06, "loss": 0.9827, "step": 5257 }, { "epoch": 0.40241849073932345, "grad_norm": 0.6133299221770038, "learning_rate": 7.427329702750512e-06, "loss": 1.4235, "step": 5258 }, { "epoch": 0.40249502525639064, "grad_norm": 0.6022368934813458, "learning_rate": 7.426403423588536e-06, "loss": 1.4802, "step": 5259 }, { "epoch": 0.4025715597734578, "grad_norm": 0.6317302915237758, "learning_rate": 7.4254770567840705e-06, "loss": 1.3699, "step": 5260 }, { "epoch": 0.402648094290525, "grad_norm": 0.5897042344740093, "learning_rate": 7.424550602394034e-06, "loss": 1.3438, "step": 5261 }, { "epoch": 0.40272462880759224, "grad_norm": 0.603244488369849, "learning_rate": 7.42362406047535e-06, "loss": 1.4747, "step": 5262 }, { "epoch": 0.40280116332465943, "grad_norm": 0.6941726737420186, "learning_rate": 7.422697431084945e-06, "loss": 1.6809, "step": 5263 }, { "epoch": 0.4028776978417266, "grad_norm": 0.5872976495311459, "learning_rate": 7.42177071427976e-06, "loss": 1.4247, "step": 5264 }, { "epoch": 0.4029542323587938, "grad_norm": 0.5831880500652454, "learning_rate": 7.420843910116732e-06, "loss": 1.4652, "step": 5265 }, { "epoch": 0.40303076687586104, "grad_norm": 0.6303133607011658, "learning_rate": 7.4199170186528045e-06, "loss": 1.6168, "step": 5266 }, { "epoch": 0.4031073013929282, "grad_norm": 0.7653821402153005, "learning_rate": 7.418990039944932e-06, "loss": 1.8707, "step": 5267 }, { "epoch": 0.4031838359099954, "grad_norm": 0.6691750067340858, "learning_rate": 7.41806297405007e-06, "loss": 1.6461, "step": 5268 }, { "epoch": 0.4032603704270626, "grad_norm": 0.6171314419075703, "learning_rate": 7.417135821025178e-06, "loss": 1.2645, "step": 5269 }, { "epoch": 0.4033369049441298, "grad_norm": 0.727685935882034, "learning_rate": 7.416208580927225e-06, "loss": 1.5609, "step": 5270 }, { "epoch": 0.403413439461197, "grad_norm": 0.5995410007153487, "learning_rate": 7.415281253813184e-06, "loss": 1.4217, "step": 5271 }, { "epoch": 0.4034899739782642, "grad_norm": 0.5985296814009624, "learning_rate": 7.414353839740032e-06, "loss": 1.4516, "step": 5272 }, { "epoch": 0.4035665084953314, "grad_norm": 0.6017597797376575, "learning_rate": 7.413426338764752e-06, "loss": 1.489, "step": 5273 }, { "epoch": 0.4036430430123986, "grad_norm": 0.5688904173839745, "learning_rate": 7.412498750944331e-06, "loss": 1.3983, "step": 5274 }, { "epoch": 0.4037195775294658, "grad_norm": 0.6893124873809876, "learning_rate": 7.411571076335765e-06, "loss": 1.4887, "step": 5275 }, { "epoch": 0.403796112046533, "grad_norm": 0.5773454734607859, "learning_rate": 7.410643314996053e-06, "loss": 1.3376, "step": 5276 }, { "epoch": 0.40387264656360017, "grad_norm": 0.6414049550151427, "learning_rate": 7.409715466982198e-06, "loss": 1.4757, "step": 5277 }, { "epoch": 0.4039491810806674, "grad_norm": 0.548754240338125, "learning_rate": 7.4087875323512106e-06, "loss": 1.2692, "step": 5278 }, { "epoch": 0.4040257155977346, "grad_norm": 0.6300760833304186, "learning_rate": 7.407859511160106e-06, "loss": 1.6004, "step": 5279 }, { "epoch": 0.4041022501148018, "grad_norm": 0.6962108693916331, "learning_rate": 7.4069314034659036e-06, "loss": 1.7166, "step": 5280 }, { "epoch": 0.40417878463186896, "grad_norm": 0.6184810454356806, "learning_rate": 7.406003209325631e-06, "loss": 1.2144, "step": 5281 }, { "epoch": 0.40425531914893614, "grad_norm": 0.5870683140416043, "learning_rate": 7.405074928796316e-06, "loss": 1.279, "step": 5282 }, { "epoch": 0.4043318536660034, "grad_norm": 0.5897279098166093, "learning_rate": 7.404146561935e-06, "loss": 1.3667, "step": 5283 }, { "epoch": 0.40440838818307057, "grad_norm": 0.707305496671967, "learning_rate": 7.40321810879872e-06, "loss": 1.459, "step": 5284 }, { "epoch": 0.40448492270013775, "grad_norm": 0.6487327830984191, "learning_rate": 7.402289569444524e-06, "loss": 1.3898, "step": 5285 }, { "epoch": 0.40456145721720493, "grad_norm": 0.6869936671588627, "learning_rate": 7.401360943929466e-06, "loss": 1.5126, "step": 5286 }, { "epoch": 0.4046379917342722, "grad_norm": 0.6081757387137661, "learning_rate": 7.400432232310601e-06, "loss": 1.5259, "step": 5287 }, { "epoch": 0.40471452625133936, "grad_norm": 0.588961363731291, "learning_rate": 7.399503434644995e-06, "loss": 1.3705, "step": 5288 }, { "epoch": 0.40479106076840654, "grad_norm": 0.5376861410595521, "learning_rate": 7.398574550989713e-06, "loss": 1.1252, "step": 5289 }, { "epoch": 0.4048675952854737, "grad_norm": 0.5263609941808107, "learning_rate": 7.397645581401829e-06, "loss": 1.1037, "step": 5290 }, { "epoch": 0.40494412980254096, "grad_norm": 0.6918493766970237, "learning_rate": 7.396716525938423e-06, "loss": 1.2019, "step": 5291 }, { "epoch": 0.40502066431960815, "grad_norm": 0.6146892756600657, "learning_rate": 7.39578738465658e-06, "loss": 1.3961, "step": 5292 }, { "epoch": 0.40509719883667533, "grad_norm": 0.566215396773704, "learning_rate": 7.394858157613384e-06, "loss": 1.2544, "step": 5293 }, { "epoch": 0.4051737333537425, "grad_norm": 1.1763024908365156, "learning_rate": 7.393928844865935e-06, "loss": 1.485, "step": 5294 }, { "epoch": 0.40525026787080975, "grad_norm": 1.8497071798034708, "learning_rate": 7.392999446471331e-06, "loss": 2.0494, "step": 5295 }, { "epoch": 0.40532680238787694, "grad_norm": 0.60845506293818, "learning_rate": 7.3920699624866755e-06, "loss": 1.4265, "step": 5296 }, { "epoch": 0.4054033369049441, "grad_norm": 0.6203073495612224, "learning_rate": 7.391140392969081e-06, "loss": 1.2416, "step": 5297 }, { "epoch": 0.4054798714220113, "grad_norm": 0.5057850411038396, "learning_rate": 7.390210737975662e-06, "loss": 1.1464, "step": 5298 }, { "epoch": 0.40555640593907855, "grad_norm": 0.6611110911064985, "learning_rate": 7.38928099756354e-06, "loss": 1.5597, "step": 5299 }, { "epoch": 0.40563294045614573, "grad_norm": 0.6816318697555593, "learning_rate": 7.388351171789839e-06, "loss": 1.4806, "step": 5300 }, { "epoch": 0.4057094749732129, "grad_norm": 0.6117608886527769, "learning_rate": 7.387421260711692e-06, "loss": 1.4363, "step": 5301 }, { "epoch": 0.4057860094902801, "grad_norm": 0.5781444353563521, "learning_rate": 7.386491264386235e-06, "loss": 1.3471, "step": 5302 }, { "epoch": 0.40586254400734734, "grad_norm": 0.6940955779357506, "learning_rate": 7.385561182870611e-06, "loss": 1.5503, "step": 5303 }, { "epoch": 0.4059390785244145, "grad_norm": 0.7349955937481477, "learning_rate": 7.3846310162219636e-06, "loss": 1.8806, "step": 5304 }, { "epoch": 0.4060156130414817, "grad_norm": 0.7080860589830342, "learning_rate": 7.383700764497449e-06, "loss": 1.6954, "step": 5305 }, { "epoch": 0.4060921475585489, "grad_norm": 0.5941520429777707, "learning_rate": 7.382770427754223e-06, "loss": 1.579, "step": 5306 }, { "epoch": 0.4061686820756161, "grad_norm": 0.5060897836264737, "learning_rate": 7.381840006049448e-06, "loss": 1.241, "step": 5307 }, { "epoch": 0.4062452165926833, "grad_norm": 0.598864675147492, "learning_rate": 7.380909499440291e-06, "loss": 1.4027, "step": 5308 }, { "epoch": 0.4063217511097505, "grad_norm": 0.616097986552794, "learning_rate": 7.379978907983928e-06, "loss": 1.6493, "step": 5309 }, { "epoch": 0.4063982856268177, "grad_norm": 0.6171817353514051, "learning_rate": 7.3790482317375354e-06, "loss": 1.2337, "step": 5310 }, { "epoch": 0.4064748201438849, "grad_norm": 0.574308614217457, "learning_rate": 7.378117470758296e-06, "loss": 1.481, "step": 5311 }, { "epoch": 0.4065513546609521, "grad_norm": 0.5683210281923312, "learning_rate": 7.377186625103399e-06, "loss": 1.4777, "step": 5312 }, { "epoch": 0.4066278891780193, "grad_norm": 0.6085568152557694, "learning_rate": 7.37625569483004e-06, "loss": 1.3593, "step": 5313 }, { "epoch": 0.40670442369508647, "grad_norm": 0.5765507365451217, "learning_rate": 7.375324679995417e-06, "loss": 1.4416, "step": 5314 }, { "epoch": 0.40678095821215365, "grad_norm": 0.6020329424241715, "learning_rate": 7.374393580656733e-06, "loss": 1.4435, "step": 5315 }, { "epoch": 0.4068574927292209, "grad_norm": 0.792407722555209, "learning_rate": 7.373462396871199e-06, "loss": 1.5958, "step": 5316 }, { "epoch": 0.4069340272462881, "grad_norm": 1.5225087953679175, "learning_rate": 7.37253112869603e-06, "loss": 1.4102, "step": 5317 }, { "epoch": 0.40701056176335526, "grad_norm": 0.7504864686261077, "learning_rate": 7.3715997761884436e-06, "loss": 1.521, "step": 5318 }, { "epoch": 0.40708709628042244, "grad_norm": 0.7439036277896026, "learning_rate": 7.370668339405669e-06, "loss": 1.3488, "step": 5319 }, { "epoch": 0.4071636307974897, "grad_norm": 0.5805329979478095, "learning_rate": 7.369736818404934e-06, "loss": 1.4229, "step": 5320 }, { "epoch": 0.40724016531455687, "grad_norm": 0.5731447078976707, "learning_rate": 7.368805213243473e-06, "loss": 1.23, "step": 5321 }, { "epoch": 0.40731669983162405, "grad_norm": 0.6385981085634055, "learning_rate": 7.367873523978529e-06, "loss": 1.4232, "step": 5322 }, { "epoch": 0.40739323434869124, "grad_norm": 0.536254983914339, "learning_rate": 7.366941750667345e-06, "loss": 1.3406, "step": 5323 }, { "epoch": 0.4074697688657585, "grad_norm": 0.6368903030854959, "learning_rate": 7.366009893367174e-06, "loss": 1.4722, "step": 5324 }, { "epoch": 0.40754630338282566, "grad_norm": 0.6640946953250255, "learning_rate": 7.3650779521352725e-06, "loss": 1.3153, "step": 5325 }, { "epoch": 0.40762283789989284, "grad_norm": 0.6184950699919327, "learning_rate": 7.3641459270289e-06, "loss": 1.3981, "step": 5326 }, { "epoch": 0.40769937241696, "grad_norm": 0.5533349016521819, "learning_rate": 7.363213818105323e-06, "loss": 1.3932, "step": 5327 }, { "epoch": 0.40777590693402727, "grad_norm": 0.5509652500162231, "learning_rate": 7.362281625421814e-06, "loss": 1.0506, "step": 5328 }, { "epoch": 0.40785244145109445, "grad_norm": 0.5430712765089122, "learning_rate": 7.36134934903565e-06, "loss": 1.1981, "step": 5329 }, { "epoch": 0.40792897596816163, "grad_norm": 0.6368620665827847, "learning_rate": 7.360416989004112e-06, "loss": 1.4132, "step": 5330 }, { "epoch": 0.4080055104852288, "grad_norm": 0.5180244360977604, "learning_rate": 7.359484545384489e-06, "loss": 1.0786, "step": 5331 }, { "epoch": 0.40808204500229606, "grad_norm": 0.4976703258671565, "learning_rate": 7.3585520182340695e-06, "loss": 1.2263, "step": 5332 }, { "epoch": 0.40815857951936324, "grad_norm": 0.5650236064029195, "learning_rate": 7.357619407610154e-06, "loss": 1.2997, "step": 5333 }, { "epoch": 0.4082351140364304, "grad_norm": 0.5492626438106695, "learning_rate": 7.356686713570043e-06, "loss": 1.2833, "step": 5334 }, { "epoch": 0.4083116485534976, "grad_norm": 0.4738317011952399, "learning_rate": 7.355753936171043e-06, "loss": 1.1658, "step": 5335 }, { "epoch": 0.40838818307056485, "grad_norm": 0.6035191452562518, "learning_rate": 7.354821075470471e-06, "loss": 1.4682, "step": 5336 }, { "epoch": 0.40846471758763203, "grad_norm": 0.5515559038324972, "learning_rate": 7.35388813152564e-06, "loss": 1.3552, "step": 5337 }, { "epoch": 0.4085412521046992, "grad_norm": 0.5771541045277606, "learning_rate": 7.352955104393877e-06, "loss": 1.3107, "step": 5338 }, { "epoch": 0.4086177866217664, "grad_norm": 0.5500988196959294, "learning_rate": 7.352021994132505e-06, "loss": 1.2348, "step": 5339 }, { "epoch": 0.40869432113883364, "grad_norm": 0.6192700325788233, "learning_rate": 7.351088800798861e-06, "loss": 1.5078, "step": 5340 }, { "epoch": 0.4087708556559008, "grad_norm": 0.9585746399579248, "learning_rate": 7.350155524450283e-06, "loss": 1.3398, "step": 5341 }, { "epoch": 0.408847390172968, "grad_norm": 0.558543653092684, "learning_rate": 7.349222165144112e-06, "loss": 1.3794, "step": 5342 }, { "epoch": 0.4089239246900352, "grad_norm": 0.6386207537103019, "learning_rate": 7.3482887229377e-06, "loss": 1.6369, "step": 5343 }, { "epoch": 0.40900045920710243, "grad_norm": 0.5982790518439153, "learning_rate": 7.347355197888396e-06, "loss": 1.4635, "step": 5344 }, { "epoch": 0.4090769937241696, "grad_norm": 0.485089695709651, "learning_rate": 7.3464215900535615e-06, "loss": 1.0415, "step": 5345 }, { "epoch": 0.4091535282412368, "grad_norm": 0.6836126420551913, "learning_rate": 7.345487899490559e-06, "loss": 1.5802, "step": 5346 }, { "epoch": 0.409230062758304, "grad_norm": 0.7134927884527628, "learning_rate": 7.344554126256759e-06, "loss": 1.6762, "step": 5347 }, { "epoch": 0.4093065972753712, "grad_norm": 0.6280594827094721, "learning_rate": 7.343620270409533e-06, "loss": 1.3175, "step": 5348 }, { "epoch": 0.4093831317924384, "grad_norm": 0.7261748116940603, "learning_rate": 7.34268633200626e-06, "loss": 1.4521, "step": 5349 }, { "epoch": 0.4094596663095056, "grad_norm": 0.5178527360634845, "learning_rate": 7.341752311104327e-06, "loss": 1.1436, "step": 5350 }, { "epoch": 0.40953620082657277, "grad_norm": 0.5769283018806195, "learning_rate": 7.340818207761119e-06, "loss": 1.2675, "step": 5351 }, { "epoch": 0.40961273534363996, "grad_norm": 0.5447050123850764, "learning_rate": 7.339884022034034e-06, "loss": 1.1824, "step": 5352 }, { "epoch": 0.4096892698607072, "grad_norm": 0.5333262159357514, "learning_rate": 7.3389497539804674e-06, "loss": 1.3354, "step": 5353 }, { "epoch": 0.4097658043777744, "grad_norm": 0.5608829294431291, "learning_rate": 7.338015403657827e-06, "loss": 1.3216, "step": 5354 }, { "epoch": 0.40984233889484156, "grad_norm": 0.5772658900396052, "learning_rate": 7.33708097112352e-06, "loss": 1.2798, "step": 5355 }, { "epoch": 0.40991887341190875, "grad_norm": 0.5491808402270231, "learning_rate": 7.33614645643496e-06, "loss": 1.2374, "step": 5356 }, { "epoch": 0.409995407928976, "grad_norm": 1.616253564371207, "learning_rate": 7.335211859649568e-06, "loss": 1.5897, "step": 5357 }, { "epoch": 0.41007194244604317, "grad_norm": 0.7152055787492559, "learning_rate": 7.334277180824767e-06, "loss": 1.3584, "step": 5358 }, { "epoch": 0.41014847696311035, "grad_norm": 0.7352160191587895, "learning_rate": 7.333342420017988e-06, "loss": 1.1496, "step": 5359 }, { "epoch": 0.41022501148017754, "grad_norm": 0.4864693791032972, "learning_rate": 7.332407577286666e-06, "loss": 1.0471, "step": 5360 }, { "epoch": 0.4103015459972448, "grad_norm": 0.5446356814950415, "learning_rate": 7.331472652688237e-06, "loss": 1.3091, "step": 5361 }, { "epoch": 0.41037808051431196, "grad_norm": 0.54316843121237, "learning_rate": 7.3305376462801484e-06, "loss": 1.1401, "step": 5362 }, { "epoch": 0.41045461503137914, "grad_norm": 0.698870425122538, "learning_rate": 7.3296025581198495e-06, "loss": 1.6187, "step": 5363 }, { "epoch": 0.41053114954844633, "grad_norm": 0.6572699558703756, "learning_rate": 7.328667388264795e-06, "loss": 1.5503, "step": 5364 }, { "epoch": 0.41060768406551357, "grad_norm": 0.7496011127545371, "learning_rate": 7.327732136772444e-06, "loss": 1.5517, "step": 5365 }, { "epoch": 0.41068421858258075, "grad_norm": 0.46943647970640734, "learning_rate": 7.326796803700263e-06, "loss": 1.0133, "step": 5366 }, { "epoch": 0.41076075309964794, "grad_norm": 0.60743341972259, "learning_rate": 7.325861389105717e-06, "loss": 1.3808, "step": 5367 }, { "epoch": 0.4108372876167151, "grad_norm": 0.5644554997371695, "learning_rate": 7.324925893046285e-06, "loss": 1.3519, "step": 5368 }, { "epoch": 0.41091382213378236, "grad_norm": 0.5483651887174397, "learning_rate": 7.323990315579447e-06, "loss": 1.0864, "step": 5369 }, { "epoch": 0.41099035665084954, "grad_norm": 0.659041635112649, "learning_rate": 7.323054656762684e-06, "loss": 1.5605, "step": 5370 }, { "epoch": 0.4110668911679167, "grad_norm": 0.5673440468249761, "learning_rate": 7.322118916653488e-06, "loss": 1.2878, "step": 5371 }, { "epoch": 0.4111434256849839, "grad_norm": 0.6022965113195061, "learning_rate": 7.321183095309353e-06, "loss": 1.5564, "step": 5372 }, { "epoch": 0.41121996020205115, "grad_norm": 0.6515696770861813, "learning_rate": 7.32024719278778e-06, "loss": 1.712, "step": 5373 }, { "epoch": 0.41129649471911833, "grad_norm": 0.5379618768547203, "learning_rate": 7.319311209146272e-06, "loss": 1.246, "step": 5374 }, { "epoch": 0.4113730292361855, "grad_norm": 0.6162133416746683, "learning_rate": 7.318375144442339e-06, "loss": 1.6324, "step": 5375 }, { "epoch": 0.4114495637532527, "grad_norm": 0.4967150549223374, "learning_rate": 7.317438998733497e-06, "loss": 1.1705, "step": 5376 }, { "epoch": 0.41152609827031994, "grad_norm": 0.8531451347363934, "learning_rate": 7.316502772077265e-06, "loss": 1.4884, "step": 5377 }, { "epoch": 0.4116026327873871, "grad_norm": 0.8904167596383271, "learning_rate": 7.315566464531166e-06, "loss": 1.5843, "step": 5378 }, { "epoch": 0.4116791673044543, "grad_norm": 0.6597440379911274, "learning_rate": 7.3146300761527315e-06, "loss": 1.4211, "step": 5379 }, { "epoch": 0.4117557018215215, "grad_norm": 0.5787617724355729, "learning_rate": 7.3136936069994945e-06, "loss": 1.403, "step": 5380 }, { "epoch": 0.41183223633858873, "grad_norm": 0.5593649196257964, "learning_rate": 7.3127570571289956e-06, "loss": 1.2804, "step": 5381 }, { "epoch": 0.4119087708556559, "grad_norm": 0.7630636285175114, "learning_rate": 7.311820426598777e-06, "loss": 1.5299, "step": 5382 }, { "epoch": 0.4119853053727231, "grad_norm": 0.593975541168105, "learning_rate": 7.310883715466392e-06, "loss": 1.4133, "step": 5383 }, { "epoch": 0.4120618398897903, "grad_norm": 0.5204207147716694, "learning_rate": 7.309946923789391e-06, "loss": 1.1365, "step": 5384 }, { "epoch": 0.41213837440685747, "grad_norm": 0.5907115764636458, "learning_rate": 7.309010051625335e-06, "loss": 1.4347, "step": 5385 }, { "epoch": 0.4122149089239247, "grad_norm": 0.6771309035333711, "learning_rate": 7.308073099031789e-06, "loss": 1.5255, "step": 5386 }, { "epoch": 0.4122914434409919, "grad_norm": 0.5725863307165321, "learning_rate": 7.30713606606632e-06, "loss": 1.3207, "step": 5387 }, { "epoch": 0.4123679779580591, "grad_norm": 0.522213672271069, "learning_rate": 7.3061989527865066e-06, "loss": 1.1558, "step": 5388 }, { "epoch": 0.41244451247512626, "grad_norm": 0.5411775934618136, "learning_rate": 7.305261759249923e-06, "loss": 1.2063, "step": 5389 }, { "epoch": 0.4125210469921935, "grad_norm": 0.6115200432656875, "learning_rate": 7.304324485514155e-06, "loss": 1.3724, "step": 5390 }, { "epoch": 0.4125975815092607, "grad_norm": 0.7938760305826714, "learning_rate": 7.303387131636791e-06, "loss": 1.3084, "step": 5391 }, { "epoch": 0.41267411602632786, "grad_norm": 0.7267368044349842, "learning_rate": 7.302449697675426e-06, "loss": 1.502, "step": 5392 }, { "epoch": 0.41275065054339505, "grad_norm": 0.8464283194143963, "learning_rate": 7.301512183687657e-06, "loss": 1.5602, "step": 5393 }, { "epoch": 0.4128271850604623, "grad_norm": 0.6351409731632227, "learning_rate": 7.300574589731089e-06, "loss": 1.0929, "step": 5394 }, { "epoch": 0.41290371957752947, "grad_norm": 0.6499865160196708, "learning_rate": 7.2996369158633304e-06, "loss": 1.5837, "step": 5395 }, { "epoch": 0.41298025409459665, "grad_norm": 0.6960809311744997, "learning_rate": 7.298699162141996e-06, "loss": 1.6961, "step": 5396 }, { "epoch": 0.41305678861166384, "grad_norm": 0.6991832422116381, "learning_rate": 7.2977613286247e-06, "loss": 1.5589, "step": 5397 }, { "epoch": 0.4131333231287311, "grad_norm": 0.8067035561877189, "learning_rate": 7.2968234153690684e-06, "loss": 1.604, "step": 5398 }, { "epoch": 0.41320985764579826, "grad_norm": 0.5912366863684193, "learning_rate": 7.295885422432733e-06, "loss": 1.4021, "step": 5399 }, { "epoch": 0.41328639216286545, "grad_norm": 0.5566775962838384, "learning_rate": 7.2949473498733206e-06, "loss": 1.2763, "step": 5400 }, { "epoch": 0.41336292667993263, "grad_norm": 0.5820114647020875, "learning_rate": 7.2940091977484715e-06, "loss": 1.3863, "step": 5401 }, { "epoch": 0.41343946119699987, "grad_norm": 0.5700370174767725, "learning_rate": 7.2930709661158305e-06, "loss": 1.3396, "step": 5402 }, { "epoch": 0.41351599571406705, "grad_norm": 0.5680012894542517, "learning_rate": 7.2921326550330415e-06, "loss": 1.1021, "step": 5403 }, { "epoch": 0.41359253023113424, "grad_norm": 0.6556308815671341, "learning_rate": 7.291194264557762e-06, "loss": 1.3792, "step": 5404 }, { "epoch": 0.4136690647482014, "grad_norm": 0.5488462726510296, "learning_rate": 7.2902557947476445e-06, "loss": 1.2235, "step": 5405 }, { "epoch": 0.41374559926526866, "grad_norm": 0.699793927062409, "learning_rate": 7.289317245660356e-06, "loss": 1.6786, "step": 5406 }, { "epoch": 0.41382213378233584, "grad_norm": 0.8513641827234691, "learning_rate": 7.288378617353561e-06, "loss": 1.7463, "step": 5407 }, { "epoch": 0.413898668299403, "grad_norm": 0.6330334302454987, "learning_rate": 7.2874399098849315e-06, "loss": 1.2364, "step": 5408 }, { "epoch": 0.4139752028164702, "grad_norm": 0.4997451144649944, "learning_rate": 7.2865011233121464e-06, "loss": 1.082, "step": 5409 }, { "epoch": 0.41405173733353745, "grad_norm": 0.6180737389779741, "learning_rate": 7.285562257692886e-06, "loss": 1.5328, "step": 5410 }, { "epoch": 0.41412827185060463, "grad_norm": 0.5980860791976741, "learning_rate": 7.284623313084836e-06, "loss": 1.5414, "step": 5411 }, { "epoch": 0.4142048063676718, "grad_norm": 0.5337062754932452, "learning_rate": 7.283684289545692e-06, "loss": 1.1702, "step": 5412 }, { "epoch": 0.414281340884739, "grad_norm": 0.6062943929362398, "learning_rate": 7.282745187133145e-06, "loss": 1.3487, "step": 5413 }, { "epoch": 0.41435787540180624, "grad_norm": 0.5641461537346516, "learning_rate": 7.2818060059049e-06, "loss": 1.4525, "step": 5414 }, { "epoch": 0.4144344099188734, "grad_norm": 0.5780595828552058, "learning_rate": 7.280866745918663e-06, "loss": 1.4252, "step": 5415 }, { "epoch": 0.4145109444359406, "grad_norm": 0.5581355441521288, "learning_rate": 7.279927407232143e-06, "loss": 1.4696, "step": 5416 }, { "epoch": 0.4145874789530078, "grad_norm": 0.58095206693978, "learning_rate": 7.278987989903058e-06, "loss": 1.3872, "step": 5417 }, { "epoch": 0.414664013470075, "grad_norm": 0.5284481074283655, "learning_rate": 7.278048493989126e-06, "loss": 1.2136, "step": 5418 }, { "epoch": 0.4147405479871422, "grad_norm": 0.6088005829265735, "learning_rate": 7.277108919548075e-06, "loss": 1.6024, "step": 5419 }, { "epoch": 0.4148170825042094, "grad_norm": 0.509832776968573, "learning_rate": 7.276169266637634e-06, "loss": 1.1877, "step": 5420 }, { "epoch": 0.4148936170212766, "grad_norm": 0.7991591648336432, "learning_rate": 7.275229535315538e-06, "loss": 1.2547, "step": 5421 }, { "epoch": 0.41497015153834377, "grad_norm": 0.5618174330156833, "learning_rate": 7.2742897256395275e-06, "loss": 1.3008, "step": 5422 }, { "epoch": 0.415046686055411, "grad_norm": 0.6173596172920434, "learning_rate": 7.273349837667347e-06, "loss": 1.5328, "step": 5423 }, { "epoch": 0.4151232205724782, "grad_norm": 0.5642489494282461, "learning_rate": 7.272409871456744e-06, "loss": 1.3832, "step": 5424 }, { "epoch": 0.4151997550895454, "grad_norm": 0.6724600524803629, "learning_rate": 7.271469827065477e-06, "loss": 1.2524, "step": 5425 }, { "epoch": 0.41527628960661256, "grad_norm": 0.5355670063005009, "learning_rate": 7.270529704551303e-06, "loss": 1.2776, "step": 5426 }, { "epoch": 0.4153528241236798, "grad_norm": 0.5851132885707899, "learning_rate": 7.2695895039719826e-06, "loss": 1.4055, "step": 5427 }, { "epoch": 0.415429358640747, "grad_norm": 0.559240965931521, "learning_rate": 7.268649225385291e-06, "loss": 1.2872, "step": 5428 }, { "epoch": 0.41550589315781417, "grad_norm": 0.5524879293003467, "learning_rate": 7.2677088688489975e-06, "loss": 1.4082, "step": 5429 }, { "epoch": 0.41558242767488135, "grad_norm": 1.524777777236438, "learning_rate": 7.26676843442088e-06, "loss": 1.732, "step": 5430 }, { "epoch": 0.4156589621919486, "grad_norm": 0.5628235154624535, "learning_rate": 7.265827922158725e-06, "loss": 1.2419, "step": 5431 }, { "epoch": 0.4157354967090158, "grad_norm": 0.5483670917203316, "learning_rate": 7.2648873321203175e-06, "loss": 1.1895, "step": 5432 }, { "epoch": 0.41581203122608296, "grad_norm": 0.6466264500844835, "learning_rate": 7.2639466643634516e-06, "loss": 1.5236, "step": 5433 }, { "epoch": 0.41588856574315014, "grad_norm": 0.7511838218425823, "learning_rate": 7.2630059189459234e-06, "loss": 1.2433, "step": 5434 }, { "epoch": 0.4159651002602174, "grad_norm": 0.5674503057279089, "learning_rate": 7.2620650959255355e-06, "loss": 1.3581, "step": 5435 }, { "epoch": 0.41604163477728456, "grad_norm": 0.6250454777984107, "learning_rate": 7.261124195360096e-06, "loss": 1.0896, "step": 5436 }, { "epoch": 0.41611816929435175, "grad_norm": 0.7187236291612816, "learning_rate": 7.260183217307416e-06, "loss": 1.7274, "step": 5437 }, { "epoch": 0.41619470381141893, "grad_norm": 0.5678210202940994, "learning_rate": 7.2592421618253115e-06, "loss": 1.421, "step": 5438 }, { "epoch": 0.41627123832848617, "grad_norm": 0.8213599443478858, "learning_rate": 7.258301028971605e-06, "loss": 1.1846, "step": 5439 }, { "epoch": 0.41634777284555335, "grad_norm": 1.0395049952096371, "learning_rate": 7.2573598188041205e-06, "loss": 1.3737, "step": 5440 }, { "epoch": 0.41642430736262054, "grad_norm": 0.48320088650516557, "learning_rate": 7.256418531380689e-06, "loss": 1.0458, "step": 5441 }, { "epoch": 0.4165008418796877, "grad_norm": 0.5440348985745737, "learning_rate": 7.25547716675915e-06, "loss": 1.201, "step": 5442 }, { "epoch": 0.41657737639675496, "grad_norm": 0.5406933700338086, "learning_rate": 7.254535724997339e-06, "loss": 1.2125, "step": 5443 }, { "epoch": 0.41665391091382215, "grad_norm": 0.5709927189966401, "learning_rate": 7.253594206153103e-06, "loss": 1.2859, "step": 5444 }, { "epoch": 0.41673044543088933, "grad_norm": 0.6488196678896563, "learning_rate": 7.252652610284291e-06, "loss": 1.5858, "step": 5445 }, { "epoch": 0.4168069799479565, "grad_norm": 0.6511348130334298, "learning_rate": 7.251710937448757e-06, "loss": 1.6742, "step": 5446 }, { "epoch": 0.41688351446502375, "grad_norm": 0.5870829968722819, "learning_rate": 7.250769187704363e-06, "loss": 1.4123, "step": 5447 }, { "epoch": 0.41696004898209094, "grad_norm": 0.5790868489509854, "learning_rate": 7.249827361108971e-06, "loss": 1.3427, "step": 5448 }, { "epoch": 0.4170365834991581, "grad_norm": 0.519134297607777, "learning_rate": 7.248885457720448e-06, "loss": 1.3201, "step": 5449 }, { "epoch": 0.4171131180162253, "grad_norm": 0.5764505484602898, "learning_rate": 7.247943477596671e-06, "loss": 1.4322, "step": 5450 }, { "epoch": 0.4171896525332925, "grad_norm": 0.6690391001631435, "learning_rate": 7.247001420795514e-06, "loss": 1.6773, "step": 5451 }, { "epoch": 0.4172661870503597, "grad_norm": 0.6624981688509979, "learning_rate": 7.246059287374862e-06, "loss": 1.4116, "step": 5452 }, { "epoch": 0.4173427215674269, "grad_norm": 0.5715659335620883, "learning_rate": 7.2451170773926035e-06, "loss": 1.3772, "step": 5453 }, { "epoch": 0.4174192560844941, "grad_norm": 0.6020051156738476, "learning_rate": 7.2441747909066305e-06, "loss": 1.5715, "step": 5454 }, { "epoch": 0.4174957906015613, "grad_norm": 1.4502622290677227, "learning_rate": 7.243232427974838e-06, "loss": 1.2589, "step": 5455 }, { "epoch": 0.4175723251186285, "grad_norm": 0.6359647432398446, "learning_rate": 7.242289988655128e-06, "loss": 1.4768, "step": 5456 }, { "epoch": 0.4176488596356957, "grad_norm": 0.6894099890966463, "learning_rate": 7.241347473005408e-06, "loss": 1.746, "step": 5457 }, { "epoch": 0.4177253941527629, "grad_norm": 0.6245042038898605, "learning_rate": 7.240404881083588e-06, "loss": 1.6446, "step": 5458 }, { "epoch": 0.41780192866983007, "grad_norm": 0.6853014995732419, "learning_rate": 7.239462212947585e-06, "loss": 1.509, "step": 5459 }, { "epoch": 0.4178784631868973, "grad_norm": 0.6200264527246389, "learning_rate": 7.2385194686553165e-06, "loss": 1.4996, "step": 5460 }, { "epoch": 0.4179549977039645, "grad_norm": 0.6665931207920744, "learning_rate": 7.2375766482647105e-06, "loss": 1.6655, "step": 5461 }, { "epoch": 0.4180315322210317, "grad_norm": 0.5844755288462868, "learning_rate": 7.236633751833696e-06, "loss": 1.4251, "step": 5462 }, { "epoch": 0.41810806673809886, "grad_norm": 0.8152148100743132, "learning_rate": 7.235690779420204e-06, "loss": 1.5984, "step": 5463 }, { "epoch": 0.4181846012551661, "grad_norm": 0.5559884274968471, "learning_rate": 7.23474773108218e-06, "loss": 1.2926, "step": 5464 }, { "epoch": 0.4182611357722333, "grad_norm": 0.6232833871001069, "learning_rate": 7.233804606877563e-06, "loss": 1.6571, "step": 5465 }, { "epoch": 0.41833767028930047, "grad_norm": 0.588026558339332, "learning_rate": 7.232861406864303e-06, "loss": 1.377, "step": 5466 }, { "epoch": 0.41841420480636765, "grad_norm": 0.5589548576349582, "learning_rate": 7.231918131100353e-06, "loss": 1.2359, "step": 5467 }, { "epoch": 0.4184907393234349, "grad_norm": 0.7485543518662353, "learning_rate": 7.230974779643669e-06, "loss": 1.4648, "step": 5468 }, { "epoch": 0.4185672738405021, "grad_norm": 0.6094646919725955, "learning_rate": 7.230031352552216e-06, "loss": 1.5993, "step": 5469 }, { "epoch": 0.41864380835756926, "grad_norm": 0.5915964349763472, "learning_rate": 7.229087849883958e-06, "loss": 1.3271, "step": 5470 }, { "epoch": 0.41872034287463644, "grad_norm": 0.5623378724252174, "learning_rate": 7.228144271696869e-06, "loss": 1.3763, "step": 5471 }, { "epoch": 0.4187968773917037, "grad_norm": 0.6525007007920559, "learning_rate": 7.227200618048924e-06, "loss": 1.6161, "step": 5472 }, { "epoch": 0.41887341190877087, "grad_norm": 0.6547531537696104, "learning_rate": 7.2262568889981056e-06, "loss": 1.6327, "step": 5473 }, { "epoch": 0.41894994642583805, "grad_norm": 0.5972191827485062, "learning_rate": 7.225313084602397e-06, "loss": 1.5538, "step": 5474 }, { "epoch": 0.41902648094290523, "grad_norm": 0.5554487991444488, "learning_rate": 7.224369204919789e-06, "loss": 1.3178, "step": 5475 }, { "epoch": 0.41910301545997247, "grad_norm": 0.6030845637747702, "learning_rate": 7.223425250008279e-06, "loss": 1.2085, "step": 5476 }, { "epoch": 0.41917954997703966, "grad_norm": 0.5631273148650776, "learning_rate": 7.222481219925863e-06, "loss": 1.3019, "step": 5477 }, { "epoch": 0.41925608449410684, "grad_norm": 0.6027479229540197, "learning_rate": 7.221537114730547e-06, "loss": 1.3781, "step": 5478 }, { "epoch": 0.419332619011174, "grad_norm": 0.5023099513126501, "learning_rate": 7.220592934480338e-06, "loss": 1.1519, "step": 5479 }, { "epoch": 0.41940915352824126, "grad_norm": 0.5890444578473484, "learning_rate": 7.219648679233251e-06, "loss": 1.1763, "step": 5480 }, { "epoch": 0.41948568804530845, "grad_norm": 0.5772603601389171, "learning_rate": 7.218704349047302e-06, "loss": 1.3591, "step": 5481 }, { "epoch": 0.41956222256237563, "grad_norm": 0.6847702091764036, "learning_rate": 7.217759943980515e-06, "loss": 1.625, "step": 5482 }, { "epoch": 0.4196387570794428, "grad_norm": 0.5354943972732863, "learning_rate": 7.216815464090917e-06, "loss": 1.1216, "step": 5483 }, { "epoch": 0.41971529159651, "grad_norm": 0.531404745196151, "learning_rate": 7.21587090943654e-06, "loss": 1.2337, "step": 5484 }, { "epoch": 0.41979182611357724, "grad_norm": 0.722902153511301, "learning_rate": 7.214926280075417e-06, "loss": 1.5013, "step": 5485 }, { "epoch": 0.4198683606306444, "grad_norm": 0.5689778091228915, "learning_rate": 7.213981576065594e-06, "loss": 1.4407, "step": 5486 }, { "epoch": 0.4199448951477116, "grad_norm": 0.6325077235665053, "learning_rate": 7.213036797465112e-06, "loss": 1.3313, "step": 5487 }, { "epoch": 0.4200214296647788, "grad_norm": 0.6552049595963009, "learning_rate": 7.212091944332025e-06, "loss": 1.5679, "step": 5488 }, { "epoch": 0.42009796418184603, "grad_norm": 0.5672611965070594, "learning_rate": 7.211147016724385e-06, "loss": 1.2777, "step": 5489 }, { "epoch": 0.4201744986989132, "grad_norm": 0.5786790888032105, "learning_rate": 7.2102020147002505e-06, "loss": 1.4183, "step": 5490 }, { "epoch": 0.4202510332159804, "grad_norm": 0.7599604840507388, "learning_rate": 7.2092569383176865e-06, "loss": 1.6525, "step": 5491 }, { "epoch": 0.4203275677330476, "grad_norm": 0.6724704196299114, "learning_rate": 7.208311787634763e-06, "loss": 1.4398, "step": 5492 }, { "epoch": 0.4204041022501148, "grad_norm": 0.6359086503146242, "learning_rate": 7.207366562709549e-06, "loss": 1.3411, "step": 5493 }, { "epoch": 0.420480636767182, "grad_norm": 0.5390646248613725, "learning_rate": 7.206421263600125e-06, "loss": 1.3255, "step": 5494 }, { "epoch": 0.4205571712842492, "grad_norm": 0.7330626858561161, "learning_rate": 7.205475890364572e-06, "loss": 1.5228, "step": 5495 }, { "epoch": 0.42063370580131637, "grad_norm": 0.6196661973518098, "learning_rate": 7.204530443060976e-06, "loss": 1.4418, "step": 5496 }, { "epoch": 0.4207102403183836, "grad_norm": 0.8599732758807759, "learning_rate": 7.203584921747429e-06, "loss": 1.8018, "step": 5497 }, { "epoch": 0.4207867748354508, "grad_norm": 0.5691315180626669, "learning_rate": 7.202639326482026e-06, "loss": 1.255, "step": 5498 }, { "epoch": 0.420863309352518, "grad_norm": 0.6655671563557094, "learning_rate": 7.201693657322869e-06, "loss": 1.3473, "step": 5499 }, { "epoch": 0.42093984386958516, "grad_norm": 0.7892920976058136, "learning_rate": 7.20074791432806e-06, "loss": 1.4553, "step": 5500 }, { "epoch": 0.4210163783866524, "grad_norm": 0.5913388432917038, "learning_rate": 7.199802097555711e-06, "loss": 1.284, "step": 5501 }, { "epoch": 0.4210929129037196, "grad_norm": 0.5963626117715776, "learning_rate": 7.198856207063933e-06, "loss": 1.3486, "step": 5502 }, { "epoch": 0.42116944742078677, "grad_norm": 0.587378414942342, "learning_rate": 7.197910242910847e-06, "loss": 1.2535, "step": 5503 }, { "epoch": 0.42124598193785395, "grad_norm": 0.5940666574380268, "learning_rate": 7.196964205154573e-06, "loss": 1.4412, "step": 5504 }, { "epoch": 0.4213225164549212, "grad_norm": 1.2928525558167474, "learning_rate": 7.1960180938532405e-06, "loss": 1.9074, "step": 5505 }, { "epoch": 0.4213990509719884, "grad_norm": 0.6749811768081199, "learning_rate": 7.195071909064982e-06, "loss": 1.7809, "step": 5506 }, { "epoch": 0.42147558548905556, "grad_norm": 0.582109757567084, "learning_rate": 7.194125650847931e-06, "loss": 1.4261, "step": 5507 }, { "epoch": 0.42155212000612274, "grad_norm": 0.5681258303510032, "learning_rate": 7.1931793192602314e-06, "loss": 1.377, "step": 5508 }, { "epoch": 0.42162865452319, "grad_norm": 0.5784925256162519, "learning_rate": 7.192232914360026e-06, "loss": 1.1906, "step": 5509 }, { "epoch": 0.42170518904025717, "grad_norm": 0.5451265926890678, "learning_rate": 7.191286436205468e-06, "loss": 1.2975, "step": 5510 }, { "epoch": 0.42178172355732435, "grad_norm": 0.5905824561660791, "learning_rate": 7.19033988485471e-06, "loss": 1.2676, "step": 5511 }, { "epoch": 0.42185825807439153, "grad_norm": 0.6247200015091755, "learning_rate": 7.189393260365911e-06, "loss": 1.0923, "step": 5512 }, { "epoch": 0.4219347925914588, "grad_norm": 0.5803120369491037, "learning_rate": 7.188446562797235e-06, "loss": 1.4008, "step": 5513 }, { "epoch": 0.42201132710852596, "grad_norm": 0.5469006009118297, "learning_rate": 7.187499792206849e-06, "loss": 1.2471, "step": 5514 }, { "epoch": 0.42208786162559314, "grad_norm": 0.6044014613009798, "learning_rate": 7.186552948652924e-06, "loss": 1.5482, "step": 5515 }, { "epoch": 0.4221643961426603, "grad_norm": 0.5357695093233136, "learning_rate": 7.18560603219364e-06, "loss": 1.4342, "step": 5516 }, { "epoch": 0.42224093065972756, "grad_norm": 0.6189329304785902, "learning_rate": 7.184659042887178e-06, "loss": 1.4886, "step": 5517 }, { "epoch": 0.42231746517679475, "grad_norm": 0.5566167540756635, "learning_rate": 7.183711980791722e-06, "loss": 1.2298, "step": 5518 }, { "epoch": 0.42239399969386193, "grad_norm": 0.54304813571722, "learning_rate": 7.182764845965465e-06, "loss": 1.2412, "step": 5519 }, { "epoch": 0.4224705342109291, "grad_norm": 0.5416024588021346, "learning_rate": 7.181817638466597e-06, "loss": 1.3144, "step": 5520 }, { "epoch": 0.4225470687279963, "grad_norm": 0.9255423996229432, "learning_rate": 7.180870358353322e-06, "loss": 1.6513, "step": 5521 }, { "epoch": 0.42262360324506354, "grad_norm": 0.8728267612825492, "learning_rate": 7.179923005683842e-06, "loss": 1.4633, "step": 5522 }, { "epoch": 0.4227001377621307, "grad_norm": 0.5298325653624193, "learning_rate": 7.178975580516365e-06, "loss": 1.2422, "step": 5523 }, { "epoch": 0.4227766722791979, "grad_norm": 1.2322341847632738, "learning_rate": 7.178028082909105e-06, "loss": 1.4929, "step": 5524 }, { "epoch": 0.4228532067962651, "grad_norm": 0.5533261445233643, "learning_rate": 7.177080512920277e-06, "loss": 1.2236, "step": 5525 }, { "epoch": 0.42292974131333233, "grad_norm": 0.5060341164912311, "learning_rate": 7.1761328706081025e-06, "loss": 1.1218, "step": 5526 }, { "epoch": 0.4230062758303995, "grad_norm": 1.0301900900326575, "learning_rate": 7.175185156030809e-06, "loss": 1.3699, "step": 5527 }, { "epoch": 0.4230828103474667, "grad_norm": 0.5980181720527074, "learning_rate": 7.174237369246628e-06, "loss": 1.3422, "step": 5528 }, { "epoch": 0.4231593448645339, "grad_norm": 0.7585485445833802, "learning_rate": 7.17328951031379e-06, "loss": 1.5049, "step": 5529 }, { "epoch": 0.4232358793816011, "grad_norm": 0.6023897008751478, "learning_rate": 7.172341579290539e-06, "loss": 1.3925, "step": 5530 }, { "epoch": 0.4233124138986683, "grad_norm": 0.486982759886755, "learning_rate": 7.171393576235116e-06, "loss": 1.0486, "step": 5531 }, { "epoch": 0.4233889484157355, "grad_norm": 0.5716624327018596, "learning_rate": 7.17044550120577e-06, "loss": 1.479, "step": 5532 }, { "epoch": 0.4234654829328027, "grad_norm": 0.6406468318398568, "learning_rate": 7.169497354260754e-06, "loss": 1.5473, "step": 5533 }, { "epoch": 0.4235420174498699, "grad_norm": 0.9809178531134629, "learning_rate": 7.1685491354583245e-06, "loss": 1.5346, "step": 5534 }, { "epoch": 0.4236185519669371, "grad_norm": 0.5404972904598168, "learning_rate": 7.167600844856742e-06, "loss": 1.3237, "step": 5535 }, { "epoch": 0.4236950864840043, "grad_norm": 0.6518732803963934, "learning_rate": 7.166652482514275e-06, "loss": 1.52, "step": 5536 }, { "epoch": 0.42377162100107146, "grad_norm": 0.5218905719842704, "learning_rate": 7.16570404848919e-06, "loss": 1.14, "step": 5537 }, { "epoch": 0.4238481555181387, "grad_norm": 0.6947119141773671, "learning_rate": 7.164755542839763e-06, "loss": 1.5124, "step": 5538 }, { "epoch": 0.4239246900352059, "grad_norm": 0.5565985343555127, "learning_rate": 7.163806965624275e-06, "loss": 1.2732, "step": 5539 }, { "epoch": 0.42400122455227307, "grad_norm": 1.0598817472159743, "learning_rate": 7.1628583169010055e-06, "loss": 1.5282, "step": 5540 }, { "epoch": 0.42407775906934025, "grad_norm": 0.5771257532817148, "learning_rate": 7.161909596728247e-06, "loss": 1.3516, "step": 5541 }, { "epoch": 0.4241542935864075, "grad_norm": 0.6526336527455021, "learning_rate": 7.1609608051642885e-06, "loss": 1.6228, "step": 5542 }, { "epoch": 0.4242308281034747, "grad_norm": 0.5455530348582904, "learning_rate": 7.160011942267427e-06, "loss": 1.2593, "step": 5543 }, { "epoch": 0.42430736262054186, "grad_norm": 0.8819348528399521, "learning_rate": 7.1590630080959666e-06, "loss": 1.4964, "step": 5544 }, { "epoch": 0.42438389713760905, "grad_norm": 0.6715883397490141, "learning_rate": 7.158114002708208e-06, "loss": 1.4594, "step": 5545 }, { "epoch": 0.4244604316546763, "grad_norm": 0.4958472845045979, "learning_rate": 7.157164926162463e-06, "loss": 1.0398, "step": 5546 }, { "epoch": 0.42453696617174347, "grad_norm": 0.5922338467564395, "learning_rate": 7.156215778517046e-06, "loss": 1.5108, "step": 5547 }, { "epoch": 0.42461350068881065, "grad_norm": 0.5401321229874925, "learning_rate": 7.155266559830275e-06, "loss": 1.1517, "step": 5548 }, { "epoch": 0.42469003520587784, "grad_norm": 0.633117758618084, "learning_rate": 7.154317270160473e-06, "loss": 1.5224, "step": 5549 }, { "epoch": 0.4247665697229451, "grad_norm": 0.5732869162368939, "learning_rate": 7.153367909565968e-06, "loss": 1.4638, "step": 5550 }, { "epoch": 0.42484310424001226, "grad_norm": 0.7573417501155911, "learning_rate": 7.152418478105089e-06, "loss": 1.6446, "step": 5551 }, { "epoch": 0.42491963875707944, "grad_norm": 0.6675129672530279, "learning_rate": 7.1514689758361755e-06, "loss": 1.6161, "step": 5552 }, { "epoch": 0.4249961732741466, "grad_norm": 0.6755701176386966, "learning_rate": 7.150519402817564e-06, "loss": 1.5608, "step": 5553 }, { "epoch": 0.4250727077912138, "grad_norm": 0.5034996419102344, "learning_rate": 7.149569759107602e-06, "loss": 1.1274, "step": 5554 }, { "epoch": 0.42514924230828105, "grad_norm": 0.5857466338814702, "learning_rate": 7.148620044764638e-06, "loss": 1.2114, "step": 5555 }, { "epoch": 0.42522577682534823, "grad_norm": 0.6544913316290422, "learning_rate": 7.147670259847024e-06, "loss": 1.5571, "step": 5556 }, { "epoch": 0.4253023113424154, "grad_norm": 0.5693362781448159, "learning_rate": 7.146720404413119e-06, "loss": 1.1807, "step": 5557 }, { "epoch": 0.4253788458594826, "grad_norm": 0.5977137796891746, "learning_rate": 7.1457704785212845e-06, "loss": 1.346, "step": 5558 }, { "epoch": 0.42545538037654984, "grad_norm": 0.5311663263786641, "learning_rate": 7.144820482229884e-06, "loss": 1.2496, "step": 5559 }, { "epoch": 0.425531914893617, "grad_norm": 0.5354282395588277, "learning_rate": 7.143870415597294e-06, "loss": 1.1586, "step": 5560 }, { "epoch": 0.4256084494106842, "grad_norm": 0.5979973334160084, "learning_rate": 7.142920278681883e-06, "loss": 1.4533, "step": 5561 }, { "epoch": 0.4256849839277514, "grad_norm": 0.6520619461173391, "learning_rate": 7.141970071542036e-06, "loss": 1.4781, "step": 5562 }, { "epoch": 0.42576151844481863, "grad_norm": 0.6466426625476431, "learning_rate": 7.141019794236134e-06, "loss": 1.8326, "step": 5563 }, { "epoch": 0.4258380529618858, "grad_norm": 0.5592265631955705, "learning_rate": 7.140069446822563e-06, "loss": 1.2213, "step": 5564 }, { "epoch": 0.425914587478953, "grad_norm": 0.5813178090403717, "learning_rate": 7.139119029359718e-06, "loss": 1.6133, "step": 5565 }, { "epoch": 0.4259911219960202, "grad_norm": 0.5987947754670031, "learning_rate": 7.138168541905994e-06, "loss": 1.4871, "step": 5566 }, { "epoch": 0.4260676565130874, "grad_norm": 0.5677310780845173, "learning_rate": 7.1372179845197945e-06, "loss": 1.3493, "step": 5567 }, { "epoch": 0.4261441910301546, "grad_norm": 0.540894267923069, "learning_rate": 7.13626735725952e-06, "loss": 1.5172, "step": 5568 }, { "epoch": 0.4262207255472218, "grad_norm": 0.46983443526614616, "learning_rate": 7.135316660183583e-06, "loss": 0.8628, "step": 5569 }, { "epoch": 0.426297260064289, "grad_norm": 0.7699263107865988, "learning_rate": 7.134365893350396e-06, "loss": 1.7397, "step": 5570 }, { "epoch": 0.4263737945813562, "grad_norm": 0.4704250353911184, "learning_rate": 7.133415056818378e-06, "loss": 1.0679, "step": 5571 }, { "epoch": 0.4264503290984234, "grad_norm": 0.6155244325938409, "learning_rate": 7.13246415064595e-06, "loss": 1.5139, "step": 5572 }, { "epoch": 0.4265268636154906, "grad_norm": 0.557995846055549, "learning_rate": 7.13151317489154e-06, "loss": 1.3369, "step": 5573 }, { "epoch": 0.42660339813255777, "grad_norm": 0.5353279653951221, "learning_rate": 7.130562129613577e-06, "loss": 1.2277, "step": 5574 }, { "epoch": 0.426679932649625, "grad_norm": 0.6601707785532084, "learning_rate": 7.129611014870498e-06, "loss": 1.6663, "step": 5575 }, { "epoch": 0.4267564671666922, "grad_norm": 0.5531301277195732, "learning_rate": 7.12865983072074e-06, "loss": 1.2986, "step": 5576 }, { "epoch": 0.42683300168375937, "grad_norm": 0.5519319659788452, "learning_rate": 7.127708577222751e-06, "loss": 1.3492, "step": 5577 }, { "epoch": 0.42690953620082656, "grad_norm": 0.4967219703477866, "learning_rate": 7.126757254434973e-06, "loss": 1.1322, "step": 5578 }, { "epoch": 0.4269860707178938, "grad_norm": 0.6336245674592936, "learning_rate": 7.125805862415861e-06, "loss": 1.3544, "step": 5579 }, { "epoch": 0.427062605234961, "grad_norm": 0.6090390254114675, "learning_rate": 7.124854401223873e-06, "loss": 1.5111, "step": 5580 }, { "epoch": 0.42713913975202816, "grad_norm": 0.80561351853177, "learning_rate": 7.123902870917466e-06, "loss": 1.3988, "step": 5581 }, { "epoch": 0.42721567426909535, "grad_norm": 0.5781756147700176, "learning_rate": 7.122951271555108e-06, "loss": 1.3793, "step": 5582 }, { "epoch": 0.4272922087861626, "grad_norm": 0.5813950942457686, "learning_rate": 7.121999603195266e-06, "loss": 1.4846, "step": 5583 }, { "epoch": 0.42736874330322977, "grad_norm": 0.5881043632545007, "learning_rate": 7.121047865896415e-06, "loss": 1.3824, "step": 5584 }, { "epoch": 0.42744527782029695, "grad_norm": 0.6815109175184374, "learning_rate": 7.120096059717032e-06, "loss": 1.8214, "step": 5585 }, { "epoch": 0.42752181233736414, "grad_norm": 0.6125320286928934, "learning_rate": 7.1191441847155975e-06, "loss": 1.4014, "step": 5586 }, { "epoch": 0.4275983468544313, "grad_norm": 0.555895916948623, "learning_rate": 7.1181922409505985e-06, "loss": 1.387, "step": 5587 }, { "epoch": 0.42767488137149856, "grad_norm": 0.6290908738376098, "learning_rate": 7.117240228480528e-06, "loss": 1.3284, "step": 5588 }, { "epoch": 0.42775141588856574, "grad_norm": 0.6174669667506849, "learning_rate": 7.1162881473638754e-06, "loss": 1.4961, "step": 5589 }, { "epoch": 0.42782795040563293, "grad_norm": 0.7164983467923292, "learning_rate": 7.115335997659143e-06, "loss": 1.5541, "step": 5590 }, { "epoch": 0.4279044849227001, "grad_norm": 0.5810450722404842, "learning_rate": 7.114383779424834e-06, "loss": 1.4086, "step": 5591 }, { "epoch": 0.42798101943976735, "grad_norm": 0.6452535167899318, "learning_rate": 7.113431492719452e-06, "loss": 1.4637, "step": 5592 }, { "epoch": 0.42805755395683454, "grad_norm": 0.6361721600006501, "learning_rate": 7.112479137601511e-06, "loss": 1.3324, "step": 5593 }, { "epoch": 0.4281340884739017, "grad_norm": 0.5976422747592867, "learning_rate": 7.111526714129526e-06, "loss": 1.6738, "step": 5594 }, { "epoch": 0.4282106229909689, "grad_norm": 0.9274979017577422, "learning_rate": 7.110574222362018e-06, "loss": 1.553, "step": 5595 }, { "epoch": 0.42828715750803614, "grad_norm": 0.6064614465685763, "learning_rate": 7.1096216623575086e-06, "loss": 1.387, "step": 5596 }, { "epoch": 0.4283636920251033, "grad_norm": 0.602580634854131, "learning_rate": 7.108669034174528e-06, "loss": 1.5072, "step": 5597 }, { "epoch": 0.4284402265421705, "grad_norm": 0.5712142244023009, "learning_rate": 7.107716337871606e-06, "loss": 1.5112, "step": 5598 }, { "epoch": 0.4285167610592377, "grad_norm": 0.5857210869800784, "learning_rate": 7.106763573507282e-06, "loss": 1.3844, "step": 5599 }, { "epoch": 0.42859329557630493, "grad_norm": 0.5661622515651935, "learning_rate": 7.1058107411400956e-06, "loss": 1.3827, "step": 5600 }, { "epoch": 0.4286698300933721, "grad_norm": 0.6041146975526278, "learning_rate": 7.104857840828592e-06, "loss": 1.4983, "step": 5601 }, { "epoch": 0.4287463646104393, "grad_norm": 0.5230037431513466, "learning_rate": 7.1039048726313205e-06, "loss": 1.15, "step": 5602 }, { "epoch": 0.4288228991275065, "grad_norm": 0.5103048329175258, "learning_rate": 7.102951836606832e-06, "loss": 1.0957, "step": 5603 }, { "epoch": 0.4288994336445737, "grad_norm": 0.7770118446876116, "learning_rate": 7.101998732813685e-06, "loss": 1.6319, "step": 5604 }, { "epoch": 0.4289759681616409, "grad_norm": 0.6208111401398398, "learning_rate": 7.1010455613104425e-06, "loss": 1.5467, "step": 5605 }, { "epoch": 0.4290525026787081, "grad_norm": 0.6162317695301099, "learning_rate": 7.1000923221556685e-06, "loss": 1.5774, "step": 5606 }, { "epoch": 0.4291290371957753, "grad_norm": 0.5579101956142071, "learning_rate": 7.0991390154079346e-06, "loss": 1.2723, "step": 5607 }, { "epoch": 0.4292055717128425, "grad_norm": 0.6647881673144516, "learning_rate": 7.098185641125812e-06, "loss": 1.3482, "step": 5608 }, { "epoch": 0.4292821062299097, "grad_norm": 0.5395268846897364, "learning_rate": 7.0972321993678825e-06, "loss": 1.2478, "step": 5609 }, { "epoch": 0.4293586407469769, "grad_norm": 0.6097038466948328, "learning_rate": 7.096278690192726e-06, "loss": 1.389, "step": 5610 }, { "epoch": 0.42943517526404407, "grad_norm": 0.701323104395292, "learning_rate": 7.0953251136589285e-06, "loss": 1.8223, "step": 5611 }, { "epoch": 0.4295117097811113, "grad_norm": 0.5531508741601759, "learning_rate": 7.094371469825083e-06, "loss": 1.2297, "step": 5612 }, { "epoch": 0.4295882442981785, "grad_norm": 0.5491914350245617, "learning_rate": 7.093417758749783e-06, "loss": 1.2282, "step": 5613 }, { "epoch": 0.4296647788152457, "grad_norm": 2.642961616050967, "learning_rate": 7.092463980491626e-06, "loss": 1.5883, "step": 5614 }, { "epoch": 0.42974131333231286, "grad_norm": 0.7772261485560417, "learning_rate": 7.091510135109217e-06, "loss": 1.542, "step": 5615 }, { "epoch": 0.4298178478493801, "grad_norm": 0.6150913603647847, "learning_rate": 7.090556222661163e-06, "loss": 1.5388, "step": 5616 }, { "epoch": 0.4298943823664473, "grad_norm": 0.5485335730911105, "learning_rate": 7.0896022432060726e-06, "loss": 1.3321, "step": 5617 }, { "epoch": 0.42997091688351446, "grad_norm": 0.7518709480503699, "learning_rate": 7.088648196802565e-06, "loss": 1.6966, "step": 5618 }, { "epoch": 0.43004745140058165, "grad_norm": 0.5306656618896685, "learning_rate": 7.0876940835092554e-06, "loss": 1.1042, "step": 5619 }, { "epoch": 0.43012398591764883, "grad_norm": 0.5452329183062682, "learning_rate": 7.086739903384773e-06, "loss": 1.2333, "step": 5620 }, { "epoch": 0.43020052043471607, "grad_norm": 0.6219388849225167, "learning_rate": 7.085785656487739e-06, "loss": 1.5746, "step": 5621 }, { "epoch": 0.43027705495178326, "grad_norm": 0.5839403611184735, "learning_rate": 7.08483134287679e-06, "loss": 1.1969, "step": 5622 }, { "epoch": 0.43035358946885044, "grad_norm": 0.770880460988862, "learning_rate": 7.083876962610561e-06, "loss": 1.6404, "step": 5623 }, { "epoch": 0.4304301239859176, "grad_norm": 0.6471514258053469, "learning_rate": 7.0829225157476896e-06, "loss": 1.3769, "step": 5624 }, { "epoch": 0.43050665850298486, "grad_norm": 0.4939277452792505, "learning_rate": 7.081968002346823e-06, "loss": 0.8981, "step": 5625 }, { "epoch": 0.43058319302005205, "grad_norm": 0.6295654589453629, "learning_rate": 7.081013422466606e-06, "loss": 1.7433, "step": 5626 }, { "epoch": 0.43065972753711923, "grad_norm": 0.8059201149436229, "learning_rate": 7.080058776165694e-06, "loss": 1.643, "step": 5627 }, { "epoch": 0.4307362620541864, "grad_norm": 0.7485332272262822, "learning_rate": 7.079104063502742e-06, "loss": 1.2649, "step": 5628 }, { "epoch": 0.43081279657125365, "grad_norm": 0.7588606798232538, "learning_rate": 7.07814928453641e-06, "loss": 1.7031, "step": 5629 }, { "epoch": 0.43088933108832084, "grad_norm": 0.6142526335323035, "learning_rate": 7.077194439325362e-06, "loss": 1.61, "step": 5630 }, { "epoch": 0.430965865605388, "grad_norm": 1.1828328853887264, "learning_rate": 7.076239527928269e-06, "loss": 1.3168, "step": 5631 }, { "epoch": 0.4310424001224552, "grad_norm": 0.5690659204147382, "learning_rate": 7.0752845504038025e-06, "loss": 1.342, "step": 5632 }, { "epoch": 0.43111893463952244, "grad_norm": 0.5385975983482745, "learning_rate": 7.074329506810638e-06, "loss": 1.1158, "step": 5633 }, { "epoch": 0.43119546915658963, "grad_norm": 0.6329695202514204, "learning_rate": 7.073374397207456e-06, "loss": 1.3823, "step": 5634 }, { "epoch": 0.4312720036736568, "grad_norm": 0.6017755870214403, "learning_rate": 7.072419221652944e-06, "loss": 1.6145, "step": 5635 }, { "epoch": 0.431348538190724, "grad_norm": 0.7660457919285405, "learning_rate": 7.0714639802057885e-06, "loss": 1.5123, "step": 5636 }, { "epoch": 0.43142507270779124, "grad_norm": 0.6490722120671235, "learning_rate": 7.0705086729246834e-06, "loss": 1.4564, "step": 5637 }, { "epoch": 0.4315016072248584, "grad_norm": 0.5488410511250832, "learning_rate": 7.069553299868323e-06, "loss": 1.319, "step": 5638 }, { "epoch": 0.4315781417419256, "grad_norm": 0.5700007656701556, "learning_rate": 7.068597861095413e-06, "loss": 1.5043, "step": 5639 }, { "epoch": 0.4316546762589928, "grad_norm": 0.5559482132422562, "learning_rate": 7.067642356664655e-06, "loss": 1.345, "step": 5640 }, { "epoch": 0.43173121077606, "grad_norm": 0.5990744872200573, "learning_rate": 7.066686786634759e-06, "loss": 1.5753, "step": 5641 }, { "epoch": 0.4318077452931272, "grad_norm": 0.6237058631424242, "learning_rate": 7.065731151064438e-06, "loss": 1.351, "step": 5642 }, { "epoch": 0.4318842798101944, "grad_norm": 0.5614956171087855, "learning_rate": 7.064775450012409e-06, "loss": 1.2282, "step": 5643 }, { "epoch": 0.4319608143272616, "grad_norm": 0.6166437003226238, "learning_rate": 7.063819683537393e-06, "loss": 1.3527, "step": 5644 }, { "epoch": 0.4320373488443288, "grad_norm": 0.5707128357533997, "learning_rate": 7.062863851698115e-06, "loss": 1.2417, "step": 5645 }, { "epoch": 0.432113883361396, "grad_norm": 0.6182700824164452, "learning_rate": 7.061907954553306e-06, "loss": 1.3029, "step": 5646 }, { "epoch": 0.4321904178784632, "grad_norm": 0.6846829394578657, "learning_rate": 7.060951992161698e-06, "loss": 1.5762, "step": 5647 }, { "epoch": 0.43226695239553037, "grad_norm": 0.5865727999431453, "learning_rate": 7.059995964582027e-06, "loss": 1.3699, "step": 5648 }, { "epoch": 0.4323434869125976, "grad_norm": 0.7731001798034177, "learning_rate": 7.059039871873034e-06, "loss": 1.8437, "step": 5649 }, { "epoch": 0.4324200214296648, "grad_norm": 0.5643119584132691, "learning_rate": 7.058083714093468e-06, "loss": 1.493, "step": 5650 }, { "epoch": 0.432496555946732, "grad_norm": 0.6588833077692636, "learning_rate": 7.057127491302075e-06, "loss": 1.5802, "step": 5651 }, { "epoch": 0.43257309046379916, "grad_norm": 0.6962792346389565, "learning_rate": 7.056171203557606e-06, "loss": 1.9088, "step": 5652 }, { "epoch": 0.43264962498086634, "grad_norm": 0.46418043296881556, "learning_rate": 7.055214850918823e-06, "loss": 1.0076, "step": 5653 }, { "epoch": 0.4327261594979336, "grad_norm": 0.5433394286388951, "learning_rate": 7.0542584334444855e-06, "loss": 1.239, "step": 5654 }, { "epoch": 0.43280269401500077, "grad_norm": 0.5634935276659808, "learning_rate": 7.053301951193356e-06, "loss": 1.175, "step": 5655 }, { "epoch": 0.43287922853206795, "grad_norm": 0.4903847398747645, "learning_rate": 7.052345404224209e-06, "loss": 1.1032, "step": 5656 }, { "epoch": 0.43295576304913513, "grad_norm": 0.5642446127293578, "learning_rate": 7.051388792595812e-06, "loss": 1.3421, "step": 5657 }, { "epoch": 0.4330322975662024, "grad_norm": 0.9167976590199948, "learning_rate": 7.050432116366947e-06, "loss": 1.6815, "step": 5658 }, { "epoch": 0.43310883208326956, "grad_norm": 0.5959625902407076, "learning_rate": 7.049475375596391e-06, "loss": 1.5044, "step": 5659 }, { "epoch": 0.43318536660033674, "grad_norm": 0.536387407818323, "learning_rate": 7.0485185703429295e-06, "loss": 1.1777, "step": 5660 }, { "epoch": 0.4332619011174039, "grad_norm": 0.6923846332358913, "learning_rate": 7.047561700665354e-06, "loss": 1.3291, "step": 5661 }, { "epoch": 0.43333843563447116, "grad_norm": 0.7413502782483542, "learning_rate": 7.046604766622455e-06, "loss": 1.6925, "step": 5662 }, { "epoch": 0.43341497015153835, "grad_norm": 0.614320655178954, "learning_rate": 7.0456477682730305e-06, "loss": 1.4059, "step": 5663 }, { "epoch": 0.43349150466860553, "grad_norm": 0.5285306248255913, "learning_rate": 7.0446907056758825e-06, "loss": 1.0996, "step": 5664 }, { "epoch": 0.4335680391856727, "grad_norm": 0.5584710529495605, "learning_rate": 7.0437335788898145e-06, "loss": 1.2366, "step": 5665 }, { "epoch": 0.43364457370273995, "grad_norm": 0.6048361196946622, "learning_rate": 7.042776387973634e-06, "loss": 1.2707, "step": 5666 }, { "epoch": 0.43372110821980714, "grad_norm": 0.6681835902715976, "learning_rate": 7.041819132986155e-06, "loss": 1.6927, "step": 5667 }, { "epoch": 0.4337976427368743, "grad_norm": 0.7140844384941879, "learning_rate": 7.0408618139861955e-06, "loss": 1.6043, "step": 5668 }, { "epoch": 0.4338741772539415, "grad_norm": 0.5838112428696439, "learning_rate": 7.0399044310325725e-06, "loss": 1.4093, "step": 5669 }, { "epoch": 0.43395071177100875, "grad_norm": 0.6744973201908555, "learning_rate": 7.038946984184114e-06, "loss": 1.5327, "step": 5670 }, { "epoch": 0.43402724628807593, "grad_norm": 0.6116328964927282, "learning_rate": 7.0379894734996466e-06, "loss": 1.3791, "step": 5671 }, { "epoch": 0.4341037808051431, "grad_norm": 0.553171422722115, "learning_rate": 7.0370318990380025e-06, "loss": 1.3056, "step": 5672 }, { "epoch": 0.4341803153222103, "grad_norm": 0.5911152051285141, "learning_rate": 7.036074260858019e-06, "loss": 1.5379, "step": 5673 }, { "epoch": 0.43425684983927754, "grad_norm": 0.5016051481926193, "learning_rate": 7.035116559018534e-06, "loss": 1.0843, "step": 5674 }, { "epoch": 0.4343333843563447, "grad_norm": 0.5091238968408509, "learning_rate": 7.034158793578396e-06, "loss": 1.0914, "step": 5675 }, { "epoch": 0.4344099188734119, "grad_norm": 0.5448031298996473, "learning_rate": 7.033200964596448e-06, "loss": 1.4301, "step": 5676 }, { "epoch": 0.4344864533904791, "grad_norm": 0.7112879229779899, "learning_rate": 7.032243072131545e-06, "loss": 1.4961, "step": 5677 }, { "epoch": 0.4345629879075463, "grad_norm": 0.6263549705480337, "learning_rate": 7.0312851162425425e-06, "loss": 1.5936, "step": 5678 }, { "epoch": 0.4346395224246135, "grad_norm": 0.5732110884057745, "learning_rate": 7.030327096988297e-06, "loss": 1.4082, "step": 5679 }, { "epoch": 0.4347160569416807, "grad_norm": 0.786437175715752, "learning_rate": 7.0293690144276776e-06, "loss": 1.7581, "step": 5680 }, { "epoch": 0.4347925914587479, "grad_norm": 0.5683624727844527, "learning_rate": 7.028410868619547e-06, "loss": 1.3467, "step": 5681 }, { "epoch": 0.4348691259758151, "grad_norm": 0.5140321218084725, "learning_rate": 7.027452659622779e-06, "loss": 1.0966, "step": 5682 }, { "epoch": 0.4349456604928823, "grad_norm": 0.5809597012745967, "learning_rate": 7.026494387496247e-06, "loss": 1.4063, "step": 5683 }, { "epoch": 0.4350221950099495, "grad_norm": 0.5775858863731076, "learning_rate": 7.025536052298832e-06, "loss": 1.3556, "step": 5684 }, { "epoch": 0.43509872952701667, "grad_norm": 0.6499819724170605, "learning_rate": 7.024577654089416e-06, "loss": 1.513, "step": 5685 }, { "epoch": 0.4351752640440839, "grad_norm": 0.5758167425216066, "learning_rate": 7.023619192926886e-06, "loss": 1.4336, "step": 5686 }, { "epoch": 0.4352517985611511, "grad_norm": 0.6865029077594751, "learning_rate": 7.022660668870133e-06, "loss": 1.7186, "step": 5687 }, { "epoch": 0.4353283330782183, "grad_norm": 0.603786502287049, "learning_rate": 7.0217020819780514e-06, "loss": 1.3766, "step": 5688 }, { "epoch": 0.43540486759528546, "grad_norm": 0.5202319575563052, "learning_rate": 7.020743432309539e-06, "loss": 1.1703, "step": 5689 }, { "epoch": 0.43548140211235264, "grad_norm": 0.8646430942106368, "learning_rate": 7.019784719923498e-06, "loss": 1.5884, "step": 5690 }, { "epoch": 0.4355579366294199, "grad_norm": 0.5334191609572957, "learning_rate": 7.018825944878838e-06, "loss": 1.1777, "step": 5691 }, { "epoch": 0.43563447114648707, "grad_norm": 0.6659173228590215, "learning_rate": 7.017867107234464e-06, "loss": 1.4448, "step": 5692 }, { "epoch": 0.43571100566355425, "grad_norm": 1.241474834218897, "learning_rate": 7.0169082070492926e-06, "loss": 1.583, "step": 5693 }, { "epoch": 0.43578754018062144, "grad_norm": 0.5720228082735463, "learning_rate": 7.01594924438224e-06, "loss": 1.413, "step": 5694 }, { "epoch": 0.4358640746976887, "grad_norm": 0.4866665151039648, "learning_rate": 7.0149902192922306e-06, "loss": 1.0967, "step": 5695 }, { "epoch": 0.43594060921475586, "grad_norm": 0.5756009613631091, "learning_rate": 7.014031131838186e-06, "loss": 1.3939, "step": 5696 }, { "epoch": 0.43601714373182304, "grad_norm": 0.7069434709687342, "learning_rate": 7.013071982079039e-06, "loss": 1.6812, "step": 5697 }, { "epoch": 0.4360936782488902, "grad_norm": 0.4899230897930891, "learning_rate": 7.01211277007372e-06, "loss": 1.2015, "step": 5698 }, { "epoch": 0.43617021276595747, "grad_norm": 0.6038095570928812, "learning_rate": 7.011153495881165e-06, "loss": 1.2376, "step": 5699 }, { "epoch": 0.43624674728302465, "grad_norm": 0.6425184372075348, "learning_rate": 7.01019415956032e-06, "loss": 1.5954, "step": 5700 }, { "epoch": 0.43632328180009183, "grad_norm": 0.5826503062495167, "learning_rate": 7.009234761170124e-06, "loss": 1.5385, "step": 5701 }, { "epoch": 0.436399816317159, "grad_norm": 0.530424821565289, "learning_rate": 7.008275300769528e-06, "loss": 1.2799, "step": 5702 }, { "epoch": 0.43647635083422626, "grad_norm": 0.5831767643273065, "learning_rate": 7.007315778417483e-06, "loss": 1.4553, "step": 5703 }, { "epoch": 0.43655288535129344, "grad_norm": 0.6085369111656181, "learning_rate": 7.006356194172945e-06, "loss": 1.4876, "step": 5704 }, { "epoch": 0.4366294198683606, "grad_norm": 0.5895693377191925, "learning_rate": 7.005396548094874e-06, "loss": 1.4973, "step": 5705 }, { "epoch": 0.4367059543854278, "grad_norm": 0.5279555188378565, "learning_rate": 7.004436840242234e-06, "loss": 1.2636, "step": 5706 }, { "epoch": 0.43678248890249505, "grad_norm": 0.9324377124653802, "learning_rate": 7.003477070673991e-06, "loss": 1.6537, "step": 5707 }, { "epoch": 0.43685902341956223, "grad_norm": 0.5055454674816368, "learning_rate": 7.002517239449118e-06, "loss": 1.1987, "step": 5708 }, { "epoch": 0.4369355579366294, "grad_norm": 0.6053979036425455, "learning_rate": 7.0015573466265885e-06, "loss": 1.4209, "step": 5709 }, { "epoch": 0.4370120924536966, "grad_norm": 0.7968093504654261, "learning_rate": 7.00059739226538e-06, "loss": 1.4647, "step": 5710 }, { "epoch": 0.43708862697076384, "grad_norm": 0.5672747679638804, "learning_rate": 6.999637376424479e-06, "loss": 1.0616, "step": 5711 }, { "epoch": 0.437165161487831, "grad_norm": 1.0497281583899214, "learning_rate": 6.998677299162867e-06, "loss": 1.5375, "step": 5712 }, { "epoch": 0.4372416960048982, "grad_norm": 0.5704851175421392, "learning_rate": 6.9977171605395375e-06, "loss": 1.3053, "step": 5713 }, { "epoch": 0.4373182305219654, "grad_norm": 0.5947987563582685, "learning_rate": 6.996756960613484e-06, "loss": 1.3499, "step": 5714 }, { "epoch": 0.43739476503903263, "grad_norm": 0.6310793586937924, "learning_rate": 6.995796699443701e-06, "loss": 1.5511, "step": 5715 }, { "epoch": 0.4374712995560998, "grad_norm": 0.6083077413596223, "learning_rate": 6.994836377089194e-06, "loss": 1.3046, "step": 5716 }, { "epoch": 0.437547834073167, "grad_norm": 0.9568536899273218, "learning_rate": 6.993875993608965e-06, "loss": 1.6653, "step": 5717 }, { "epoch": 0.4376243685902342, "grad_norm": 0.5952576849759745, "learning_rate": 6.992915549062022e-06, "loss": 1.4685, "step": 5718 }, { "epoch": 0.4377009031073014, "grad_norm": 0.581490154095319, "learning_rate": 6.991955043507383e-06, "loss": 1.3382, "step": 5719 }, { "epoch": 0.4377774376243686, "grad_norm": 0.5927726480419196, "learning_rate": 6.990994477004058e-06, "loss": 1.2744, "step": 5720 }, { "epoch": 0.4378539721414358, "grad_norm": 0.5186914498319001, "learning_rate": 6.990033849611071e-06, "loss": 1.0235, "step": 5721 }, { "epoch": 0.43793050665850297, "grad_norm": 0.6047471401304879, "learning_rate": 6.989073161387444e-06, "loss": 1.414, "step": 5722 }, { "epoch": 0.43800704117557016, "grad_norm": 0.5834443708613687, "learning_rate": 6.988112412392205e-06, "loss": 1.4007, "step": 5723 }, { "epoch": 0.4380835756926374, "grad_norm": 0.548230840553026, "learning_rate": 6.987151602684387e-06, "loss": 1.3021, "step": 5724 }, { "epoch": 0.4381601102097046, "grad_norm": 0.5985093337343527, "learning_rate": 6.986190732323022e-06, "loss": 1.3733, "step": 5725 }, { "epoch": 0.43823664472677176, "grad_norm": 0.6602282605580326, "learning_rate": 6.985229801367151e-06, "loss": 1.3629, "step": 5726 }, { "epoch": 0.43831317924383895, "grad_norm": 0.618182193557307, "learning_rate": 6.984268809875816e-06, "loss": 1.6143, "step": 5727 }, { "epoch": 0.4383897137609062, "grad_norm": 0.49919127927110496, "learning_rate": 6.983307757908063e-06, "loss": 0.9036, "step": 5728 }, { "epoch": 0.43846624827797337, "grad_norm": 0.5536378344740478, "learning_rate": 6.98234664552294e-06, "loss": 1.2448, "step": 5729 }, { "epoch": 0.43854278279504055, "grad_norm": 0.607940413260914, "learning_rate": 6.981385472779505e-06, "loss": 1.3075, "step": 5730 }, { "epoch": 0.43861931731210774, "grad_norm": 0.7365581633609309, "learning_rate": 6.980424239736811e-06, "loss": 1.412, "step": 5731 }, { "epoch": 0.438695851829175, "grad_norm": 0.8688394649373999, "learning_rate": 6.979462946453921e-06, "loss": 1.3373, "step": 5732 }, { "epoch": 0.43877238634624216, "grad_norm": 0.6548891996037911, "learning_rate": 6.9785015929899e-06, "loss": 1.209, "step": 5733 }, { "epoch": 0.43884892086330934, "grad_norm": 0.6130307287209763, "learning_rate": 6.977540179403815e-06, "loss": 1.3575, "step": 5734 }, { "epoch": 0.43892545538037653, "grad_norm": 0.5674004540996601, "learning_rate": 6.976578705754741e-06, "loss": 1.3738, "step": 5735 }, { "epoch": 0.43900198989744377, "grad_norm": 0.7393320624982559, "learning_rate": 6.97561717210175e-06, "loss": 1.7473, "step": 5736 }, { "epoch": 0.43907852441451095, "grad_norm": 0.6720702901693616, "learning_rate": 6.974655578503925e-06, "loss": 1.5548, "step": 5737 }, { "epoch": 0.43915505893157813, "grad_norm": 0.6270849791931539, "learning_rate": 6.973693925020347e-06, "loss": 1.6071, "step": 5738 }, { "epoch": 0.4392315934486453, "grad_norm": 0.6132294947857018, "learning_rate": 6.972732211710103e-06, "loss": 1.5466, "step": 5739 }, { "epoch": 0.43930812796571256, "grad_norm": 0.5940117789661153, "learning_rate": 6.9717704386322834e-06, "loss": 1.5086, "step": 5740 }, { "epoch": 0.43938466248277974, "grad_norm": 0.7314054733499256, "learning_rate": 6.970808605845984e-06, "loss": 1.4376, "step": 5741 }, { "epoch": 0.4394611969998469, "grad_norm": 0.5680829638846301, "learning_rate": 6.969846713410302e-06, "loss": 1.3291, "step": 5742 }, { "epoch": 0.4395377315169141, "grad_norm": 0.5619389301338196, "learning_rate": 6.968884761384338e-06, "loss": 1.4371, "step": 5743 }, { "epoch": 0.43961426603398135, "grad_norm": 0.6180911858060482, "learning_rate": 6.967922749827198e-06, "loss": 1.5733, "step": 5744 }, { "epoch": 0.43969080055104853, "grad_norm": 0.9464989139049741, "learning_rate": 6.966960678797991e-06, "loss": 1.6086, "step": 5745 }, { "epoch": 0.4397673350681157, "grad_norm": 0.7183386825575743, "learning_rate": 6.965998548355829e-06, "loss": 1.5481, "step": 5746 }, { "epoch": 0.4398438695851829, "grad_norm": 0.6532889538973458, "learning_rate": 6.965036358559828e-06, "loss": 1.6722, "step": 5747 }, { "epoch": 0.43992040410225014, "grad_norm": 0.5782388093967267, "learning_rate": 6.964074109469109e-06, "loss": 1.3855, "step": 5748 }, { "epoch": 0.4399969386193173, "grad_norm": 0.48550054527631603, "learning_rate": 6.963111801142795e-06, "loss": 1.1085, "step": 5749 }, { "epoch": 0.4400734731363845, "grad_norm": 0.6608525169468991, "learning_rate": 6.962149433640012e-06, "loss": 1.5728, "step": 5750 }, { "epoch": 0.4401500076534517, "grad_norm": 0.8093575564787749, "learning_rate": 6.96118700701989e-06, "loss": 1.3872, "step": 5751 }, { "epoch": 0.44022654217051893, "grad_norm": 0.5875479089171425, "learning_rate": 6.960224521341568e-06, "loss": 1.3289, "step": 5752 }, { "epoch": 0.4403030766875861, "grad_norm": 0.5184288814858885, "learning_rate": 6.959261976664178e-06, "loss": 1.3155, "step": 5753 }, { "epoch": 0.4403796112046533, "grad_norm": 0.8068776440223686, "learning_rate": 6.9582993730468664e-06, "loss": 1.502, "step": 5754 }, { "epoch": 0.4404561457217205, "grad_norm": 0.58629214113054, "learning_rate": 6.9573367105487755e-06, "loss": 1.4465, "step": 5755 }, { "epoch": 0.44053268023878767, "grad_norm": 0.718242781810663, "learning_rate": 6.956373989229056e-06, "loss": 1.3251, "step": 5756 }, { "epoch": 0.4406092147558549, "grad_norm": 0.5683758297678182, "learning_rate": 6.955411209146859e-06, "loss": 1.3987, "step": 5757 }, { "epoch": 0.4406857492729221, "grad_norm": 0.6512875705905887, "learning_rate": 6.9544483703613415e-06, "loss": 1.6128, "step": 5758 }, { "epoch": 0.4407622837899893, "grad_norm": 0.49138597077296825, "learning_rate": 6.953485472931664e-06, "loss": 1.0095, "step": 5759 }, { "epoch": 0.44083881830705646, "grad_norm": 0.6061923780340439, "learning_rate": 6.9525225169169876e-06, "loss": 1.3933, "step": 5760 }, { "epoch": 0.4409153528241237, "grad_norm": 0.5526879006492375, "learning_rate": 6.95155950237648e-06, "loss": 1.2085, "step": 5761 }, { "epoch": 0.4409918873411909, "grad_norm": 0.6070499478993753, "learning_rate": 6.950596429369312e-06, "loss": 1.6535, "step": 5762 }, { "epoch": 0.44106842185825806, "grad_norm": 0.7897767706388634, "learning_rate": 6.949633297954657e-06, "loss": 1.5375, "step": 5763 }, { "epoch": 0.44114495637532525, "grad_norm": 0.736265946563942, "learning_rate": 6.948670108191694e-06, "loss": 1.5968, "step": 5764 }, { "epoch": 0.4412214908923925, "grad_norm": 0.7142977402033258, "learning_rate": 6.947706860139603e-06, "loss": 1.3361, "step": 5765 }, { "epoch": 0.44129802540945967, "grad_norm": 0.5623440767499568, "learning_rate": 6.9467435538575716e-06, "loss": 1.3511, "step": 5766 }, { "epoch": 0.44137455992652685, "grad_norm": 0.7265346008920016, "learning_rate": 6.945780189404782e-06, "loss": 1.5565, "step": 5767 }, { "epoch": 0.44145109444359404, "grad_norm": 0.6484289145258122, "learning_rate": 6.944816766840435e-06, "loss": 1.2493, "step": 5768 }, { "epoch": 0.4415276289606613, "grad_norm": 0.5670678046722484, "learning_rate": 6.943853286223718e-06, "loss": 1.3875, "step": 5769 }, { "epoch": 0.44160416347772846, "grad_norm": 0.5314884776031455, "learning_rate": 6.942889747613836e-06, "loss": 1.1793, "step": 5770 }, { "epoch": 0.44168069799479565, "grad_norm": 0.49343536557009704, "learning_rate": 6.941926151069988e-06, "loss": 1.0314, "step": 5771 }, { "epoch": 0.44175723251186283, "grad_norm": 0.5036797249296557, "learning_rate": 6.9409624966513815e-06, "loss": 1.1925, "step": 5772 }, { "epoch": 0.44183376702893007, "grad_norm": 0.5892644544082717, "learning_rate": 6.939998784417226e-06, "loss": 1.5782, "step": 5773 }, { "epoch": 0.44191030154599725, "grad_norm": 0.7155616964325731, "learning_rate": 6.9390350144267345e-06, "loss": 1.6814, "step": 5774 }, { "epoch": 0.44198683606306444, "grad_norm": 0.6076319909478761, "learning_rate": 6.9380711867391235e-06, "loss": 1.5492, "step": 5775 }, { "epoch": 0.4420633705801316, "grad_norm": 0.716200083439225, "learning_rate": 6.937107301413617e-06, "loss": 1.6071, "step": 5776 }, { "epoch": 0.44213990509719886, "grad_norm": 0.602360347714149, "learning_rate": 6.9361433585094364e-06, "loss": 1.3201, "step": 5777 }, { "epoch": 0.44221643961426604, "grad_norm": 0.5684013691876264, "learning_rate": 6.935179358085807e-06, "loss": 1.134, "step": 5778 }, { "epoch": 0.4422929741313332, "grad_norm": 0.542329665621679, "learning_rate": 6.9342153002019645e-06, "loss": 1.4007, "step": 5779 }, { "epoch": 0.4423695086484004, "grad_norm": 0.5123271708294757, "learning_rate": 6.933251184917141e-06, "loss": 1.0989, "step": 5780 }, { "epoch": 0.44244604316546765, "grad_norm": 0.7809941000413477, "learning_rate": 6.932287012290574e-06, "loss": 1.2228, "step": 5781 }, { "epoch": 0.44252257768253483, "grad_norm": 0.5376059959852361, "learning_rate": 6.931322782381507e-06, "loss": 1.2302, "step": 5782 }, { "epoch": 0.442599112199602, "grad_norm": 0.5034673146195427, "learning_rate": 6.930358495249183e-06, "loss": 1.0857, "step": 5783 }, { "epoch": 0.4426756467166692, "grad_norm": 1.8736234694085785, "learning_rate": 6.929394150952851e-06, "loss": 1.5771, "step": 5784 }, { "epoch": 0.44275218123373644, "grad_norm": 0.7850677891155334, "learning_rate": 6.928429749551766e-06, "loss": 1.5158, "step": 5785 }, { "epoch": 0.4428287157508036, "grad_norm": 0.545751488242769, "learning_rate": 6.92746529110518e-06, "loss": 1.2236, "step": 5786 }, { "epoch": 0.4429052502678708, "grad_norm": 0.6026292224781999, "learning_rate": 6.926500775672356e-06, "loss": 1.4174, "step": 5787 }, { "epoch": 0.442981784784938, "grad_norm": 1.3948118676798646, "learning_rate": 6.925536203312553e-06, "loss": 1.3668, "step": 5788 }, { "epoch": 0.4430583193020052, "grad_norm": 0.5172349757936284, "learning_rate": 6.92457157408504e-06, "loss": 1.0622, "step": 5789 }, { "epoch": 0.4431348538190724, "grad_norm": 0.5498130973840717, "learning_rate": 6.9236068880490835e-06, "loss": 1.3526, "step": 5790 }, { "epoch": 0.4432113883361396, "grad_norm": 0.6657231972720574, "learning_rate": 6.92264214526396e-06, "loss": 1.2101, "step": 5791 }, { "epoch": 0.4432879228532068, "grad_norm": 0.6151575549488103, "learning_rate": 6.9216773457889456e-06, "loss": 1.3847, "step": 5792 }, { "epoch": 0.44336445737027397, "grad_norm": 0.5976355397918072, "learning_rate": 6.920712489683319e-06, "loss": 1.3894, "step": 5793 }, { "epoch": 0.4434409918873412, "grad_norm": 0.55645630775135, "learning_rate": 6.919747577006364e-06, "loss": 1.2934, "step": 5794 }, { "epoch": 0.4435175264044084, "grad_norm": 0.6105548865289125, "learning_rate": 6.918782607817369e-06, "loss": 1.2937, "step": 5795 }, { "epoch": 0.4435940609214756, "grad_norm": 0.5634197999963606, "learning_rate": 6.917817582175625e-06, "loss": 1.3312, "step": 5796 }, { "epoch": 0.44367059543854276, "grad_norm": 0.9227306653666856, "learning_rate": 6.916852500140423e-06, "loss": 1.6688, "step": 5797 }, { "epoch": 0.44374712995561, "grad_norm": 0.6204590392741772, "learning_rate": 6.915887361771065e-06, "loss": 1.3647, "step": 5798 }, { "epoch": 0.4438236644726772, "grad_norm": 1.2382092167653285, "learning_rate": 6.9149221671268475e-06, "loss": 1.578, "step": 5799 }, { "epoch": 0.44390019898974437, "grad_norm": 0.5527740520217146, "learning_rate": 6.913956916267077e-06, "loss": 1.3327, "step": 5800 }, { "epoch": 0.44397673350681155, "grad_norm": 0.606023516341631, "learning_rate": 6.912991609251063e-06, "loss": 1.3713, "step": 5801 }, { "epoch": 0.4440532680238788, "grad_norm": 0.5272013612568982, "learning_rate": 6.912026246138113e-06, "loss": 1.2629, "step": 5802 }, { "epoch": 0.444129802540946, "grad_norm": 0.598173550348848, "learning_rate": 6.911060826987545e-06, "loss": 1.322, "step": 5803 }, { "epoch": 0.44420633705801316, "grad_norm": 0.5660220593498235, "learning_rate": 6.910095351858677e-06, "loss": 1.5252, "step": 5804 }, { "epoch": 0.44428287157508034, "grad_norm": 0.5664140443208023, "learning_rate": 6.909129820810829e-06, "loss": 1.4635, "step": 5805 }, { "epoch": 0.4443594060921476, "grad_norm": 0.5997190735810622, "learning_rate": 6.908164233903327e-06, "loss": 1.2745, "step": 5806 }, { "epoch": 0.44443594060921476, "grad_norm": 0.631115433365912, "learning_rate": 6.9071985911955005e-06, "loss": 1.6101, "step": 5807 }, { "epoch": 0.44451247512628195, "grad_norm": 0.5639907461510819, "learning_rate": 6.906232892746679e-06, "loss": 1.3399, "step": 5808 }, { "epoch": 0.44458900964334913, "grad_norm": 0.5091716441733118, "learning_rate": 6.905267138616201e-06, "loss": 1.1553, "step": 5809 }, { "epoch": 0.44466554416041637, "grad_norm": 0.6484666675447996, "learning_rate": 6.904301328863403e-06, "loss": 1.51, "step": 5810 }, { "epoch": 0.44474207867748355, "grad_norm": 2.6490948024308776, "learning_rate": 6.903335463547628e-06, "loss": 1.6855, "step": 5811 }, { "epoch": 0.44481861319455074, "grad_norm": 0.5375817932843887, "learning_rate": 6.902369542728221e-06, "loss": 1.1921, "step": 5812 }, { "epoch": 0.4448951477116179, "grad_norm": 0.6812962245983467, "learning_rate": 6.901403566464533e-06, "loss": 1.5158, "step": 5813 }, { "epoch": 0.44497168222868516, "grad_norm": 0.5686095371611581, "learning_rate": 6.900437534815915e-06, "loss": 1.1221, "step": 5814 }, { "epoch": 0.44504821674575235, "grad_norm": 0.5458771470275303, "learning_rate": 6.899471447841724e-06, "loss": 1.3658, "step": 5815 }, { "epoch": 0.44512475126281953, "grad_norm": 0.5912208240376662, "learning_rate": 6.898505305601315e-06, "loss": 1.3835, "step": 5816 }, { "epoch": 0.4452012857798867, "grad_norm": 0.5939653408903658, "learning_rate": 6.897539108154056e-06, "loss": 1.4564, "step": 5817 }, { "epoch": 0.44527782029695395, "grad_norm": 0.5448248044052969, "learning_rate": 6.8965728555593124e-06, "loss": 1.4319, "step": 5818 }, { "epoch": 0.44535435481402114, "grad_norm": 0.5501198329436718, "learning_rate": 6.89560654787645e-06, "loss": 1.2799, "step": 5819 }, { "epoch": 0.4454308893310883, "grad_norm": 2.187107946978504, "learning_rate": 6.894640185164846e-06, "loss": 1.5104, "step": 5820 }, { "epoch": 0.4455074238481555, "grad_norm": 0.5796238826546113, "learning_rate": 6.8936737674838735e-06, "loss": 1.543, "step": 5821 }, { "epoch": 0.44558395836522274, "grad_norm": 0.6461873462577812, "learning_rate": 6.892707294892913e-06, "loss": 1.5845, "step": 5822 }, { "epoch": 0.4456604928822899, "grad_norm": 0.5040660121431564, "learning_rate": 6.891740767451348e-06, "loss": 1.0734, "step": 5823 }, { "epoch": 0.4457370273993571, "grad_norm": 0.596628101112772, "learning_rate": 6.890774185218564e-06, "loss": 1.3385, "step": 5824 }, { "epoch": 0.4458135619164243, "grad_norm": 0.5125244488220809, "learning_rate": 6.889807548253952e-06, "loss": 1.2759, "step": 5825 }, { "epoch": 0.4458900964334915, "grad_norm": 0.5711099514057376, "learning_rate": 6.888840856616904e-06, "loss": 1.318, "step": 5826 }, { "epoch": 0.4459666309505587, "grad_norm": 0.6278014845352275, "learning_rate": 6.887874110366815e-06, "loss": 1.6097, "step": 5827 }, { "epoch": 0.4460431654676259, "grad_norm": 0.587632793780652, "learning_rate": 6.886907309563089e-06, "loss": 1.3827, "step": 5828 }, { "epoch": 0.4461196999846931, "grad_norm": 0.7438342949423609, "learning_rate": 6.8859404542651256e-06, "loss": 1.5353, "step": 5829 }, { "epoch": 0.44619623450176027, "grad_norm": 0.8662479695715645, "learning_rate": 6.8849735445323305e-06, "loss": 1.5684, "step": 5830 }, { "epoch": 0.4462727690188275, "grad_norm": 0.6529377237127313, "learning_rate": 6.884006580424116e-06, "loss": 1.6382, "step": 5831 }, { "epoch": 0.4463493035358947, "grad_norm": 0.6073244756403624, "learning_rate": 6.883039561999896e-06, "loss": 1.5296, "step": 5832 }, { "epoch": 0.4464258380529619, "grad_norm": 0.5851307024649737, "learning_rate": 6.882072489319085e-06, "loss": 1.4607, "step": 5833 }, { "epoch": 0.44650237257002906, "grad_norm": 0.5398401541139738, "learning_rate": 6.881105362441103e-06, "loss": 1.3869, "step": 5834 }, { "epoch": 0.4465789070870963, "grad_norm": 0.49663966808602644, "learning_rate": 6.880138181425373e-06, "loss": 1.1602, "step": 5835 }, { "epoch": 0.4466554416041635, "grad_norm": 0.633028477369639, "learning_rate": 6.8791709463313224e-06, "loss": 1.4966, "step": 5836 }, { "epoch": 0.44673197612123067, "grad_norm": 0.7236572665812122, "learning_rate": 6.878203657218382e-06, "loss": 1.6421, "step": 5837 }, { "epoch": 0.44680851063829785, "grad_norm": 0.5376061548921635, "learning_rate": 6.877236314145983e-06, "loss": 1.2633, "step": 5838 }, { "epoch": 0.4468850451553651, "grad_norm": 2.15175635518004, "learning_rate": 6.876268917173561e-06, "loss": 1.6296, "step": 5839 }, { "epoch": 0.4469615796724323, "grad_norm": 0.8057481318913291, "learning_rate": 6.8753014663605574e-06, "loss": 1.8426, "step": 5840 }, { "epoch": 0.44703811418949946, "grad_norm": 0.5414570950265186, "learning_rate": 6.874333961766416e-06, "loss": 1.177, "step": 5841 }, { "epoch": 0.44711464870656664, "grad_norm": 0.6175815141135189, "learning_rate": 6.873366403450583e-06, "loss": 1.4261, "step": 5842 }, { "epoch": 0.4471911832236339, "grad_norm": 0.8456738767575814, "learning_rate": 6.8723987914725055e-06, "loss": 1.6145, "step": 5843 }, { "epoch": 0.44726771774070107, "grad_norm": 0.5669591117875797, "learning_rate": 6.871431125891639e-06, "loss": 1.4145, "step": 5844 }, { "epoch": 0.44734425225776825, "grad_norm": 0.624178018752147, "learning_rate": 6.870463406767439e-06, "loss": 1.3092, "step": 5845 }, { "epoch": 0.44742078677483543, "grad_norm": 0.5510369374487276, "learning_rate": 6.869495634159365e-06, "loss": 1.2323, "step": 5846 }, { "epoch": 0.44749732129190267, "grad_norm": 0.5503983784135508, "learning_rate": 6.86852780812688e-06, "loss": 1.2434, "step": 5847 }, { "epoch": 0.44757385580896986, "grad_norm": 0.6022923071903624, "learning_rate": 6.86755992872945e-06, "loss": 1.3968, "step": 5848 }, { "epoch": 0.44765039032603704, "grad_norm": 0.6209621117904377, "learning_rate": 6.866591996026545e-06, "loss": 1.4994, "step": 5849 }, { "epoch": 0.4477269248431042, "grad_norm": 0.6917674462160492, "learning_rate": 6.865624010077636e-06, "loss": 1.5147, "step": 5850 }, { "epoch": 0.44780345936017146, "grad_norm": 0.567377464319868, "learning_rate": 6.864655970942202e-06, "loss": 1.3019, "step": 5851 }, { "epoch": 0.44787999387723865, "grad_norm": 0.5257767863367265, "learning_rate": 6.863687878679719e-06, "loss": 1.1799, "step": 5852 }, { "epoch": 0.44795652839430583, "grad_norm": 0.9547269042049282, "learning_rate": 6.862719733349671e-06, "loss": 1.5698, "step": 5853 }, { "epoch": 0.448033062911373, "grad_norm": 0.5773398109201717, "learning_rate": 6.861751535011543e-06, "loss": 1.2933, "step": 5854 }, { "epoch": 0.44810959742844025, "grad_norm": 0.5374889477618446, "learning_rate": 6.860783283724826e-06, "loss": 1.2275, "step": 5855 }, { "epoch": 0.44818613194550744, "grad_norm": 0.5368815306481333, "learning_rate": 6.859814979549009e-06, "loss": 1.2982, "step": 5856 }, { "epoch": 0.4482626664625746, "grad_norm": 0.5728353198152704, "learning_rate": 6.85884662254359e-06, "loss": 1.4576, "step": 5857 }, { "epoch": 0.4483392009796418, "grad_norm": 0.5591595272845455, "learning_rate": 6.857878212768068e-06, "loss": 1.3244, "step": 5858 }, { "epoch": 0.448415735496709, "grad_norm": 0.6552199767787027, "learning_rate": 6.856909750281941e-06, "loss": 1.3246, "step": 5859 }, { "epoch": 0.44849227001377623, "grad_norm": 0.6703345520234059, "learning_rate": 6.855941235144721e-06, "loss": 1.4615, "step": 5860 }, { "epoch": 0.4485688045308434, "grad_norm": 0.5505784481290146, "learning_rate": 6.8549726674159115e-06, "loss": 1.1514, "step": 5861 }, { "epoch": 0.4486453390479106, "grad_norm": 0.5228104887266417, "learning_rate": 6.854004047155025e-06, "loss": 1.1131, "step": 5862 }, { "epoch": 0.4487218735649778, "grad_norm": 0.9810714451528408, "learning_rate": 6.853035374421577e-06, "loss": 1.7061, "step": 5863 }, { "epoch": 0.448798408082045, "grad_norm": 0.7788463172420049, "learning_rate": 6.8520666492750854e-06, "loss": 1.7274, "step": 5864 }, { "epoch": 0.4488749425991122, "grad_norm": 0.5781410926220198, "learning_rate": 6.851097871775072e-06, "loss": 1.3436, "step": 5865 }, { "epoch": 0.4489514771161794, "grad_norm": 0.7146015876444413, "learning_rate": 6.85012904198106e-06, "loss": 1.5734, "step": 5866 }, { "epoch": 0.44902801163324657, "grad_norm": 0.780936109975855, "learning_rate": 6.8491601599525805e-06, "loss": 1.1063, "step": 5867 }, { "epoch": 0.4491045461503138, "grad_norm": 0.5223806684622104, "learning_rate": 6.848191225749159e-06, "loss": 1.2739, "step": 5868 }, { "epoch": 0.449181080667381, "grad_norm": 0.6049314003882038, "learning_rate": 6.847222239430335e-06, "loss": 1.4307, "step": 5869 }, { "epoch": 0.4492576151844482, "grad_norm": 0.5828912323704963, "learning_rate": 6.846253201055645e-06, "loss": 1.4114, "step": 5870 }, { "epoch": 0.44933414970151536, "grad_norm": 0.6199846169306727, "learning_rate": 6.845284110684628e-06, "loss": 1.5312, "step": 5871 }, { "epoch": 0.4494106842185826, "grad_norm": 0.6476802323517259, "learning_rate": 6.844314968376827e-06, "loss": 1.355, "step": 5872 }, { "epoch": 0.4494872187356498, "grad_norm": 1.7099633094943396, "learning_rate": 6.8433457741917916e-06, "loss": 1.5368, "step": 5873 }, { "epoch": 0.44956375325271697, "grad_norm": 0.627395828731304, "learning_rate": 6.842376528189068e-06, "loss": 1.5133, "step": 5874 }, { "epoch": 0.44964028776978415, "grad_norm": 0.5800720760632896, "learning_rate": 6.841407230428216e-06, "loss": 1.4489, "step": 5875 }, { "epoch": 0.4497168222868514, "grad_norm": 0.8136011605021249, "learning_rate": 6.840437880968786e-06, "loss": 1.8881, "step": 5876 }, { "epoch": 0.4497933568039186, "grad_norm": 0.5860858593633393, "learning_rate": 6.83946847987034e-06, "loss": 1.3533, "step": 5877 }, { "epoch": 0.44986989132098576, "grad_norm": 0.8903253182140859, "learning_rate": 6.838499027192443e-06, "loss": 1.6353, "step": 5878 }, { "epoch": 0.44994642583805294, "grad_norm": 0.6383607911308531, "learning_rate": 6.837529522994658e-06, "loss": 1.6937, "step": 5879 }, { "epoch": 0.4500229603551202, "grad_norm": 0.6265837750473017, "learning_rate": 6.836559967336556e-06, "loss": 1.4976, "step": 5880 }, { "epoch": 0.45009949487218737, "grad_norm": 0.5822351388620898, "learning_rate": 6.835590360277707e-06, "loss": 1.4078, "step": 5881 }, { "epoch": 0.45017602938925455, "grad_norm": 0.555483355048409, "learning_rate": 6.8346207018776915e-06, "loss": 1.323, "step": 5882 }, { "epoch": 0.45025256390632173, "grad_norm": 0.5148992340486993, "learning_rate": 6.833650992196083e-06, "loss": 1.2587, "step": 5883 }, { "epoch": 0.450329098423389, "grad_norm": 0.6388886261541977, "learning_rate": 6.832681231292466e-06, "loss": 1.4918, "step": 5884 }, { "epoch": 0.45040563294045616, "grad_norm": 0.5223698657328878, "learning_rate": 6.831711419226423e-06, "loss": 1.0794, "step": 5885 }, { "epoch": 0.45048216745752334, "grad_norm": 0.6258075237876881, "learning_rate": 6.830741556057547e-06, "loss": 1.5253, "step": 5886 }, { "epoch": 0.4505587019745905, "grad_norm": 0.6019805717633623, "learning_rate": 6.829771641845424e-06, "loss": 1.3851, "step": 5887 }, { "epoch": 0.45063523649165776, "grad_norm": 0.5939449492424294, "learning_rate": 6.828801676649651e-06, "loss": 1.402, "step": 5888 }, { "epoch": 0.45071177100872495, "grad_norm": 0.5463995889341808, "learning_rate": 6.827831660529825e-06, "loss": 1.2501, "step": 5889 }, { "epoch": 0.45078830552579213, "grad_norm": 0.5922436212071122, "learning_rate": 6.826861593545548e-06, "loss": 1.3182, "step": 5890 }, { "epoch": 0.4508648400428593, "grad_norm": 0.8756604233326281, "learning_rate": 6.825891475756423e-06, "loss": 1.4816, "step": 5891 }, { "epoch": 0.4509413745599265, "grad_norm": 0.7079644185609172, "learning_rate": 6.8249213072220564e-06, "loss": 1.6648, "step": 5892 }, { "epoch": 0.45101790907699374, "grad_norm": 0.5931734640563289, "learning_rate": 6.82395108800206e-06, "loss": 1.3576, "step": 5893 }, { "epoch": 0.4510944435940609, "grad_norm": 0.5818116851010962, "learning_rate": 6.822980818156044e-06, "loss": 1.3041, "step": 5894 }, { "epoch": 0.4511709781111281, "grad_norm": 0.610468984484329, "learning_rate": 6.8220104977436275e-06, "loss": 1.5308, "step": 5895 }, { "epoch": 0.4512475126281953, "grad_norm": 0.7091268465165327, "learning_rate": 6.821040126824428e-06, "loss": 1.8152, "step": 5896 }, { "epoch": 0.45132404714526253, "grad_norm": 0.6107978016015165, "learning_rate": 6.820069705458068e-06, "loss": 1.3777, "step": 5897 }, { "epoch": 0.4514005816623297, "grad_norm": 0.5491212804418426, "learning_rate": 6.819099233704174e-06, "loss": 1.3825, "step": 5898 }, { "epoch": 0.4514771161793969, "grad_norm": 0.5646041918970611, "learning_rate": 6.818128711622375e-06, "loss": 1.4309, "step": 5899 }, { "epoch": 0.4515536506964641, "grad_norm": 0.6013971153081076, "learning_rate": 6.817158139272302e-06, "loss": 1.4446, "step": 5900 }, { "epoch": 0.4516301852135313, "grad_norm": 0.8495280333840335, "learning_rate": 6.81618751671359e-06, "loss": 1.2706, "step": 5901 }, { "epoch": 0.4517067197305985, "grad_norm": 0.49486626144124424, "learning_rate": 6.815216844005876e-06, "loss": 0.8329, "step": 5902 }, { "epoch": 0.4517832542476657, "grad_norm": 0.6011624943755794, "learning_rate": 6.814246121208804e-06, "loss": 1.3174, "step": 5903 }, { "epoch": 0.4518597887647329, "grad_norm": 0.713751757052544, "learning_rate": 6.813275348382016e-06, "loss": 1.8467, "step": 5904 }, { "epoch": 0.4519363232818001, "grad_norm": 0.5944703829732537, "learning_rate": 6.812304525585159e-06, "loss": 1.5238, "step": 5905 }, { "epoch": 0.4520128577988673, "grad_norm": 0.5923296873658089, "learning_rate": 6.811333652877883e-06, "loss": 1.5423, "step": 5906 }, { "epoch": 0.4520893923159345, "grad_norm": 0.7108700406086412, "learning_rate": 6.8103627303198415e-06, "loss": 1.1392, "step": 5907 }, { "epoch": 0.45216592683300166, "grad_norm": 0.5313465009665386, "learning_rate": 6.809391757970692e-06, "loss": 1.2384, "step": 5908 }, { "epoch": 0.4522424613500689, "grad_norm": 0.7502845676617038, "learning_rate": 6.808420735890095e-06, "loss": 1.6699, "step": 5909 }, { "epoch": 0.4523189958671361, "grad_norm": 0.593934875195089, "learning_rate": 6.807449664137709e-06, "loss": 1.3662, "step": 5910 }, { "epoch": 0.45239553038420327, "grad_norm": 0.5281636904733347, "learning_rate": 6.8064785427732024e-06, "loss": 1.2396, "step": 5911 }, { "epoch": 0.45247206490127045, "grad_norm": 0.5892159919300678, "learning_rate": 6.805507371856243e-06, "loss": 1.3515, "step": 5912 }, { "epoch": 0.4525485994183377, "grad_norm": 0.49327166359512586, "learning_rate": 6.804536151446503e-06, "loss": 0.8039, "step": 5913 }, { "epoch": 0.4526251339354049, "grad_norm": 0.6097186314808337, "learning_rate": 6.803564881603657e-06, "loss": 1.5736, "step": 5914 }, { "epoch": 0.45270166845247206, "grad_norm": 0.6339500678107483, "learning_rate": 6.802593562387383e-06, "loss": 1.5046, "step": 5915 }, { "epoch": 0.45277820296953925, "grad_norm": 0.5930157726002846, "learning_rate": 6.80162219385736e-06, "loss": 1.2422, "step": 5916 }, { "epoch": 0.4528547374866065, "grad_norm": 0.6781141588519907, "learning_rate": 6.800650776073274e-06, "loss": 1.6354, "step": 5917 }, { "epoch": 0.45293127200367367, "grad_norm": 0.5213795974675686, "learning_rate": 6.799679309094809e-06, "loss": 0.9785, "step": 5918 }, { "epoch": 0.45300780652074085, "grad_norm": 0.5368921759261948, "learning_rate": 6.798707792981658e-06, "loss": 1.2862, "step": 5919 }, { "epoch": 0.45308434103780804, "grad_norm": 0.6185359614628653, "learning_rate": 6.7977362277935136e-06, "loss": 1.5826, "step": 5920 }, { "epoch": 0.4531608755548753, "grad_norm": 0.5987840448080833, "learning_rate": 6.796764613590069e-06, "loss": 1.1944, "step": 5921 }, { "epoch": 0.45323741007194246, "grad_norm": 0.7705615497381832, "learning_rate": 6.795792950431025e-06, "loss": 1.4633, "step": 5922 }, { "epoch": 0.45331394458900964, "grad_norm": 0.5020344961902692, "learning_rate": 6.794821238376082e-06, "loss": 0.9843, "step": 5923 }, { "epoch": 0.4533904791060768, "grad_norm": 0.5685350273251929, "learning_rate": 6.793849477484948e-06, "loss": 1.2111, "step": 5924 }, { "epoch": 0.453467013623144, "grad_norm": 0.6034824484403084, "learning_rate": 6.792877667817328e-06, "loss": 1.5205, "step": 5925 }, { "epoch": 0.45354354814021125, "grad_norm": 0.7086234102259097, "learning_rate": 6.791905809432935e-06, "loss": 1.6264, "step": 5926 }, { "epoch": 0.45362008265727843, "grad_norm": 1.3247761120196866, "learning_rate": 6.79093390239148e-06, "loss": 1.5668, "step": 5927 }, { "epoch": 0.4536966171743456, "grad_norm": 0.5013096800034736, "learning_rate": 6.789961946752684e-06, "loss": 1.0672, "step": 5928 }, { "epoch": 0.4537731516914128, "grad_norm": 0.7976212796928053, "learning_rate": 6.788989942576262e-06, "loss": 1.3094, "step": 5929 }, { "epoch": 0.45384968620848004, "grad_norm": 0.6059491110962711, "learning_rate": 6.78801788992194e-06, "loss": 1.4939, "step": 5930 }, { "epoch": 0.4539262207255472, "grad_norm": 0.5754704372357463, "learning_rate": 6.787045788849444e-06, "loss": 1.2221, "step": 5931 }, { "epoch": 0.4540027552426144, "grad_norm": 0.6504488038945335, "learning_rate": 6.7860736394185e-06, "loss": 1.657, "step": 5932 }, { "epoch": 0.4540792897596816, "grad_norm": 1.114121841530628, "learning_rate": 6.785101441688844e-06, "loss": 1.6434, "step": 5933 }, { "epoch": 0.45415582427674883, "grad_norm": 0.5116696146459916, "learning_rate": 6.784129195720206e-06, "loss": 0.9597, "step": 5934 }, { "epoch": 0.454232358793816, "grad_norm": 0.6155580720696086, "learning_rate": 6.783156901572327e-06, "loss": 1.3112, "step": 5935 }, { "epoch": 0.4543088933108832, "grad_norm": 0.6505500519811831, "learning_rate": 6.782184559304948e-06, "loss": 1.6111, "step": 5936 }, { "epoch": 0.4543854278279504, "grad_norm": 0.759335387117584, "learning_rate": 6.78121216897781e-06, "loss": 1.5676, "step": 5937 }, { "epoch": 0.4544619623450176, "grad_norm": 0.7292608644118274, "learning_rate": 6.780239730650662e-06, "loss": 1.424, "step": 5938 }, { "epoch": 0.4545384968620848, "grad_norm": 0.5054470407390074, "learning_rate": 6.779267244383253e-06, "loss": 1.1309, "step": 5939 }, { "epoch": 0.454615031379152, "grad_norm": 0.5871922610290122, "learning_rate": 6.778294710235333e-06, "loss": 1.4416, "step": 5940 }, { "epoch": 0.4546915658962192, "grad_norm": 0.6197885817854362, "learning_rate": 6.7773221282666605e-06, "loss": 1.5687, "step": 5941 }, { "epoch": 0.4547681004132864, "grad_norm": 0.6053609443026969, "learning_rate": 6.776349498536991e-06, "loss": 1.2335, "step": 5942 }, { "epoch": 0.4548446349303536, "grad_norm": 0.5543836948789733, "learning_rate": 6.77537682110609e-06, "loss": 1.0566, "step": 5943 }, { "epoch": 0.4549211694474208, "grad_norm": 0.8406733722584467, "learning_rate": 6.774404096033717e-06, "loss": 1.5063, "step": 5944 }, { "epoch": 0.45499770396448797, "grad_norm": 0.5940230544710166, "learning_rate": 6.773431323379643e-06, "loss": 1.4662, "step": 5945 }, { "epoch": 0.4550742384815552, "grad_norm": 0.8436006835827513, "learning_rate": 6.772458503203636e-06, "loss": 1.6032, "step": 5946 }, { "epoch": 0.4551507729986224, "grad_norm": 0.5192007949581731, "learning_rate": 6.771485635565469e-06, "loss": 1.1964, "step": 5947 }, { "epoch": 0.45522730751568957, "grad_norm": 1.330637061490651, "learning_rate": 6.770512720524919e-06, "loss": 1.3227, "step": 5948 }, { "epoch": 0.45530384203275676, "grad_norm": 0.6176116595250576, "learning_rate": 6.769539758141763e-06, "loss": 1.5088, "step": 5949 }, { "epoch": 0.455380376549824, "grad_norm": 0.5873571335534816, "learning_rate": 6.768566748475784e-06, "loss": 1.4168, "step": 5950 }, { "epoch": 0.4554569110668912, "grad_norm": 0.6344299454359217, "learning_rate": 6.767593691586767e-06, "loss": 1.4133, "step": 5951 }, { "epoch": 0.45553344558395836, "grad_norm": 0.6698982363227315, "learning_rate": 6.766620587534499e-06, "loss": 1.515, "step": 5952 }, { "epoch": 0.45560998010102555, "grad_norm": 0.580722162277104, "learning_rate": 6.765647436378769e-06, "loss": 1.2995, "step": 5953 }, { "epoch": 0.4556865146180928, "grad_norm": 0.5441531517433432, "learning_rate": 6.7646742381793725e-06, "loss": 1.1448, "step": 5954 }, { "epoch": 0.45576304913515997, "grad_norm": 0.5723920670302672, "learning_rate": 6.763700992996105e-06, "loss": 1.4445, "step": 5955 }, { "epoch": 0.45583958365222715, "grad_norm": 0.8577789115082093, "learning_rate": 6.762727700888765e-06, "loss": 1.2464, "step": 5956 }, { "epoch": 0.45591611816929434, "grad_norm": 0.6280749041534235, "learning_rate": 6.761754361917156e-06, "loss": 1.4727, "step": 5957 }, { "epoch": 0.4559926526863615, "grad_norm": 1.2492126758957787, "learning_rate": 6.760780976141081e-06, "loss": 1.2498, "step": 5958 }, { "epoch": 0.45606918720342876, "grad_norm": 0.5959214467213696, "learning_rate": 6.759807543620348e-06, "loss": 1.4119, "step": 5959 }, { "epoch": 0.45614572172049594, "grad_norm": 0.7252345713283989, "learning_rate": 6.7588340644147695e-06, "loss": 1.6028, "step": 5960 }, { "epoch": 0.45622225623756313, "grad_norm": 0.716975904165217, "learning_rate": 6.757860538584157e-06, "loss": 1.4375, "step": 5961 }, { "epoch": 0.4562987907546303, "grad_norm": 0.5676442276789564, "learning_rate": 6.756886966188328e-06, "loss": 1.3884, "step": 5962 }, { "epoch": 0.45637532527169755, "grad_norm": 0.5636736596529958, "learning_rate": 6.7559133472870995e-06, "loss": 1.3353, "step": 5963 }, { "epoch": 0.45645185978876474, "grad_norm": 0.554381747955119, "learning_rate": 6.754939681940296e-06, "loss": 1.2117, "step": 5964 }, { "epoch": 0.4565283943058319, "grad_norm": 0.8787702844166855, "learning_rate": 6.753965970207742e-06, "loss": 1.7317, "step": 5965 }, { "epoch": 0.4566049288228991, "grad_norm": 1.7203739416888137, "learning_rate": 6.7529922121492655e-06, "loss": 1.7649, "step": 5966 }, { "epoch": 0.45668146333996634, "grad_norm": 0.5812605361027114, "learning_rate": 6.7520184078246944e-06, "loss": 1.2885, "step": 5967 }, { "epoch": 0.4567579978570335, "grad_norm": 0.5519236804338411, "learning_rate": 6.751044557293865e-06, "loss": 1.3009, "step": 5968 }, { "epoch": 0.4568345323741007, "grad_norm": 0.6072198133578876, "learning_rate": 6.7500706606166136e-06, "loss": 1.4441, "step": 5969 }, { "epoch": 0.4569110668911679, "grad_norm": 0.6146995930982645, "learning_rate": 6.749096717852778e-06, "loss": 1.602, "step": 5970 }, { "epoch": 0.45698760140823513, "grad_norm": 0.5521791331631798, "learning_rate": 6.748122729062201e-06, "loss": 1.3389, "step": 5971 }, { "epoch": 0.4570641359253023, "grad_norm": 0.47983854601973414, "learning_rate": 6.747148694304726e-06, "loss": 0.9628, "step": 5972 }, { "epoch": 0.4571406704423695, "grad_norm": 0.6780457618744108, "learning_rate": 6.746174613640203e-06, "loss": 1.7069, "step": 5973 }, { "epoch": 0.4572172049594367, "grad_norm": 0.5764962077653646, "learning_rate": 6.7452004871284805e-06, "loss": 1.1618, "step": 5974 }, { "epoch": 0.4572937394765039, "grad_norm": 0.6346920685443396, "learning_rate": 6.744226314829412e-06, "loss": 1.5443, "step": 5975 }, { "epoch": 0.4573702739935711, "grad_norm": 0.6676431534655622, "learning_rate": 6.743252096802854e-06, "loss": 1.6436, "step": 5976 }, { "epoch": 0.4574468085106383, "grad_norm": 0.5733874057253574, "learning_rate": 6.742277833108667e-06, "loss": 1.3408, "step": 5977 }, { "epoch": 0.4575233430277055, "grad_norm": 0.5932626406063238, "learning_rate": 6.7413035238067094e-06, "loss": 1.4308, "step": 5978 }, { "epoch": 0.4575998775447727, "grad_norm": 0.5766636748162269, "learning_rate": 6.740329168956847e-06, "loss": 1.2958, "step": 5979 }, { "epoch": 0.4576764120618399, "grad_norm": 0.7413247822458359, "learning_rate": 6.739354768618949e-06, "loss": 1.5138, "step": 5980 }, { "epoch": 0.4577529465789071, "grad_norm": 0.5932023619466407, "learning_rate": 6.738380322852881e-06, "loss": 1.5082, "step": 5981 }, { "epoch": 0.45782948109597427, "grad_norm": 0.5691145420072959, "learning_rate": 6.73740583171852e-06, "loss": 1.2783, "step": 5982 }, { "epoch": 0.4579060156130415, "grad_norm": 0.6623588373052053, "learning_rate": 6.736431295275741e-06, "loss": 1.5156, "step": 5983 }, { "epoch": 0.4579825501301087, "grad_norm": 0.6199792969666384, "learning_rate": 6.7354567135844215e-06, "loss": 1.4206, "step": 5984 }, { "epoch": 0.4580590846471759, "grad_norm": 1.278922619671154, "learning_rate": 6.734482086704443e-06, "loss": 1.495, "step": 5985 }, { "epoch": 0.45813561916424306, "grad_norm": 0.6186307473079615, "learning_rate": 6.733507414695688e-06, "loss": 1.3709, "step": 5986 }, { "epoch": 0.4582121536813103, "grad_norm": 0.5996805193612995, "learning_rate": 6.732532697618046e-06, "loss": 1.3991, "step": 5987 }, { "epoch": 0.4582886881983775, "grad_norm": 0.5959252330329712, "learning_rate": 6.731557935531404e-06, "loss": 1.5057, "step": 5988 }, { "epoch": 0.45836522271544466, "grad_norm": 0.5742513743458192, "learning_rate": 6.730583128495657e-06, "loss": 1.1395, "step": 5989 }, { "epoch": 0.45844175723251185, "grad_norm": 0.8399683518158539, "learning_rate": 6.729608276570697e-06, "loss": 1.4759, "step": 5990 }, { "epoch": 0.4585182917495791, "grad_norm": 0.6145343316360287, "learning_rate": 6.728633379816425e-06, "loss": 1.5768, "step": 5991 }, { "epoch": 0.45859482626664627, "grad_norm": 0.5801456295383282, "learning_rate": 6.727658438292739e-06, "loss": 1.3176, "step": 5992 }, { "epoch": 0.45867136078371346, "grad_norm": 0.531588786380242, "learning_rate": 6.726683452059542e-06, "loss": 1.1236, "step": 5993 }, { "epoch": 0.45874789530078064, "grad_norm": 0.5011853259892977, "learning_rate": 6.725708421176744e-06, "loss": 1.0677, "step": 5994 }, { "epoch": 0.4588244298178478, "grad_norm": 0.5635613250785301, "learning_rate": 6.724733345704251e-06, "loss": 1.3695, "step": 5995 }, { "epoch": 0.45890096433491506, "grad_norm": 0.5986256149363586, "learning_rate": 6.723758225701973e-06, "loss": 1.4211, "step": 5996 }, { "epoch": 0.45897749885198225, "grad_norm": 0.5906421658340706, "learning_rate": 6.722783061229826e-06, "loss": 1.475, "step": 5997 }, { "epoch": 0.45905403336904943, "grad_norm": 0.7617810648091577, "learning_rate": 6.721807852347727e-06, "loss": 1.6, "step": 5998 }, { "epoch": 0.4591305678861166, "grad_norm": 0.5657230745940355, "learning_rate": 6.720832599115597e-06, "loss": 1.3213, "step": 5999 }, { "epoch": 0.45920710240318385, "grad_norm": 0.6041714466198156, "learning_rate": 6.7198573015933566e-06, "loss": 1.4325, "step": 6000 }, { "epoch": 0.45928363692025104, "grad_norm": 0.949650029108231, "learning_rate": 6.718881959840931e-06, "loss": 1.275, "step": 6001 }, { "epoch": 0.4593601714373182, "grad_norm": 0.5804650266338772, "learning_rate": 6.7179065739182495e-06, "loss": 1.616, "step": 6002 }, { "epoch": 0.4594367059543854, "grad_norm": 0.5850977254617679, "learning_rate": 6.716931143885242e-06, "loss": 1.3503, "step": 6003 }, { "epoch": 0.45951324047145264, "grad_norm": 0.6838760092955012, "learning_rate": 6.715955669801841e-06, "loss": 1.4086, "step": 6004 }, { "epoch": 0.45958977498851983, "grad_norm": 0.5635312512023505, "learning_rate": 6.714980151727986e-06, "loss": 1.4119, "step": 6005 }, { "epoch": 0.459666309505587, "grad_norm": 0.5658269424216248, "learning_rate": 6.714004589723611e-06, "loss": 1.3908, "step": 6006 }, { "epoch": 0.4597428440226542, "grad_norm": 0.6316144819759187, "learning_rate": 6.713028983848658e-06, "loss": 1.6413, "step": 6007 }, { "epoch": 0.45981937853972144, "grad_norm": 0.5442620846885375, "learning_rate": 6.712053334163074e-06, "loss": 1.429, "step": 6008 }, { "epoch": 0.4598959130567886, "grad_norm": 0.5802835431967422, "learning_rate": 6.711077640726806e-06, "loss": 1.4045, "step": 6009 }, { "epoch": 0.4599724475738558, "grad_norm": 0.6151401754070615, "learning_rate": 6.7101019035998e-06, "loss": 1.4385, "step": 6010 }, { "epoch": 0.460048982090923, "grad_norm": 0.5283227918974325, "learning_rate": 6.709126122842011e-06, "loss": 1.0857, "step": 6011 }, { "epoch": 0.4601255166079902, "grad_norm": 0.6394357142315937, "learning_rate": 6.708150298513394e-06, "loss": 1.4346, "step": 6012 }, { "epoch": 0.4602020511250574, "grad_norm": 0.6318148011218535, "learning_rate": 6.707174430673904e-06, "loss": 1.3155, "step": 6013 }, { "epoch": 0.4602785856421246, "grad_norm": 0.5543247576415451, "learning_rate": 6.706198519383504e-06, "loss": 1.2697, "step": 6014 }, { "epoch": 0.4603551201591918, "grad_norm": 0.6283182099562123, "learning_rate": 6.705222564702156e-06, "loss": 1.5386, "step": 6015 }, { "epoch": 0.460431654676259, "grad_norm": 0.4635214465241354, "learning_rate": 6.704246566689825e-06, "loss": 0.9738, "step": 6016 }, { "epoch": 0.4605081891933262, "grad_norm": 0.5396584053612402, "learning_rate": 6.70327052540648e-06, "loss": 1.1771, "step": 6017 }, { "epoch": 0.4605847237103934, "grad_norm": 0.693877752170828, "learning_rate": 6.702294440912091e-06, "loss": 1.6027, "step": 6018 }, { "epoch": 0.46066125822746057, "grad_norm": 0.5625584639952513, "learning_rate": 6.701318313266633e-06, "loss": 1.3351, "step": 6019 }, { "epoch": 0.4607377927445278, "grad_norm": 0.5727655184258756, "learning_rate": 6.70034214253008e-06, "loss": 1.3808, "step": 6020 }, { "epoch": 0.460814327261595, "grad_norm": 0.9431852406150764, "learning_rate": 6.699365928762413e-06, "loss": 1.5679, "step": 6021 }, { "epoch": 0.4608908617786622, "grad_norm": 0.5153742224856496, "learning_rate": 6.698389672023612e-06, "loss": 1.1527, "step": 6022 }, { "epoch": 0.46096739629572936, "grad_norm": 0.6500320549386936, "learning_rate": 6.697413372373662e-06, "loss": 1.4797, "step": 6023 }, { "epoch": 0.4610439308127966, "grad_norm": 0.5076316516718694, "learning_rate": 6.696437029872551e-06, "loss": 1.1612, "step": 6024 }, { "epoch": 0.4611204653298638, "grad_norm": 0.5809559838818815, "learning_rate": 6.695460644580264e-06, "loss": 1.3692, "step": 6025 }, { "epoch": 0.46119699984693097, "grad_norm": 0.5989357255495609, "learning_rate": 6.694484216556798e-06, "loss": 1.2228, "step": 6026 }, { "epoch": 0.46127353436399815, "grad_norm": 0.640053376403188, "learning_rate": 6.693507745862146e-06, "loss": 1.6767, "step": 6027 }, { "epoch": 0.46135006888106533, "grad_norm": 0.5526858494476561, "learning_rate": 6.692531232556304e-06, "loss": 1.3586, "step": 6028 }, { "epoch": 0.4614266033981326, "grad_norm": 0.5475855927428469, "learning_rate": 6.691554676699273e-06, "loss": 1.2871, "step": 6029 }, { "epoch": 0.46150313791519976, "grad_norm": 0.5371883720918442, "learning_rate": 6.690578078351054e-06, "loss": 1.381, "step": 6030 }, { "epoch": 0.46157967243226694, "grad_norm": 0.5085026539553965, "learning_rate": 6.689601437571655e-06, "loss": 1.221, "step": 6031 }, { "epoch": 0.4616562069493341, "grad_norm": 0.7230567784087938, "learning_rate": 6.6886247544210825e-06, "loss": 1.5076, "step": 6032 }, { "epoch": 0.46173274146640136, "grad_norm": 0.5195662253909207, "learning_rate": 6.687648028959345e-06, "loss": 1.1611, "step": 6033 }, { "epoch": 0.46180927598346855, "grad_norm": 0.49958513871864824, "learning_rate": 6.686671261246458e-06, "loss": 0.9993, "step": 6034 }, { "epoch": 0.46188581050053573, "grad_norm": 0.7200811874776842, "learning_rate": 6.685694451342437e-06, "loss": 1.3692, "step": 6035 }, { "epoch": 0.4619623450176029, "grad_norm": 0.6603836979162833, "learning_rate": 6.6847175993072965e-06, "loss": 1.4097, "step": 6036 }, { "epoch": 0.46203887953467015, "grad_norm": 0.6796146416998177, "learning_rate": 6.683740705201061e-06, "loss": 1.3264, "step": 6037 }, { "epoch": 0.46211541405173734, "grad_norm": 0.6089195740837522, "learning_rate": 6.682763769083753e-06, "loss": 1.3112, "step": 6038 }, { "epoch": 0.4621919485688045, "grad_norm": 0.6760407686189616, "learning_rate": 6.681786791015399e-06, "loss": 1.5466, "step": 6039 }, { "epoch": 0.4622684830858717, "grad_norm": 0.5922137535372366, "learning_rate": 6.680809771056024e-06, "loss": 1.4387, "step": 6040 }, { "epoch": 0.46234501760293895, "grad_norm": 0.6641881308974763, "learning_rate": 6.6798327092656625e-06, "loss": 1.6011, "step": 6041 }, { "epoch": 0.46242155212000613, "grad_norm": 0.5616495384132824, "learning_rate": 6.678855605704346e-06, "loss": 1.2756, "step": 6042 }, { "epoch": 0.4624980866370733, "grad_norm": 0.5613396219793865, "learning_rate": 6.677878460432113e-06, "loss": 1.3558, "step": 6043 }, { "epoch": 0.4625746211541405, "grad_norm": 0.8052018709690499, "learning_rate": 6.676901273509001e-06, "loss": 1.4815, "step": 6044 }, { "epoch": 0.46265115567120774, "grad_norm": 0.5966865139139779, "learning_rate": 6.6759240449950506e-06, "loss": 1.5151, "step": 6045 }, { "epoch": 0.4627276901882749, "grad_norm": 0.6185434316801462, "learning_rate": 6.6749467749503065e-06, "loss": 1.5438, "step": 6046 }, { "epoch": 0.4628042247053421, "grad_norm": 0.6380371405186253, "learning_rate": 6.673969463434814e-06, "loss": 1.543, "step": 6047 }, { "epoch": 0.4628807592224093, "grad_norm": 0.6032820336446283, "learning_rate": 6.672992110508624e-06, "loss": 1.371, "step": 6048 }, { "epoch": 0.4629572937394765, "grad_norm": 0.6380390891040019, "learning_rate": 6.672014716231786e-06, "loss": 1.7499, "step": 6049 }, { "epoch": 0.4630338282565437, "grad_norm": 1.1537910517227568, "learning_rate": 6.671037280664355e-06, "loss": 1.5576, "step": 6050 }, { "epoch": 0.4631103627736109, "grad_norm": 0.6216325958705884, "learning_rate": 6.670059803866389e-06, "loss": 1.4116, "step": 6051 }, { "epoch": 0.4631868972906781, "grad_norm": 0.8364557661642126, "learning_rate": 6.669082285897944e-06, "loss": 1.4477, "step": 6052 }, { "epoch": 0.4632634318077453, "grad_norm": 0.5866266675598778, "learning_rate": 6.668104726819084e-06, "loss": 1.3438, "step": 6053 }, { "epoch": 0.4633399663248125, "grad_norm": 0.5583911561738707, "learning_rate": 6.6671271266898715e-06, "loss": 1.2641, "step": 6054 }, { "epoch": 0.4634165008418797, "grad_norm": 0.5589090387011875, "learning_rate": 6.6661494855703745e-06, "loss": 1.2231, "step": 6055 }, { "epoch": 0.46349303535894687, "grad_norm": 0.5328285758419856, "learning_rate": 6.665171803520662e-06, "loss": 1.203, "step": 6056 }, { "epoch": 0.4635695698760141, "grad_norm": 0.577863085135672, "learning_rate": 6.664194080600806e-06, "loss": 1.3467, "step": 6057 }, { "epoch": 0.4636461043930813, "grad_norm": 0.6126924856945976, "learning_rate": 6.663216316870877e-06, "loss": 1.3303, "step": 6058 }, { "epoch": 0.4637226389101485, "grad_norm": 0.6241871769505078, "learning_rate": 6.6622385123909585e-06, "loss": 1.4981, "step": 6059 }, { "epoch": 0.46379917342721566, "grad_norm": 0.5380245482901038, "learning_rate": 6.661260667221124e-06, "loss": 1.1494, "step": 6060 }, { "epoch": 0.46387570794428284, "grad_norm": 0.5476125708342147, "learning_rate": 6.660282781421458e-06, "loss": 1.1409, "step": 6061 }, { "epoch": 0.4639522424613501, "grad_norm": 0.650657424957081, "learning_rate": 6.659304855052044e-06, "loss": 1.6489, "step": 6062 }, { "epoch": 0.46402877697841727, "grad_norm": 0.5674871073637201, "learning_rate": 6.658326888172968e-06, "loss": 1.2757, "step": 6063 }, { "epoch": 0.46410531149548445, "grad_norm": 0.5907703454051959, "learning_rate": 6.657348880844321e-06, "loss": 1.419, "step": 6064 }, { "epoch": 0.46418184601255164, "grad_norm": 0.7230987573011765, "learning_rate": 6.656370833126194e-06, "loss": 1.3793, "step": 6065 }, { "epoch": 0.4642583805296189, "grad_norm": 0.6059431704411005, "learning_rate": 6.655392745078679e-06, "loss": 1.4058, "step": 6066 }, { "epoch": 0.46433491504668606, "grad_norm": 0.5191579494375131, "learning_rate": 6.654414616761877e-06, "loss": 1.186, "step": 6067 }, { "epoch": 0.46441144956375324, "grad_norm": 0.8972021803016073, "learning_rate": 6.653436448235882e-06, "loss": 1.5675, "step": 6068 }, { "epoch": 0.4644879840808204, "grad_norm": 0.4944293424247877, "learning_rate": 6.652458239560799e-06, "loss": 1.0205, "step": 6069 }, { "epoch": 0.46456451859788767, "grad_norm": 0.6335610570238336, "learning_rate": 6.651479990796732e-06, "loss": 1.5129, "step": 6070 }, { "epoch": 0.46464105311495485, "grad_norm": 0.6113415388257876, "learning_rate": 6.650501702003787e-06, "loss": 1.4759, "step": 6071 }, { "epoch": 0.46471758763202203, "grad_norm": 0.5480163359896247, "learning_rate": 6.649523373242073e-06, "loss": 1.1855, "step": 6072 }, { "epoch": 0.4647941221490892, "grad_norm": 1.3562474170113357, "learning_rate": 6.6485450045717e-06, "loss": 1.2626, "step": 6073 }, { "epoch": 0.46487065666615646, "grad_norm": 0.6668699270529929, "learning_rate": 6.647566596052784e-06, "loss": 1.3823, "step": 6074 }, { "epoch": 0.46494719118322364, "grad_norm": 0.7190689549381277, "learning_rate": 6.646588147745441e-06, "loss": 1.6414, "step": 6075 }, { "epoch": 0.4650237257002908, "grad_norm": 0.5749392163386022, "learning_rate": 6.645609659709788e-06, "loss": 1.4241, "step": 6076 }, { "epoch": 0.465100260217358, "grad_norm": 0.5839639924633651, "learning_rate": 6.6446311320059484e-06, "loss": 1.4168, "step": 6077 }, { "epoch": 0.46517679473442525, "grad_norm": 1.471747046863545, "learning_rate": 6.643652564694044e-06, "loss": 2.0586, "step": 6078 }, { "epoch": 0.46525332925149243, "grad_norm": 0.5607838129687375, "learning_rate": 6.642673957834202e-06, "loss": 1.3392, "step": 6079 }, { "epoch": 0.4653298637685596, "grad_norm": 0.6232394585129772, "learning_rate": 6.6416953114865526e-06, "loss": 1.5648, "step": 6080 }, { "epoch": 0.4654063982856268, "grad_norm": 0.5621830364831687, "learning_rate": 6.640716625711223e-06, "loss": 1.3685, "step": 6081 }, { "epoch": 0.46548293280269404, "grad_norm": 0.6030310281932305, "learning_rate": 6.639737900568348e-06, "loss": 1.3308, "step": 6082 }, { "epoch": 0.4655594673197612, "grad_norm": 0.6300407474177209, "learning_rate": 6.638759136118066e-06, "loss": 1.5679, "step": 6083 }, { "epoch": 0.4656360018368284, "grad_norm": 0.5874989481974845, "learning_rate": 6.6377803324205125e-06, "loss": 1.3369, "step": 6084 }, { "epoch": 0.4657125363538956, "grad_norm": 0.5553643309361551, "learning_rate": 6.636801489535829e-06, "loss": 1.3803, "step": 6085 }, { "epoch": 0.46578907087096283, "grad_norm": 0.5671606090220114, "learning_rate": 6.635822607524158e-06, "loss": 1.2876, "step": 6086 }, { "epoch": 0.46586560538803, "grad_norm": 0.5079751105632282, "learning_rate": 6.634843686445644e-06, "loss": 1.1064, "step": 6087 }, { "epoch": 0.4659421399050972, "grad_norm": 0.7290034567201287, "learning_rate": 6.633864726360438e-06, "loss": 1.4938, "step": 6088 }, { "epoch": 0.4660186744221644, "grad_norm": 0.6349675069136727, "learning_rate": 6.632885727328687e-06, "loss": 1.7343, "step": 6089 }, { "epoch": 0.4660952089392316, "grad_norm": 0.6081161381537342, "learning_rate": 6.631906689410546e-06, "loss": 1.468, "step": 6090 }, { "epoch": 0.4661717434562988, "grad_norm": 0.6811909872094752, "learning_rate": 6.630927612666168e-06, "loss": 1.4341, "step": 6091 }, { "epoch": 0.466248277973366, "grad_norm": 0.6078193793074623, "learning_rate": 6.629948497155712e-06, "loss": 1.4625, "step": 6092 }, { "epoch": 0.46632481249043317, "grad_norm": 0.5467467080016123, "learning_rate": 6.628969342939335e-06, "loss": 1.2606, "step": 6093 }, { "epoch": 0.46640134700750036, "grad_norm": 0.5502919978597088, "learning_rate": 6.627990150077204e-06, "loss": 1.2504, "step": 6094 }, { "epoch": 0.4664778815245676, "grad_norm": 0.543355028524182, "learning_rate": 6.62701091862948e-06, "loss": 1.2615, "step": 6095 }, { "epoch": 0.4665544160416348, "grad_norm": 0.653206689516414, "learning_rate": 6.626031648656332e-06, "loss": 1.4513, "step": 6096 }, { "epoch": 0.46663095055870196, "grad_norm": 0.49569958461170294, "learning_rate": 6.625052340217927e-06, "loss": 1.0959, "step": 6097 }, { "epoch": 0.46670748507576915, "grad_norm": 0.6434652333029874, "learning_rate": 6.624072993374437e-06, "loss": 1.5688, "step": 6098 }, { "epoch": 0.4667840195928364, "grad_norm": 0.608242430780633, "learning_rate": 6.623093608186038e-06, "loss": 1.5622, "step": 6099 }, { "epoch": 0.46686055410990357, "grad_norm": 0.6511322992366855, "learning_rate": 6.622114184712903e-06, "loss": 1.6046, "step": 6100 }, { "epoch": 0.46693708862697075, "grad_norm": 0.6178804145033452, "learning_rate": 6.621134723015214e-06, "loss": 1.4818, "step": 6101 }, { "epoch": 0.46701362314403794, "grad_norm": 0.4971761201823645, "learning_rate": 6.620155223153151e-06, "loss": 0.92, "step": 6102 }, { "epoch": 0.4670901576611052, "grad_norm": 0.6520166033872201, "learning_rate": 6.619175685186896e-06, "loss": 1.4077, "step": 6103 }, { "epoch": 0.46716669217817236, "grad_norm": 0.5628734323474798, "learning_rate": 6.6181961091766375e-06, "loss": 1.3727, "step": 6104 }, { "epoch": 0.46724322669523954, "grad_norm": 0.5756820678317224, "learning_rate": 6.617216495182561e-06, "loss": 1.5173, "step": 6105 }, { "epoch": 0.46731976121230673, "grad_norm": 0.5884624108701582, "learning_rate": 6.616236843264858e-06, "loss": 1.3272, "step": 6106 }, { "epoch": 0.46739629572937397, "grad_norm": 0.5748539388848807, "learning_rate": 6.615257153483722e-06, "loss": 1.5127, "step": 6107 }, { "epoch": 0.46747283024644115, "grad_norm": 0.5961126617167469, "learning_rate": 6.614277425899347e-06, "loss": 1.4674, "step": 6108 }, { "epoch": 0.46754936476350833, "grad_norm": 0.5285203403102169, "learning_rate": 6.6132976605719305e-06, "loss": 1.1518, "step": 6109 }, { "epoch": 0.4676258992805755, "grad_norm": 0.6391056145651873, "learning_rate": 6.612317857561671e-06, "loss": 1.5797, "step": 6110 }, { "epoch": 0.46770243379764276, "grad_norm": 0.5576562121655649, "learning_rate": 6.611338016928774e-06, "loss": 1.2916, "step": 6111 }, { "epoch": 0.46777896831470994, "grad_norm": 0.8460736579856886, "learning_rate": 6.610358138733439e-06, "loss": 1.439, "step": 6112 }, { "epoch": 0.4678555028317771, "grad_norm": 0.5897357455990316, "learning_rate": 6.609378223035877e-06, "loss": 1.2411, "step": 6113 }, { "epoch": 0.4679320373488443, "grad_norm": 0.6837704511241859, "learning_rate": 6.608398269896296e-06, "loss": 1.7714, "step": 6114 }, { "epoch": 0.46800857186591155, "grad_norm": 0.7773417997596351, "learning_rate": 6.607418279374906e-06, "loss": 1.63, "step": 6115 }, { "epoch": 0.46808510638297873, "grad_norm": 0.9021000965474809, "learning_rate": 6.6064382515319205e-06, "loss": 1.3035, "step": 6116 }, { "epoch": 0.4681616409000459, "grad_norm": 0.7382287363201194, "learning_rate": 6.605458186427555e-06, "loss": 1.4988, "step": 6117 }, { "epoch": 0.4682381754171131, "grad_norm": 0.6467079530894293, "learning_rate": 6.60447808412203e-06, "loss": 1.4888, "step": 6118 }, { "epoch": 0.46831470993418034, "grad_norm": 0.6424574461635565, "learning_rate": 6.603497944675565e-06, "loss": 1.5443, "step": 6119 }, { "epoch": 0.4683912444512475, "grad_norm": 0.5920902525738718, "learning_rate": 6.602517768148382e-06, "loss": 1.4444, "step": 6120 }, { "epoch": 0.4684677789683147, "grad_norm": 0.6764638842323162, "learning_rate": 6.601537554600705e-06, "loss": 1.6349, "step": 6121 }, { "epoch": 0.4685443134853819, "grad_norm": 0.5338439501410938, "learning_rate": 6.600557304092763e-06, "loss": 1.2438, "step": 6122 }, { "epoch": 0.46862084800244913, "grad_norm": 0.5675029731011331, "learning_rate": 6.599577016684784e-06, "loss": 1.3781, "step": 6123 }, { "epoch": 0.4686973825195163, "grad_norm": 0.6658641688427136, "learning_rate": 6.598596692437001e-06, "loss": 1.4642, "step": 6124 }, { "epoch": 0.4687739170365835, "grad_norm": 0.5154379833008678, "learning_rate": 6.597616331409648e-06, "loss": 1.1335, "step": 6125 }, { "epoch": 0.4688504515536507, "grad_norm": 0.6715975867785828, "learning_rate": 6.596635933662961e-06, "loss": 1.6426, "step": 6126 }, { "epoch": 0.46892698607071787, "grad_norm": 0.7180632786238106, "learning_rate": 6.595655499257179e-06, "loss": 1.9134, "step": 6127 }, { "epoch": 0.4690035205877851, "grad_norm": 1.0592588543953503, "learning_rate": 6.594675028252544e-06, "loss": 1.4218, "step": 6128 }, { "epoch": 0.4690800551048523, "grad_norm": 0.5431028039100232, "learning_rate": 6.593694520709296e-06, "loss": 1.1841, "step": 6129 }, { "epoch": 0.4691565896219195, "grad_norm": 0.784881670357494, "learning_rate": 6.592713976687683e-06, "loss": 1.668, "step": 6130 }, { "epoch": 0.46923312413898666, "grad_norm": 0.574161019241277, "learning_rate": 6.591733396247949e-06, "loss": 1.3695, "step": 6131 }, { "epoch": 0.4693096586560539, "grad_norm": 0.6385753678646285, "learning_rate": 6.590752779450347e-06, "loss": 1.4364, "step": 6132 }, { "epoch": 0.4693861931731211, "grad_norm": 0.5875399914966538, "learning_rate": 6.589772126355129e-06, "loss": 1.3558, "step": 6133 }, { "epoch": 0.46946272769018826, "grad_norm": 0.5387432983724489, "learning_rate": 6.588791437022547e-06, "loss": 1.1162, "step": 6134 }, { "epoch": 0.46953926220725545, "grad_norm": 0.652030114754431, "learning_rate": 6.58781071151286e-06, "loss": 1.3187, "step": 6135 }, { "epoch": 0.4696157967243227, "grad_norm": 0.7434594344788653, "learning_rate": 6.586829949886326e-06, "loss": 1.6407, "step": 6136 }, { "epoch": 0.46969233124138987, "grad_norm": 0.5401553017842533, "learning_rate": 6.585849152203205e-06, "loss": 1.1199, "step": 6137 }, { "epoch": 0.46976886575845705, "grad_norm": 0.58752457727031, "learning_rate": 6.58486831852376e-06, "loss": 1.2646, "step": 6138 }, { "epoch": 0.46984540027552424, "grad_norm": 0.6001081412119396, "learning_rate": 6.583887448908259e-06, "loss": 1.4483, "step": 6139 }, { "epoch": 0.4699219347925915, "grad_norm": 0.6217106938470798, "learning_rate": 6.582906543416967e-06, "loss": 1.4489, "step": 6140 }, { "epoch": 0.46999846930965866, "grad_norm": 0.639727848355363, "learning_rate": 6.581925602110155e-06, "loss": 1.3055, "step": 6141 }, { "epoch": 0.47007500382672585, "grad_norm": 0.5884853526566708, "learning_rate": 6.580944625048093e-06, "loss": 1.432, "step": 6142 }, { "epoch": 0.47015153834379303, "grad_norm": 0.5437554839797187, "learning_rate": 6.579963612291057e-06, "loss": 1.2293, "step": 6143 }, { "epoch": 0.47022807286086027, "grad_norm": 0.5492744062501642, "learning_rate": 6.578982563899323e-06, "loss": 1.1097, "step": 6144 }, { "epoch": 0.47030460737792745, "grad_norm": 0.5987875754225983, "learning_rate": 6.578001479933169e-06, "loss": 1.4848, "step": 6145 }, { "epoch": 0.47038114189499464, "grad_norm": 0.5580547777948728, "learning_rate": 6.577020360452877e-06, "loss": 1.4344, "step": 6146 }, { "epoch": 0.4704576764120618, "grad_norm": 0.5894493132224768, "learning_rate": 6.576039205518728e-06, "loss": 1.2977, "step": 6147 }, { "epoch": 0.47053421092912906, "grad_norm": 0.6174363166931293, "learning_rate": 6.575058015191009e-06, "loss": 1.5389, "step": 6148 }, { "epoch": 0.47061074544619624, "grad_norm": 0.6509959027413473, "learning_rate": 6.574076789530006e-06, "loss": 1.6271, "step": 6149 }, { "epoch": 0.4706872799632634, "grad_norm": 0.5448546073669942, "learning_rate": 6.573095528596011e-06, "loss": 1.1314, "step": 6150 }, { "epoch": 0.4707638144803306, "grad_norm": 0.5961510323558668, "learning_rate": 6.572114232449311e-06, "loss": 1.3007, "step": 6151 }, { "epoch": 0.47084034899739785, "grad_norm": 0.6924355869982001, "learning_rate": 6.571132901150204e-06, "loss": 1.3635, "step": 6152 }, { "epoch": 0.47091688351446503, "grad_norm": 0.6537375977256787, "learning_rate": 6.570151534758983e-06, "loss": 1.588, "step": 6153 }, { "epoch": 0.4709934180315322, "grad_norm": 0.595272646694587, "learning_rate": 6.569170133335946e-06, "loss": 1.4771, "step": 6154 }, { "epoch": 0.4710699525485994, "grad_norm": 0.6020689620787523, "learning_rate": 6.568188696941396e-06, "loss": 1.5624, "step": 6155 }, { "epoch": 0.47114648706566664, "grad_norm": 0.5397798374593662, "learning_rate": 6.5672072256356345e-06, "loss": 1.3256, "step": 6156 }, { "epoch": 0.4712230215827338, "grad_norm": 0.5138203558680763, "learning_rate": 6.566225719478965e-06, "loss": 1.0032, "step": 6157 }, { "epoch": 0.471299556099801, "grad_norm": 0.5017064486688231, "learning_rate": 6.565244178531693e-06, "loss": 0.9999, "step": 6158 }, { "epoch": 0.4713760906168682, "grad_norm": 0.5227310597416975, "learning_rate": 6.5642626028541316e-06, "loss": 1.1337, "step": 6159 }, { "epoch": 0.47145262513393543, "grad_norm": 0.5532819822291705, "learning_rate": 6.563280992506587e-06, "loss": 1.2913, "step": 6160 }, { "epoch": 0.4715291596510026, "grad_norm": 0.6776917623394635, "learning_rate": 6.562299347549375e-06, "loss": 1.5226, "step": 6161 }, { "epoch": 0.4716056941680698, "grad_norm": 0.5941181461271762, "learning_rate": 6.56131766804281e-06, "loss": 1.2237, "step": 6162 }, { "epoch": 0.471682228685137, "grad_norm": 0.657484556648269, "learning_rate": 6.560335954047211e-06, "loss": 1.5811, "step": 6163 }, { "epoch": 0.47175876320220417, "grad_norm": 0.5871538209876067, "learning_rate": 6.5593542056228945e-06, "loss": 1.2144, "step": 6164 }, { "epoch": 0.4718352977192714, "grad_norm": 0.8366305668114372, "learning_rate": 6.558372422830183e-06, "loss": 1.3625, "step": 6165 }, { "epoch": 0.4719118322363386, "grad_norm": 0.6797134083087276, "learning_rate": 6.557390605729401e-06, "loss": 1.6836, "step": 6166 }, { "epoch": 0.4719883667534058, "grad_norm": 0.5912316197491374, "learning_rate": 6.556408754380873e-06, "loss": 1.5326, "step": 6167 }, { "epoch": 0.47206490127047296, "grad_norm": 0.8129253390633073, "learning_rate": 6.555426868844929e-06, "loss": 1.6579, "step": 6168 }, { "epoch": 0.4721414357875402, "grad_norm": 0.5362493624351913, "learning_rate": 6.554444949181898e-06, "loss": 1.2381, "step": 6169 }, { "epoch": 0.4722179703046074, "grad_norm": 0.5347481835838045, "learning_rate": 6.553462995452111e-06, "loss": 1.1547, "step": 6170 }, { "epoch": 0.47229450482167457, "grad_norm": 0.8752950044386316, "learning_rate": 6.552481007715904e-06, "loss": 1.5604, "step": 6171 }, { "epoch": 0.47237103933874175, "grad_norm": 0.7169111815494729, "learning_rate": 6.551498986033612e-06, "loss": 1.3394, "step": 6172 }, { "epoch": 0.472447573855809, "grad_norm": 0.5013455850491416, "learning_rate": 6.550516930465572e-06, "loss": 1.076, "step": 6173 }, { "epoch": 0.4725241083728762, "grad_norm": 0.6050883173090211, "learning_rate": 6.5495348410721285e-06, "loss": 1.3832, "step": 6174 }, { "epoch": 0.47260064288994336, "grad_norm": 0.5143251450299109, "learning_rate": 6.548552717913621e-06, "loss": 1.1217, "step": 6175 }, { "epoch": 0.47267717740701054, "grad_norm": 0.507753960294388, "learning_rate": 6.547570561050393e-06, "loss": 1.0868, "step": 6176 }, { "epoch": 0.4727537119240778, "grad_norm": 0.5555032483450274, "learning_rate": 6.546588370542795e-06, "loss": 1.177, "step": 6177 }, { "epoch": 0.47283024644114496, "grad_norm": 0.6169753934626061, "learning_rate": 6.545606146451172e-06, "loss": 1.2509, "step": 6178 }, { "epoch": 0.47290678095821215, "grad_norm": 1.084247127320877, "learning_rate": 6.544623888835877e-06, "loss": 1.6235, "step": 6179 }, { "epoch": 0.47298331547527933, "grad_norm": 0.7065134221776324, "learning_rate": 6.543641597757262e-06, "loss": 1.2816, "step": 6180 }, { "epoch": 0.47305984999234657, "grad_norm": 0.48087663746490017, "learning_rate": 6.542659273275681e-06, "loss": 0.9863, "step": 6181 }, { "epoch": 0.47313638450941375, "grad_norm": 0.5854601847839574, "learning_rate": 6.541676915451494e-06, "loss": 1.4266, "step": 6182 }, { "epoch": 0.47321291902648094, "grad_norm": 0.6106545876494935, "learning_rate": 6.540694524345057e-06, "loss": 1.4406, "step": 6183 }, { "epoch": 0.4732894535435481, "grad_norm": 0.6320434406111359, "learning_rate": 6.539712100016731e-06, "loss": 1.5124, "step": 6184 }, { "epoch": 0.47336598806061536, "grad_norm": 0.6418205618767489, "learning_rate": 6.538729642526882e-06, "loss": 1.3962, "step": 6185 }, { "epoch": 0.47344252257768255, "grad_norm": 0.4836518921124787, "learning_rate": 6.537747151935873e-06, "loss": 1.0345, "step": 6186 }, { "epoch": 0.47351905709474973, "grad_norm": 0.5477247630987988, "learning_rate": 6.536764628304071e-06, "loss": 1.2535, "step": 6187 }, { "epoch": 0.4735955916118169, "grad_norm": 0.7331183440660444, "learning_rate": 6.5357820716918455e-06, "loss": 1.7203, "step": 6188 }, { "epoch": 0.47367212612888415, "grad_norm": 0.5993441395520993, "learning_rate": 6.534799482159568e-06, "loss": 1.3895, "step": 6189 }, { "epoch": 0.47374866064595134, "grad_norm": 0.5955106191833363, "learning_rate": 6.533816859767611e-06, "loss": 1.4995, "step": 6190 }, { "epoch": 0.4738251951630185, "grad_norm": 0.6992864100688312, "learning_rate": 6.532834204576351e-06, "loss": 1.4639, "step": 6191 }, { "epoch": 0.4739017296800857, "grad_norm": 0.5684854326696238, "learning_rate": 6.531851516646164e-06, "loss": 1.392, "step": 6192 }, { "epoch": 0.47397826419715294, "grad_norm": 0.5219095595633945, "learning_rate": 6.530868796037431e-06, "loss": 1.1135, "step": 6193 }, { "epoch": 0.4740547987142201, "grad_norm": 0.5840452372835055, "learning_rate": 6.529886042810532e-06, "loss": 1.3247, "step": 6194 }, { "epoch": 0.4741313332312873, "grad_norm": 0.6856369251511134, "learning_rate": 6.528903257025848e-06, "loss": 1.7067, "step": 6195 }, { "epoch": 0.4742078677483545, "grad_norm": 0.6074058495238711, "learning_rate": 6.527920438743769e-06, "loss": 1.4668, "step": 6196 }, { "epoch": 0.4742844022654217, "grad_norm": 0.6146841691417203, "learning_rate": 6.52693758802468e-06, "loss": 1.5147, "step": 6197 }, { "epoch": 0.4743609367824889, "grad_norm": 0.4992440153924983, "learning_rate": 6.5259547049289706e-06, "loss": 1.1174, "step": 6198 }, { "epoch": 0.4744374712995561, "grad_norm": 0.5645159618949159, "learning_rate": 6.524971789517031e-06, "loss": 1.2452, "step": 6199 }, { "epoch": 0.4745140058166233, "grad_norm": 0.4957562431898239, "learning_rate": 6.523988841849255e-06, "loss": 1.1142, "step": 6200 }, { "epoch": 0.47459054033369047, "grad_norm": 0.5447104362981163, "learning_rate": 6.5230058619860396e-06, "loss": 1.0619, "step": 6201 }, { "epoch": 0.4746670748507577, "grad_norm": 0.5802995629105465, "learning_rate": 6.522022849987779e-06, "loss": 1.3382, "step": 6202 }, { "epoch": 0.4747436093678249, "grad_norm": 0.6193363717497833, "learning_rate": 6.521039805914874e-06, "loss": 1.3821, "step": 6203 }, { "epoch": 0.4748201438848921, "grad_norm": 0.5918838621667324, "learning_rate": 6.520056729827727e-06, "loss": 1.4157, "step": 6204 }, { "epoch": 0.47489667840195926, "grad_norm": 0.5947191966447103, "learning_rate": 6.51907362178674e-06, "loss": 1.6024, "step": 6205 }, { "epoch": 0.4749732129190265, "grad_norm": 0.60851299492721, "learning_rate": 6.518090481852316e-06, "loss": 1.4457, "step": 6206 }, { "epoch": 0.4750497474360937, "grad_norm": 0.6485249301781435, "learning_rate": 6.5171073100848645e-06, "loss": 1.6442, "step": 6207 }, { "epoch": 0.47512628195316087, "grad_norm": 0.6389507895217311, "learning_rate": 6.516124106544796e-06, "loss": 1.526, "step": 6208 }, { "epoch": 0.47520281647022805, "grad_norm": 0.4791712628507261, "learning_rate": 6.515140871292518e-06, "loss": 1.0922, "step": 6209 }, { "epoch": 0.4752793509872953, "grad_norm": 0.5864312599998984, "learning_rate": 6.514157604388444e-06, "loss": 1.5404, "step": 6210 }, { "epoch": 0.4753558855043625, "grad_norm": 0.6170510886164733, "learning_rate": 6.5131743058929904e-06, "loss": 1.6154, "step": 6211 }, { "epoch": 0.47543242002142966, "grad_norm": 0.5629337334411818, "learning_rate": 6.512190975866575e-06, "loss": 1.3807, "step": 6212 }, { "epoch": 0.47550895453849684, "grad_norm": 0.8298392863944821, "learning_rate": 6.511207614369613e-06, "loss": 1.843, "step": 6213 }, { "epoch": 0.4755854890555641, "grad_norm": 0.4882669433699139, "learning_rate": 6.510224221462527e-06, "loss": 0.991, "step": 6214 }, { "epoch": 0.47566202357263127, "grad_norm": 0.5949510271571571, "learning_rate": 6.50924079720574e-06, "loss": 1.529, "step": 6215 }, { "epoch": 0.47573855808969845, "grad_norm": 0.5006778921284034, "learning_rate": 6.508257341659677e-06, "loss": 1.1036, "step": 6216 }, { "epoch": 0.47581509260676563, "grad_norm": 0.48008335914792916, "learning_rate": 6.507273854884762e-06, "loss": 1.0232, "step": 6217 }, { "epoch": 0.47589162712383287, "grad_norm": 0.551201970058002, "learning_rate": 6.506290336941427e-06, "loss": 1.3036, "step": 6218 }, { "epoch": 0.47596816164090006, "grad_norm": 1.8123286667403353, "learning_rate": 6.505306787890097e-06, "loss": 1.6582, "step": 6219 }, { "epoch": 0.47604469615796724, "grad_norm": 0.6176405223470498, "learning_rate": 6.50432320779121e-06, "loss": 1.4749, "step": 6220 }, { "epoch": 0.4761212306750344, "grad_norm": 0.5909334974385649, "learning_rate": 6.503339596705197e-06, "loss": 1.2985, "step": 6221 }, { "epoch": 0.47619776519210166, "grad_norm": 0.5635731506091842, "learning_rate": 6.502355954692493e-06, "loss": 1.5202, "step": 6222 }, { "epoch": 0.47627429970916885, "grad_norm": 0.5942134876774681, "learning_rate": 6.501372281813539e-06, "loss": 1.3635, "step": 6223 }, { "epoch": 0.47635083422623603, "grad_norm": 0.7380277328474647, "learning_rate": 6.500388578128773e-06, "loss": 1.3858, "step": 6224 }, { "epoch": 0.4764273687433032, "grad_norm": 0.6304017358351195, "learning_rate": 6.499404843698636e-06, "loss": 1.6097, "step": 6225 }, { "epoch": 0.47650390326037045, "grad_norm": 0.6319298892401098, "learning_rate": 6.4984210785835734e-06, "loss": 1.6001, "step": 6226 }, { "epoch": 0.47658043777743764, "grad_norm": 0.6407803855776115, "learning_rate": 6.497437282844029e-06, "loss": 1.3511, "step": 6227 }, { "epoch": 0.4766569722945048, "grad_norm": 0.6467571223664526, "learning_rate": 6.496453456540451e-06, "loss": 1.5133, "step": 6228 }, { "epoch": 0.476733506811572, "grad_norm": 0.5161291931629362, "learning_rate": 6.495469599733289e-06, "loss": 1.0747, "step": 6229 }, { "epoch": 0.4768100413286392, "grad_norm": 0.6036788804670908, "learning_rate": 6.494485712482993e-06, "loss": 1.3006, "step": 6230 }, { "epoch": 0.47688657584570643, "grad_norm": 0.5721309341514134, "learning_rate": 6.493501794850018e-06, "loss": 1.2084, "step": 6231 }, { "epoch": 0.4769631103627736, "grad_norm": 0.5397825301651101, "learning_rate": 6.492517846894817e-06, "loss": 1.1819, "step": 6232 }, { "epoch": 0.4770396448798408, "grad_norm": 0.5690394664586585, "learning_rate": 6.491533868677845e-06, "loss": 1.3056, "step": 6233 }, { "epoch": 0.477116179396908, "grad_norm": 0.5571396142346369, "learning_rate": 6.490549860259565e-06, "loss": 1.4374, "step": 6234 }, { "epoch": 0.4771927139139752, "grad_norm": 0.5901626616758692, "learning_rate": 6.4895658217004345e-06, "loss": 1.4798, "step": 6235 }, { "epoch": 0.4772692484310424, "grad_norm": 0.6294545969556019, "learning_rate": 6.4885817530609154e-06, "loss": 1.6061, "step": 6236 }, { "epoch": 0.4773457829481096, "grad_norm": 0.6230760120578978, "learning_rate": 6.487597654401474e-06, "loss": 1.258, "step": 6237 }, { "epoch": 0.47742231746517677, "grad_norm": 0.6944650360388018, "learning_rate": 6.486613525782574e-06, "loss": 1.758, "step": 6238 }, { "epoch": 0.477498851982244, "grad_norm": 0.6925924441699143, "learning_rate": 6.485629367264686e-06, "loss": 1.2697, "step": 6239 }, { "epoch": 0.4775753864993112, "grad_norm": 0.5879247242124378, "learning_rate": 6.484645178908277e-06, "loss": 1.3469, "step": 6240 }, { "epoch": 0.4776519210163784, "grad_norm": 0.6067881737495464, "learning_rate": 6.483660960773819e-06, "loss": 1.3763, "step": 6241 }, { "epoch": 0.47772845553344556, "grad_norm": 0.7119975031465219, "learning_rate": 6.482676712921786e-06, "loss": 1.4758, "step": 6242 }, { "epoch": 0.4778049900505128, "grad_norm": 0.6145755806421237, "learning_rate": 6.481692435412653e-06, "loss": 1.5414, "step": 6243 }, { "epoch": 0.47788152456758, "grad_norm": 0.5391481764298599, "learning_rate": 6.4807081283068964e-06, "loss": 1.3257, "step": 6244 }, { "epoch": 0.47795805908464717, "grad_norm": 0.5745639085425986, "learning_rate": 6.479723791664996e-06, "loss": 1.2846, "step": 6245 }, { "epoch": 0.47803459360171435, "grad_norm": 0.5395604639700669, "learning_rate": 6.478739425547431e-06, "loss": 1.1411, "step": 6246 }, { "epoch": 0.4781111281187816, "grad_norm": 0.6179351661229191, "learning_rate": 6.477755030014686e-06, "loss": 1.6739, "step": 6247 }, { "epoch": 0.4781876626358488, "grad_norm": 0.5410247263149902, "learning_rate": 6.476770605127242e-06, "loss": 1.1668, "step": 6248 }, { "epoch": 0.47826419715291596, "grad_norm": 0.6012623426553194, "learning_rate": 6.475786150945585e-06, "loss": 1.6673, "step": 6249 }, { "epoch": 0.47834073166998314, "grad_norm": 0.6892383750546757, "learning_rate": 6.474801667530208e-06, "loss": 1.3776, "step": 6250 }, { "epoch": 0.4784172661870504, "grad_norm": 1.8079924206930784, "learning_rate": 6.4738171549415965e-06, "loss": 1.1399, "step": 6251 }, { "epoch": 0.47849380070411757, "grad_norm": 0.6247048940395535, "learning_rate": 6.472832613240243e-06, "loss": 1.2475, "step": 6252 }, { "epoch": 0.47857033522118475, "grad_norm": 0.6705833199119119, "learning_rate": 6.471848042486639e-06, "loss": 1.2613, "step": 6253 }, { "epoch": 0.47864686973825193, "grad_norm": 0.5850062979463745, "learning_rate": 6.470863442741281e-06, "loss": 1.5011, "step": 6254 }, { "epoch": 0.4787234042553192, "grad_norm": 0.6687065477747841, "learning_rate": 6.469878814064665e-06, "loss": 1.7038, "step": 6255 }, { "epoch": 0.47879993877238636, "grad_norm": 0.7695526874481757, "learning_rate": 6.468894156517291e-06, "loss": 1.8136, "step": 6256 }, { "epoch": 0.47887647328945354, "grad_norm": 0.6143480124137709, "learning_rate": 6.467909470159658e-06, "loss": 1.3713, "step": 6257 }, { "epoch": 0.4789530078065207, "grad_norm": 0.5804568755466852, "learning_rate": 6.466924755052267e-06, "loss": 1.2213, "step": 6258 }, { "epoch": 0.47902954232358796, "grad_norm": 0.5565061035120071, "learning_rate": 6.465940011255625e-06, "loss": 1.3571, "step": 6259 }, { "epoch": 0.47910607684065515, "grad_norm": 0.8159710871103397, "learning_rate": 6.464955238830235e-06, "loss": 1.737, "step": 6260 }, { "epoch": 0.47918261135772233, "grad_norm": 0.5397167701705051, "learning_rate": 6.463970437836606e-06, "loss": 1.2204, "step": 6261 }, { "epoch": 0.4792591458747895, "grad_norm": 1.1373023849460553, "learning_rate": 6.462985608335246e-06, "loss": 1.3554, "step": 6262 }, { "epoch": 0.4793356803918567, "grad_norm": 0.5250826117057118, "learning_rate": 6.462000750386665e-06, "loss": 1.2398, "step": 6263 }, { "epoch": 0.47941221490892394, "grad_norm": 0.6842607262680294, "learning_rate": 6.461015864051379e-06, "loss": 1.8155, "step": 6264 }, { "epoch": 0.4794887494259911, "grad_norm": 0.7042286973322092, "learning_rate": 6.460030949389898e-06, "loss": 1.5117, "step": 6265 }, { "epoch": 0.4795652839430583, "grad_norm": 0.5269198202292508, "learning_rate": 6.4590460064627414e-06, "loss": 1.2464, "step": 6266 }, { "epoch": 0.4796418184601255, "grad_norm": 0.6543180337174552, "learning_rate": 6.458061035330425e-06, "loss": 1.4118, "step": 6267 }, { "epoch": 0.47971835297719273, "grad_norm": 0.5430454709179893, "learning_rate": 6.45707603605347e-06, "loss": 1.3293, "step": 6268 }, { "epoch": 0.4797948874942599, "grad_norm": 0.5541526133761362, "learning_rate": 6.4560910086923955e-06, "loss": 1.2743, "step": 6269 }, { "epoch": 0.4798714220113271, "grad_norm": 2.633859669235553, "learning_rate": 6.455105953307727e-06, "loss": 1.4219, "step": 6270 }, { "epoch": 0.4799479565283943, "grad_norm": 0.950039686849304, "learning_rate": 6.454120869959989e-06, "loss": 1.7637, "step": 6271 }, { "epoch": 0.4800244910454615, "grad_norm": 0.6412035350189518, "learning_rate": 6.453135758709706e-06, "loss": 1.7185, "step": 6272 }, { "epoch": 0.4801010255625287, "grad_norm": 0.9550699966377345, "learning_rate": 6.452150619617408e-06, "loss": 1.4859, "step": 6273 }, { "epoch": 0.4801775600795959, "grad_norm": 0.4973474479665333, "learning_rate": 6.451165452743624e-06, "loss": 1.1021, "step": 6274 }, { "epoch": 0.4802540945966631, "grad_norm": 0.5924902997191832, "learning_rate": 6.450180258148887e-06, "loss": 1.3406, "step": 6275 }, { "epoch": 0.4803306291137303, "grad_norm": 0.5811043770557867, "learning_rate": 6.44919503589373e-06, "loss": 1.5228, "step": 6276 }, { "epoch": 0.4804071636307975, "grad_norm": 0.6196485750792152, "learning_rate": 6.448209786038682e-06, "loss": 1.7453, "step": 6277 }, { "epoch": 0.4804836981478647, "grad_norm": 0.5806920166163667, "learning_rate": 6.44722450864429e-06, "loss": 1.2669, "step": 6278 }, { "epoch": 0.48056023266493186, "grad_norm": 0.7915307752429251, "learning_rate": 6.446239203771086e-06, "loss": 1.7151, "step": 6279 }, { "epoch": 0.4806367671819991, "grad_norm": 1.230545067969959, "learning_rate": 6.445253871479609e-06, "loss": 1.4346, "step": 6280 }, { "epoch": 0.4807133016990663, "grad_norm": 0.6439439210045228, "learning_rate": 6.444268511830405e-06, "loss": 1.2592, "step": 6281 }, { "epoch": 0.48078983621613347, "grad_norm": 0.6038798422496472, "learning_rate": 6.443283124884014e-06, "loss": 1.3313, "step": 6282 }, { "epoch": 0.48086637073320065, "grad_norm": 0.6234888885601987, "learning_rate": 6.4422977107009855e-06, "loss": 1.5238, "step": 6283 }, { "epoch": 0.4809429052502679, "grad_norm": 0.47878459620442054, "learning_rate": 6.441312269341862e-06, "loss": 1.0229, "step": 6284 }, { "epoch": 0.4810194397673351, "grad_norm": 0.6649271602838935, "learning_rate": 6.440326800867194e-06, "loss": 1.4697, "step": 6285 }, { "epoch": 0.48109597428440226, "grad_norm": 0.6876126308048247, "learning_rate": 6.43934130533753e-06, "loss": 1.9135, "step": 6286 }, { "epoch": 0.48117250880146945, "grad_norm": 0.6142760123395886, "learning_rate": 6.4383557828134235e-06, "loss": 1.5383, "step": 6287 }, { "epoch": 0.4812490433185367, "grad_norm": 0.5539036624934235, "learning_rate": 6.437370233355426e-06, "loss": 1.2521, "step": 6288 }, { "epoch": 0.48132557783560387, "grad_norm": 0.586190965660945, "learning_rate": 6.436384657024095e-06, "loss": 1.348, "step": 6289 }, { "epoch": 0.48140211235267105, "grad_norm": 0.6594855038089991, "learning_rate": 6.435399053879987e-06, "loss": 1.6103, "step": 6290 }, { "epoch": 0.48147864686973824, "grad_norm": 0.6146904655808318, "learning_rate": 6.4344134239836575e-06, "loss": 1.286, "step": 6291 }, { "epoch": 0.4815551813868055, "grad_norm": 0.5546168079835889, "learning_rate": 6.43342776739567e-06, "loss": 1.3581, "step": 6292 }, { "epoch": 0.48163171590387266, "grad_norm": 0.5856626265492593, "learning_rate": 6.432442084176584e-06, "loss": 1.3975, "step": 6293 }, { "epoch": 0.48170825042093984, "grad_norm": 0.6826590104370766, "learning_rate": 6.431456374386964e-06, "loss": 1.7201, "step": 6294 }, { "epoch": 0.481784784938007, "grad_norm": 0.5704838707113266, "learning_rate": 6.430470638087375e-06, "loss": 1.4959, "step": 6295 }, { "epoch": 0.48186131945507427, "grad_norm": 0.6529339989272296, "learning_rate": 6.429484875338381e-06, "loss": 1.3972, "step": 6296 }, { "epoch": 0.48193785397214145, "grad_norm": 0.6045645878623488, "learning_rate": 6.428499086200554e-06, "loss": 1.469, "step": 6297 }, { "epoch": 0.48201438848920863, "grad_norm": 0.58144560879394, "learning_rate": 6.427513270734463e-06, "loss": 1.4993, "step": 6298 }, { "epoch": 0.4820909230062758, "grad_norm": 0.5482859093683478, "learning_rate": 6.426527429000678e-06, "loss": 1.1089, "step": 6299 }, { "epoch": 0.482167457523343, "grad_norm": 0.6506489958905627, "learning_rate": 6.4255415610597734e-06, "loss": 1.2689, "step": 6300 }, { "epoch": 0.48224399204041024, "grad_norm": 0.5418500961775798, "learning_rate": 6.424555666972322e-06, "loss": 1.294, "step": 6301 }, { "epoch": 0.4823205265574774, "grad_norm": 0.5820077794306129, "learning_rate": 6.423569746798901e-06, "loss": 1.4015, "step": 6302 }, { "epoch": 0.4823970610745446, "grad_norm": 0.5973037862525642, "learning_rate": 6.42258380060009e-06, "loss": 1.2579, "step": 6303 }, { "epoch": 0.4824735955916118, "grad_norm": 1.9452504543362994, "learning_rate": 6.421597828436464e-06, "loss": 1.6086, "step": 6304 }, { "epoch": 0.48255013010867903, "grad_norm": 0.5588854906197683, "learning_rate": 6.420611830368611e-06, "loss": 1.0658, "step": 6305 }, { "epoch": 0.4826266646257462, "grad_norm": 0.5581905621091177, "learning_rate": 6.419625806457107e-06, "loss": 1.3848, "step": 6306 }, { "epoch": 0.4827031991428134, "grad_norm": 0.5191389735934903, "learning_rate": 6.418639756762539e-06, "loss": 1.2567, "step": 6307 }, { "epoch": 0.4827797336598806, "grad_norm": 0.5290831829279046, "learning_rate": 6.417653681345493e-06, "loss": 1.1658, "step": 6308 }, { "epoch": 0.4828562681769478, "grad_norm": 0.5448174978621705, "learning_rate": 6.4166675802665575e-06, "loss": 1.3576, "step": 6309 }, { "epoch": 0.482932802694015, "grad_norm": 0.7070314757003603, "learning_rate": 6.41568145358632e-06, "loss": 1.776, "step": 6310 }, { "epoch": 0.4830093372110822, "grad_norm": 1.0246971650473422, "learning_rate": 6.414695301365372e-06, "loss": 2.0481, "step": 6311 }, { "epoch": 0.4830858717281494, "grad_norm": 0.5681931944272987, "learning_rate": 6.4137091236643045e-06, "loss": 1.3022, "step": 6312 }, { "epoch": 0.4831624062452166, "grad_norm": 0.5714813800927153, "learning_rate": 6.41272292054371e-06, "loss": 1.4591, "step": 6313 }, { "epoch": 0.4832389407622838, "grad_norm": 0.5159674622934519, "learning_rate": 6.411736692064188e-06, "loss": 1.1539, "step": 6314 }, { "epoch": 0.483315475279351, "grad_norm": 0.6299014680459261, "learning_rate": 6.41075043828633e-06, "loss": 1.3573, "step": 6315 }, { "epoch": 0.48339200979641817, "grad_norm": 0.549092481354445, "learning_rate": 6.409764159270738e-06, "loss": 1.2551, "step": 6316 }, { "epoch": 0.4834685443134854, "grad_norm": 0.6408772961405194, "learning_rate": 6.408777855078014e-06, "loss": 1.688, "step": 6317 }, { "epoch": 0.4835450788305526, "grad_norm": 0.5691568412343524, "learning_rate": 6.407791525768751e-06, "loss": 1.4659, "step": 6318 }, { "epoch": 0.48362161334761977, "grad_norm": 0.8075259899339174, "learning_rate": 6.4068051714035625e-06, "loss": 1.846, "step": 6319 }, { "epoch": 0.48369814786468696, "grad_norm": 0.6438596942521823, "learning_rate": 6.405818792043047e-06, "loss": 1.5146, "step": 6320 }, { "epoch": 0.4837746823817542, "grad_norm": 0.6225931459855432, "learning_rate": 6.4048323877478114e-06, "loss": 1.5672, "step": 6321 }, { "epoch": 0.4838512168988214, "grad_norm": 0.5287054692953748, "learning_rate": 6.403845958578464e-06, "loss": 1.1699, "step": 6322 }, { "epoch": 0.48392775141588856, "grad_norm": 0.6473372551230426, "learning_rate": 6.402859504595612e-06, "loss": 1.6485, "step": 6323 }, { "epoch": 0.48400428593295575, "grad_norm": 1.371526357604555, "learning_rate": 6.401873025859869e-06, "loss": 1.8257, "step": 6324 }, { "epoch": 0.484080820450023, "grad_norm": 0.7876856800265521, "learning_rate": 6.400886522431846e-06, "loss": 1.5559, "step": 6325 }, { "epoch": 0.48415735496709017, "grad_norm": 0.6606549810836397, "learning_rate": 6.399899994372156e-06, "loss": 1.5205, "step": 6326 }, { "epoch": 0.48423388948415735, "grad_norm": 0.5649456448332382, "learning_rate": 6.398913441741415e-06, "loss": 1.4877, "step": 6327 }, { "epoch": 0.48431042400122454, "grad_norm": 0.48407257677966603, "learning_rate": 6.397926864600242e-06, "loss": 1.1009, "step": 6328 }, { "epoch": 0.4843869585182918, "grad_norm": 0.5215663797556296, "learning_rate": 6.396940263009249e-06, "loss": 1.1958, "step": 6329 }, { "epoch": 0.48446349303535896, "grad_norm": 0.6976357824041771, "learning_rate": 6.395953637029062e-06, "loss": 1.4153, "step": 6330 }, { "epoch": 0.48454002755242614, "grad_norm": 0.5626589792156202, "learning_rate": 6.3949669867203e-06, "loss": 1.2117, "step": 6331 }, { "epoch": 0.48461656206949333, "grad_norm": 0.597397443190564, "learning_rate": 6.3939803121435855e-06, "loss": 1.3491, "step": 6332 }, { "epoch": 0.4846930965865605, "grad_norm": 0.6921124816318626, "learning_rate": 6.392993613359542e-06, "loss": 1.5823, "step": 6333 }, { "epoch": 0.48476963110362775, "grad_norm": 0.5219260466019934, "learning_rate": 6.392006890428795e-06, "loss": 1.1728, "step": 6334 }, { "epoch": 0.48484616562069494, "grad_norm": 0.5849444989941619, "learning_rate": 6.391020143411974e-06, "loss": 1.4212, "step": 6335 }, { "epoch": 0.4849227001377621, "grad_norm": 0.5684120008127658, "learning_rate": 6.390033372369708e-06, "loss": 1.4824, "step": 6336 }, { "epoch": 0.4849992346548293, "grad_norm": 0.6026312973565421, "learning_rate": 6.389046577362623e-06, "loss": 1.3948, "step": 6337 }, { "epoch": 0.48507576917189654, "grad_norm": 0.6058660341102217, "learning_rate": 6.388059758451355e-06, "loss": 1.1963, "step": 6338 }, { "epoch": 0.4851523036889637, "grad_norm": 0.6311494654895817, "learning_rate": 6.387072915696535e-06, "loss": 1.1734, "step": 6339 }, { "epoch": 0.4852288382060309, "grad_norm": 0.5267059683486935, "learning_rate": 6.386086049158797e-06, "loss": 1.1223, "step": 6340 }, { "epoch": 0.4853053727230981, "grad_norm": 0.5442062066807799, "learning_rate": 6.38509915889878e-06, "loss": 1.3391, "step": 6341 }, { "epoch": 0.48538190724016533, "grad_norm": 0.6185391973124366, "learning_rate": 6.384112244977117e-06, "loss": 1.526, "step": 6342 }, { "epoch": 0.4854584417572325, "grad_norm": 0.5762349154972863, "learning_rate": 6.383125307454452e-06, "loss": 1.2407, "step": 6343 }, { "epoch": 0.4855349762742997, "grad_norm": 0.7853540063388259, "learning_rate": 6.38213834639142e-06, "loss": 1.3438, "step": 6344 }, { "epoch": 0.4856115107913669, "grad_norm": 1.1014374559699722, "learning_rate": 6.381151361848668e-06, "loss": 1.3161, "step": 6345 }, { "epoch": 0.4856880453084341, "grad_norm": 0.6328601925359707, "learning_rate": 6.380164353886835e-06, "loss": 1.4755, "step": 6346 }, { "epoch": 0.4857645798255013, "grad_norm": 0.5466981079997356, "learning_rate": 6.379177322566569e-06, "loss": 1.2025, "step": 6347 }, { "epoch": 0.4858411143425685, "grad_norm": 0.5838539535633623, "learning_rate": 6.378190267948514e-06, "loss": 1.2905, "step": 6348 }, { "epoch": 0.4859176488596357, "grad_norm": 0.6408789733660336, "learning_rate": 6.377203190093318e-06, "loss": 1.6515, "step": 6349 }, { "epoch": 0.4859941833767029, "grad_norm": 0.5912252892545065, "learning_rate": 6.376216089061632e-06, "loss": 1.5517, "step": 6350 }, { "epoch": 0.4860707178937701, "grad_norm": 0.631092396099642, "learning_rate": 6.375228964914103e-06, "loss": 1.5232, "step": 6351 }, { "epoch": 0.4861472524108373, "grad_norm": 0.5976480024470011, "learning_rate": 6.374241817711386e-06, "loss": 1.4409, "step": 6352 }, { "epoch": 0.48622378692790447, "grad_norm": 0.6547055939863089, "learning_rate": 6.373254647514132e-06, "loss": 1.5929, "step": 6353 }, { "epoch": 0.4863003214449717, "grad_norm": 0.6306909105966789, "learning_rate": 6.372267454382998e-06, "loss": 1.4562, "step": 6354 }, { "epoch": 0.4863768559620389, "grad_norm": 0.5383720593713393, "learning_rate": 6.371280238378637e-06, "loss": 1.2298, "step": 6355 }, { "epoch": 0.4864533904791061, "grad_norm": 0.5823493506821756, "learning_rate": 6.370292999561709e-06, "loss": 1.3645, "step": 6356 }, { "epoch": 0.48652992499617326, "grad_norm": 0.6808292776818877, "learning_rate": 6.369305737992873e-06, "loss": 1.4925, "step": 6357 }, { "epoch": 0.4866064595132405, "grad_norm": 0.5702953430299759, "learning_rate": 6.368318453732787e-06, "loss": 1.3166, "step": 6358 }, { "epoch": 0.4866829940303077, "grad_norm": 0.6419326592970672, "learning_rate": 6.367331146842114e-06, "loss": 1.4861, "step": 6359 }, { "epoch": 0.48675952854737486, "grad_norm": 0.4917793554019296, "learning_rate": 6.366343817381518e-06, "loss": 0.9282, "step": 6360 }, { "epoch": 0.48683606306444205, "grad_norm": 0.6854052221133822, "learning_rate": 6.365356465411664e-06, "loss": 1.5617, "step": 6361 }, { "epoch": 0.4869125975815093, "grad_norm": 0.5621893997392439, "learning_rate": 6.364369090993215e-06, "loss": 1.2294, "step": 6362 }, { "epoch": 0.48698913209857647, "grad_norm": 0.6290243405407889, "learning_rate": 6.36338169418684e-06, "loss": 1.6634, "step": 6363 }, { "epoch": 0.48706566661564366, "grad_norm": 0.6911398502858531, "learning_rate": 6.362394275053207e-06, "loss": 1.461, "step": 6364 }, { "epoch": 0.48714220113271084, "grad_norm": 0.6979309373830586, "learning_rate": 6.361406833652987e-06, "loss": 1.605, "step": 6365 }, { "epoch": 0.487218735649778, "grad_norm": 2.0165168755973792, "learning_rate": 6.360419370046851e-06, "loss": 1.4399, "step": 6366 }, { "epoch": 0.48729527016684526, "grad_norm": 0.5656948865745031, "learning_rate": 6.35943188429547e-06, "loss": 1.3218, "step": 6367 }, { "epoch": 0.48737180468391245, "grad_norm": 0.5614882450622222, "learning_rate": 6.358444376459521e-06, "loss": 1.3942, "step": 6368 }, { "epoch": 0.48744833920097963, "grad_norm": 0.576150083534331, "learning_rate": 6.357456846599677e-06, "loss": 1.4406, "step": 6369 }, { "epoch": 0.4875248737180468, "grad_norm": 0.6439317123269032, "learning_rate": 6.356469294776617e-06, "loss": 1.4833, "step": 6370 }, { "epoch": 0.48760140823511405, "grad_norm": 0.6128130152553182, "learning_rate": 6.355481721051016e-06, "loss": 1.4176, "step": 6371 }, { "epoch": 0.48767794275218124, "grad_norm": 0.5586795205046193, "learning_rate": 6.354494125483558e-06, "loss": 0.9867, "step": 6372 }, { "epoch": 0.4877544772692484, "grad_norm": 0.6039420721771203, "learning_rate": 6.353506508134919e-06, "loss": 1.3154, "step": 6373 }, { "epoch": 0.4878310117863156, "grad_norm": 0.5381353978022568, "learning_rate": 6.352518869065784e-06, "loss": 1.2157, "step": 6374 }, { "epoch": 0.48790754630338284, "grad_norm": 0.5834500433649773, "learning_rate": 6.351531208336836e-06, "loss": 1.2789, "step": 6375 }, { "epoch": 0.48798408082045003, "grad_norm": 0.5741021923594108, "learning_rate": 6.350543526008761e-06, "loss": 1.4479, "step": 6376 }, { "epoch": 0.4880606153375172, "grad_norm": 0.5566414493405082, "learning_rate": 6.3495558221422424e-06, "loss": 1.3246, "step": 6377 }, { "epoch": 0.4881371498545844, "grad_norm": 0.667759575331685, "learning_rate": 6.348568096797969e-06, "loss": 1.7734, "step": 6378 }, { "epoch": 0.48821368437165164, "grad_norm": 0.5589316461713681, "learning_rate": 6.347580350036629e-06, "loss": 1.1813, "step": 6379 }, { "epoch": 0.4882902188887188, "grad_norm": 0.6435649096373797, "learning_rate": 6.346592581918914e-06, "loss": 1.4818, "step": 6380 }, { "epoch": 0.488366753405786, "grad_norm": 0.7132663438467918, "learning_rate": 6.3456047925055155e-06, "loss": 1.4477, "step": 6381 }, { "epoch": 0.4884432879228532, "grad_norm": 0.5251774995304156, "learning_rate": 6.344616981857124e-06, "loss": 1.085, "step": 6382 }, { "epoch": 0.4885198224399204, "grad_norm": 0.5672829854464702, "learning_rate": 6.343629150034436e-06, "loss": 1.397, "step": 6383 }, { "epoch": 0.4885963569569876, "grad_norm": 0.5470711334360802, "learning_rate": 6.342641297098145e-06, "loss": 1.3324, "step": 6384 }, { "epoch": 0.4886728914740548, "grad_norm": 0.6409206733190791, "learning_rate": 6.341653423108949e-06, "loss": 1.5937, "step": 6385 }, { "epoch": 0.488749425991122, "grad_norm": 0.6339023925926923, "learning_rate": 6.340665528127545e-06, "loss": 1.4098, "step": 6386 }, { "epoch": 0.4888259605081892, "grad_norm": 0.5396250550893641, "learning_rate": 6.339677612214633e-06, "loss": 1.2803, "step": 6387 }, { "epoch": 0.4889024950252564, "grad_norm": 0.6709807387481236, "learning_rate": 6.3386896754309135e-06, "loss": 1.3491, "step": 6388 }, { "epoch": 0.4889790295423236, "grad_norm": 0.6565744905559003, "learning_rate": 6.337701717837086e-06, "loss": 1.3132, "step": 6389 }, { "epoch": 0.48905556405939077, "grad_norm": 0.5386898624157934, "learning_rate": 6.3367137394938564e-06, "loss": 1.2489, "step": 6390 }, { "epoch": 0.489132098576458, "grad_norm": 0.6139663691104589, "learning_rate": 6.3357257404619285e-06, "loss": 1.5468, "step": 6391 }, { "epoch": 0.4892086330935252, "grad_norm": 0.49990365962866773, "learning_rate": 6.3347377208020065e-06, "loss": 1.1677, "step": 6392 }, { "epoch": 0.4892851676105924, "grad_norm": 0.6355050755819971, "learning_rate": 6.333749680574799e-06, "loss": 1.585, "step": 6393 }, { "epoch": 0.48936170212765956, "grad_norm": 0.625439125383856, "learning_rate": 6.332761619841014e-06, "loss": 1.4644, "step": 6394 }, { "epoch": 0.4894382366447268, "grad_norm": 0.6584854175647853, "learning_rate": 6.331773538661359e-06, "loss": 1.0934, "step": 6395 }, { "epoch": 0.489514771161794, "grad_norm": 0.438707970983748, "learning_rate": 6.330785437096547e-06, "loss": 0.8553, "step": 6396 }, { "epoch": 0.48959130567886117, "grad_norm": 0.5361366784590635, "learning_rate": 6.329797315207288e-06, "loss": 1.0927, "step": 6397 }, { "epoch": 0.48966784019592835, "grad_norm": 0.6076553809001662, "learning_rate": 6.328809173054297e-06, "loss": 1.3753, "step": 6398 }, { "epoch": 0.48974437471299553, "grad_norm": 0.5428620983173797, "learning_rate": 6.327821010698286e-06, "loss": 1.297, "step": 6399 }, { "epoch": 0.4898209092300628, "grad_norm": 0.6138489535939818, "learning_rate": 6.326832828199974e-06, "loss": 1.5569, "step": 6400 }, { "epoch": 0.48989744374712996, "grad_norm": 0.6117519820962065, "learning_rate": 6.325844625620075e-06, "loss": 1.4379, "step": 6401 }, { "epoch": 0.48997397826419714, "grad_norm": 1.2087451276541616, "learning_rate": 6.324856403019309e-06, "loss": 1.3324, "step": 6402 }, { "epoch": 0.4900505127812643, "grad_norm": 0.7980145655725139, "learning_rate": 6.323868160458392e-06, "loss": 1.5587, "step": 6403 }, { "epoch": 0.49012704729833156, "grad_norm": 0.49235301731913195, "learning_rate": 6.322879897998049e-06, "loss": 1.1686, "step": 6404 }, { "epoch": 0.49020358181539875, "grad_norm": 0.9096210946267591, "learning_rate": 6.321891615699e-06, "loss": 1.5161, "step": 6405 }, { "epoch": 0.49028011633246593, "grad_norm": 0.6671701791203543, "learning_rate": 6.3209033136219665e-06, "loss": 1.4414, "step": 6406 }, { "epoch": 0.4903566508495331, "grad_norm": 0.6634313924455839, "learning_rate": 6.319914991827676e-06, "loss": 1.5736, "step": 6407 }, { "epoch": 0.49043318536660035, "grad_norm": 0.5038841551845334, "learning_rate": 6.31892665037685e-06, "loss": 1.1514, "step": 6408 }, { "epoch": 0.49050971988366754, "grad_norm": 0.536754153148871, "learning_rate": 6.317938289330218e-06, "loss": 1.139, "step": 6409 }, { "epoch": 0.4905862544007347, "grad_norm": 0.6794728578635183, "learning_rate": 6.3169499087485076e-06, "loss": 1.3965, "step": 6410 }, { "epoch": 0.4906627889178019, "grad_norm": 1.132228171832059, "learning_rate": 6.315961508692448e-06, "loss": 1.5977, "step": 6411 }, { "epoch": 0.49073932343486915, "grad_norm": 0.5952891723662034, "learning_rate": 6.314973089222767e-06, "loss": 1.4487, "step": 6412 }, { "epoch": 0.49081585795193633, "grad_norm": 0.5491460097419821, "learning_rate": 6.3139846504001985e-06, "loss": 1.118, "step": 6413 }, { "epoch": 0.4908923924690035, "grad_norm": 0.5790095197204737, "learning_rate": 6.312996192285474e-06, "loss": 1.463, "step": 6414 }, { "epoch": 0.4909689269860707, "grad_norm": 0.6556230725669271, "learning_rate": 6.312007714939329e-06, "loss": 1.5354, "step": 6415 }, { "epoch": 0.49104546150313794, "grad_norm": 0.7272863065399309, "learning_rate": 6.311019218422496e-06, "loss": 1.3226, "step": 6416 }, { "epoch": 0.4911219960202051, "grad_norm": 0.5933087863041163, "learning_rate": 6.310030702795713e-06, "loss": 1.422, "step": 6417 }, { "epoch": 0.4911985305372723, "grad_norm": 0.5519088639980219, "learning_rate": 6.309042168119717e-06, "loss": 1.2708, "step": 6418 }, { "epoch": 0.4912750650543395, "grad_norm": 0.5518108177398354, "learning_rate": 6.308053614455245e-06, "loss": 1.313, "step": 6419 }, { "epoch": 0.4913515995714067, "grad_norm": 0.548586645753809, "learning_rate": 6.30706504186304e-06, "loss": 1.2254, "step": 6420 }, { "epoch": 0.4914281340884739, "grad_norm": 0.4981304880981458, "learning_rate": 6.3060764504038395e-06, "loss": 1.1403, "step": 6421 }, { "epoch": 0.4915046686055411, "grad_norm": 0.5732318603288614, "learning_rate": 6.305087840138388e-06, "loss": 1.1954, "step": 6422 }, { "epoch": 0.4915812031226083, "grad_norm": 0.5608069665765403, "learning_rate": 6.304099211127427e-06, "loss": 1.3402, "step": 6423 }, { "epoch": 0.4916577376396755, "grad_norm": 0.8656605127228328, "learning_rate": 6.3031105634317015e-06, "loss": 1.5161, "step": 6424 }, { "epoch": 0.4917342721567427, "grad_norm": 0.7063701890870029, "learning_rate": 6.302121897111957e-06, "loss": 1.54, "step": 6425 }, { "epoch": 0.4918108066738099, "grad_norm": 0.6001556116024914, "learning_rate": 6.301133212228939e-06, "loss": 1.5526, "step": 6426 }, { "epoch": 0.49188734119087707, "grad_norm": 0.5574442304357752, "learning_rate": 6.300144508843397e-06, "loss": 1.3743, "step": 6427 }, { "epoch": 0.4919638757079443, "grad_norm": 0.5637123736079973, "learning_rate": 6.299155787016078e-06, "loss": 1.4727, "step": 6428 }, { "epoch": 0.4920404102250115, "grad_norm": 0.5808049032146668, "learning_rate": 6.298167046807735e-06, "loss": 1.424, "step": 6429 }, { "epoch": 0.4921169447420787, "grad_norm": 0.6686451352706723, "learning_rate": 6.297178288279116e-06, "loss": 1.7228, "step": 6430 }, { "epoch": 0.49219347925914586, "grad_norm": 0.5617461570207753, "learning_rate": 6.296189511490975e-06, "loss": 1.2523, "step": 6431 }, { "epoch": 0.49227001377621304, "grad_norm": 0.584037254032285, "learning_rate": 6.295200716504065e-06, "loss": 1.2919, "step": 6432 }, { "epoch": 0.4923465482932803, "grad_norm": 0.6162794494280314, "learning_rate": 6.294211903379141e-06, "loss": 1.3263, "step": 6433 }, { "epoch": 0.49242308281034747, "grad_norm": 1.0157325352930682, "learning_rate": 6.293223072176957e-06, "loss": 1.3713, "step": 6434 }, { "epoch": 0.49249961732741465, "grad_norm": 0.7830393804852122, "learning_rate": 6.2922342229582724e-06, "loss": 1.9204, "step": 6435 }, { "epoch": 0.49257615184448184, "grad_norm": 2.7296669637085995, "learning_rate": 6.291245355783841e-06, "loss": 1.6738, "step": 6436 }, { "epoch": 0.4926526863615491, "grad_norm": 0.6051657232649758, "learning_rate": 6.290256470714426e-06, "loss": 1.3753, "step": 6437 }, { "epoch": 0.49272922087861626, "grad_norm": 0.5479725636987457, "learning_rate": 6.2892675678107866e-06, "loss": 1.3021, "step": 6438 }, { "epoch": 0.49280575539568344, "grad_norm": 0.5731331323430805, "learning_rate": 6.288278647133681e-06, "loss": 1.3934, "step": 6439 }, { "epoch": 0.4928822899127506, "grad_norm": 0.6131988461719402, "learning_rate": 6.287289708743876e-06, "loss": 1.4818, "step": 6440 }, { "epoch": 0.49295882442981787, "grad_norm": 0.6399396899080091, "learning_rate": 6.286300752702129e-06, "loss": 1.539, "step": 6441 }, { "epoch": 0.49303535894688505, "grad_norm": 0.621464759430897, "learning_rate": 6.2853117790692105e-06, "loss": 1.5211, "step": 6442 }, { "epoch": 0.49311189346395223, "grad_norm": 0.6691041320969972, "learning_rate": 6.284322787905883e-06, "loss": 1.6413, "step": 6443 }, { "epoch": 0.4931884279810194, "grad_norm": 0.5055119450111816, "learning_rate": 6.283333779272914e-06, "loss": 1.3188, "step": 6444 }, { "epoch": 0.49326496249808666, "grad_norm": 0.582089424038105, "learning_rate": 6.28234475323107e-06, "loss": 1.408, "step": 6445 }, { "epoch": 0.49334149701515384, "grad_norm": 0.5413228489321892, "learning_rate": 6.281355709841121e-06, "loss": 1.1834, "step": 6446 }, { "epoch": 0.493418031532221, "grad_norm": 0.5450066377067613, "learning_rate": 6.280366649163834e-06, "loss": 1.3142, "step": 6447 }, { "epoch": 0.4934945660492882, "grad_norm": 0.5104864469097589, "learning_rate": 6.279377571259984e-06, "loss": 1.1293, "step": 6448 }, { "epoch": 0.49357110056635545, "grad_norm": 0.5559229647509601, "learning_rate": 6.27838847619034e-06, "loss": 1.3885, "step": 6449 }, { "epoch": 0.49364763508342263, "grad_norm": 1.0646520995431528, "learning_rate": 6.277399364015675e-06, "loss": 1.7808, "step": 6450 }, { "epoch": 0.4937241696004898, "grad_norm": 0.5868842219161806, "learning_rate": 6.276410234796765e-06, "loss": 1.4441, "step": 6451 }, { "epoch": 0.493800704117557, "grad_norm": 0.6637984759414121, "learning_rate": 6.275421088594384e-06, "loss": 1.4008, "step": 6452 }, { "epoch": 0.49387723863462424, "grad_norm": 0.6199080981307784, "learning_rate": 6.274431925469306e-06, "loss": 1.5268, "step": 6453 }, { "epoch": 0.4939537731516914, "grad_norm": 0.6937269708617142, "learning_rate": 6.2734427454823136e-06, "loss": 1.6447, "step": 6454 }, { "epoch": 0.4940303076687586, "grad_norm": 0.5532094171913822, "learning_rate": 6.272453548694179e-06, "loss": 1.2126, "step": 6455 }, { "epoch": 0.4941068421858258, "grad_norm": 0.6267367135815937, "learning_rate": 6.271464335165685e-06, "loss": 1.519, "step": 6456 }, { "epoch": 0.49418337670289303, "grad_norm": 0.7723793226704257, "learning_rate": 6.2704751049576095e-06, "loss": 1.444, "step": 6457 }, { "epoch": 0.4942599112199602, "grad_norm": 0.6148834990990282, "learning_rate": 6.269485858130735e-06, "loss": 1.5762, "step": 6458 }, { "epoch": 0.4943364457370274, "grad_norm": 0.5767304625723784, "learning_rate": 6.268496594745845e-06, "loss": 1.4162, "step": 6459 }, { "epoch": 0.4944129802540946, "grad_norm": 0.6488293275305752, "learning_rate": 6.267507314863721e-06, "loss": 1.6845, "step": 6460 }, { "epoch": 0.4944895147711618, "grad_norm": 0.6645440140183089, "learning_rate": 6.266518018545147e-06, "loss": 1.6026, "step": 6461 }, { "epoch": 0.494566049288229, "grad_norm": 0.6610916239481716, "learning_rate": 6.2655287058509105e-06, "loss": 1.4189, "step": 6462 }, { "epoch": 0.4946425838052962, "grad_norm": 0.5832066190177919, "learning_rate": 6.264539376841796e-06, "loss": 1.3137, "step": 6463 }, { "epoch": 0.49471911832236337, "grad_norm": 0.5524164775109599, "learning_rate": 6.2635500315785914e-06, "loss": 1.4127, "step": 6464 }, { "epoch": 0.4947956528394306, "grad_norm": 0.5861672436389957, "learning_rate": 6.262560670122085e-06, "loss": 1.4274, "step": 6465 }, { "epoch": 0.4948721873564978, "grad_norm": 0.6304366529986015, "learning_rate": 6.261571292533067e-06, "loss": 1.3985, "step": 6466 }, { "epoch": 0.494948721873565, "grad_norm": 0.6004140327419785, "learning_rate": 6.260581898872325e-06, "loss": 1.4974, "step": 6467 }, { "epoch": 0.49502525639063216, "grad_norm": 0.5546593775093449, "learning_rate": 6.259592489200653e-06, "loss": 1.1804, "step": 6468 }, { "epoch": 0.49510179090769935, "grad_norm": 0.51397588631114, "learning_rate": 6.258603063578843e-06, "loss": 1.1197, "step": 6469 }, { "epoch": 0.4951783254247666, "grad_norm": 0.5681359746987799, "learning_rate": 6.257613622067686e-06, "loss": 1.2532, "step": 6470 }, { "epoch": 0.49525485994183377, "grad_norm": 1.0639061962327276, "learning_rate": 6.25662416472798e-06, "loss": 2.0788, "step": 6471 }, { "epoch": 0.49533139445890095, "grad_norm": 0.6770662409986449, "learning_rate": 6.255634691620518e-06, "loss": 1.7655, "step": 6472 }, { "epoch": 0.49540792897596814, "grad_norm": 0.5943193944137162, "learning_rate": 6.2546452028060955e-06, "loss": 1.4189, "step": 6473 }, { "epoch": 0.4954844634930354, "grad_norm": 0.5480793640023178, "learning_rate": 6.253655698345509e-06, "loss": 1.3824, "step": 6474 }, { "epoch": 0.49556099801010256, "grad_norm": 0.6239413374855534, "learning_rate": 6.252666178299561e-06, "loss": 1.5514, "step": 6475 }, { "epoch": 0.49563753252716974, "grad_norm": 0.5983922167269232, "learning_rate": 6.251676642729047e-06, "loss": 1.4386, "step": 6476 }, { "epoch": 0.49571406704423693, "grad_norm": 0.5975218877121777, "learning_rate": 6.250687091694767e-06, "loss": 1.3966, "step": 6477 }, { "epoch": 0.49579060156130417, "grad_norm": 1.7562049094479937, "learning_rate": 6.249697525257523e-06, "loss": 1.7107, "step": 6478 }, { "epoch": 0.49586713607837135, "grad_norm": 0.5028468615811753, "learning_rate": 6.248707943478116e-06, "loss": 1.0491, "step": 6479 }, { "epoch": 0.49594367059543853, "grad_norm": 0.7600939707756097, "learning_rate": 6.24771834641735e-06, "loss": 1.8297, "step": 6480 }, { "epoch": 0.4960202051125057, "grad_norm": 0.6353316768807237, "learning_rate": 6.2467287341360285e-06, "loss": 1.6245, "step": 6481 }, { "epoch": 0.49609673962957296, "grad_norm": 0.6239441785443602, "learning_rate": 6.2457391066949545e-06, "loss": 1.303, "step": 6482 }, { "epoch": 0.49617327414664014, "grad_norm": 0.585557217730066, "learning_rate": 6.244749464154936e-06, "loss": 1.2963, "step": 6483 }, { "epoch": 0.4962498086637073, "grad_norm": 0.6042594293036015, "learning_rate": 6.24375980657678e-06, "loss": 1.3615, "step": 6484 }, { "epoch": 0.4963263431807745, "grad_norm": 0.6585540717555023, "learning_rate": 6.242770134021291e-06, "loss": 1.4089, "step": 6485 }, { "epoch": 0.49640287769784175, "grad_norm": 0.5041947482334614, "learning_rate": 6.24178044654928e-06, "loss": 0.8931, "step": 6486 }, { "epoch": 0.49647941221490893, "grad_norm": 0.6787965079590055, "learning_rate": 6.240790744221557e-06, "loss": 1.2609, "step": 6487 }, { "epoch": 0.4965559467319761, "grad_norm": 0.6224670593062817, "learning_rate": 6.239801027098929e-06, "loss": 1.4881, "step": 6488 }, { "epoch": 0.4966324812490433, "grad_norm": 0.6841497080673895, "learning_rate": 6.238811295242209e-06, "loss": 1.8418, "step": 6489 }, { "epoch": 0.49670901576611054, "grad_norm": 0.8460863198935936, "learning_rate": 6.23782154871221e-06, "loss": 1.4772, "step": 6490 }, { "epoch": 0.4967855502831777, "grad_norm": 0.5626074361290867, "learning_rate": 6.2368317875697435e-06, "loss": 1.1879, "step": 6491 }, { "epoch": 0.4968620848002449, "grad_norm": 0.542716143166819, "learning_rate": 6.235842011875624e-06, "loss": 1.3004, "step": 6492 }, { "epoch": 0.4969386193173121, "grad_norm": 0.5794724805650779, "learning_rate": 6.234852221690666e-06, "loss": 1.4498, "step": 6493 }, { "epoch": 0.49701515383437933, "grad_norm": 0.521450688920846, "learning_rate": 6.233862417075687e-06, "loss": 1.1309, "step": 6494 }, { "epoch": 0.4970916883514465, "grad_norm": 0.5051182724618103, "learning_rate": 6.232872598091501e-06, "loss": 1.053, "step": 6495 }, { "epoch": 0.4971682228685137, "grad_norm": 0.7568984993789012, "learning_rate": 6.231882764798926e-06, "loss": 1.9256, "step": 6496 }, { "epoch": 0.4972447573855809, "grad_norm": 0.4819626130088549, "learning_rate": 6.230892917258781e-06, "loss": 0.8894, "step": 6497 }, { "epoch": 0.4973212919026481, "grad_norm": 0.5592483751040426, "learning_rate": 6.229903055531885e-06, "loss": 1.3276, "step": 6498 }, { "epoch": 0.4973978264197153, "grad_norm": 0.6373231947453257, "learning_rate": 6.228913179679058e-06, "loss": 1.7147, "step": 6499 }, { "epoch": 0.4974743609367825, "grad_norm": 0.9793993658919192, "learning_rate": 6.227923289761121e-06, "loss": 1.4057, "step": 6500 }, { "epoch": 0.4975508954538497, "grad_norm": 0.4933908534807161, "learning_rate": 6.226933385838895e-06, "loss": 1.131, "step": 6501 }, { "epoch": 0.49762742997091686, "grad_norm": 0.6100110450315973, "learning_rate": 6.225943467973204e-06, "loss": 1.4561, "step": 6502 }, { "epoch": 0.4977039644879841, "grad_norm": 0.6794841116175416, "learning_rate": 6.224953536224872e-06, "loss": 1.7567, "step": 6503 }, { "epoch": 0.4977804990050513, "grad_norm": 0.551929141322729, "learning_rate": 6.22396359065472e-06, "loss": 1.2394, "step": 6504 }, { "epoch": 0.49785703352211846, "grad_norm": 0.5949927933373942, "learning_rate": 6.222973631323577e-06, "loss": 1.3352, "step": 6505 }, { "epoch": 0.49793356803918565, "grad_norm": 0.7422595286149264, "learning_rate": 6.221983658292267e-06, "loss": 1.6487, "step": 6506 }, { "epoch": 0.4980101025562529, "grad_norm": 0.7112895801402893, "learning_rate": 6.220993671621617e-06, "loss": 1.5123, "step": 6507 }, { "epoch": 0.49808663707332007, "grad_norm": 0.5721188679917973, "learning_rate": 6.220003671372456e-06, "loss": 1.3473, "step": 6508 }, { "epoch": 0.49816317159038725, "grad_norm": 0.9035224290147578, "learning_rate": 6.219013657605611e-06, "loss": 1.2595, "step": 6509 }, { "epoch": 0.49823970610745444, "grad_norm": 0.5339125181433872, "learning_rate": 6.218023630381912e-06, "loss": 1.3541, "step": 6510 }, { "epoch": 0.4983162406245217, "grad_norm": 0.6443614061590213, "learning_rate": 6.2170335897621904e-06, "loss": 1.508, "step": 6511 }, { "epoch": 0.49839277514158886, "grad_norm": 0.6157956787657076, "learning_rate": 6.216043535807276e-06, "loss": 1.4456, "step": 6512 }, { "epoch": 0.49846930965865605, "grad_norm": 0.5959382293951166, "learning_rate": 6.215053468578e-06, "loss": 1.3125, "step": 6513 }, { "epoch": 0.49854584417572323, "grad_norm": 0.5462374633578692, "learning_rate": 6.2140633881351976e-06, "loss": 1.1877, "step": 6514 }, { "epoch": 0.49862237869279047, "grad_norm": 0.5415175812543777, "learning_rate": 6.2130732945396995e-06, "loss": 1.367, "step": 6515 }, { "epoch": 0.49869891320985765, "grad_norm": 0.5885809038376847, "learning_rate": 6.212083187852342e-06, "loss": 1.1709, "step": 6516 }, { "epoch": 0.49877544772692484, "grad_norm": 0.5517744834680814, "learning_rate": 6.211093068133959e-06, "loss": 1.4696, "step": 6517 }, { "epoch": 0.498851982243992, "grad_norm": 1.1226746579762006, "learning_rate": 6.210102935445385e-06, "loss": 1.3086, "step": 6518 }, { "epoch": 0.49892851676105926, "grad_norm": 0.6845844969102572, "learning_rate": 6.209112789847462e-06, "loss": 1.3477, "step": 6519 }, { "epoch": 0.49900505127812644, "grad_norm": 0.5317536418720032, "learning_rate": 6.208122631401022e-06, "loss": 1.2926, "step": 6520 }, { "epoch": 0.4990815857951936, "grad_norm": 0.9055066798572273, "learning_rate": 6.207132460166906e-06, "loss": 1.244, "step": 6521 }, { "epoch": 0.4991581203122608, "grad_norm": 0.522838638379174, "learning_rate": 6.206142276205952e-06, "loss": 1.1992, "step": 6522 }, { "epoch": 0.49923465482932805, "grad_norm": 0.6032553990409668, "learning_rate": 6.205152079579001e-06, "loss": 1.3587, "step": 6523 }, { "epoch": 0.49931118934639523, "grad_norm": 0.7163978943345911, "learning_rate": 6.204161870346892e-06, "loss": 1.3522, "step": 6524 }, { "epoch": 0.4993877238634624, "grad_norm": 0.5171184817191888, "learning_rate": 6.203171648570469e-06, "loss": 1.1943, "step": 6525 }, { "epoch": 0.4994642583805296, "grad_norm": 0.480765979034935, "learning_rate": 6.20218141431057e-06, "loss": 0.9711, "step": 6526 }, { "epoch": 0.49954079289759684, "grad_norm": 0.5923855931010303, "learning_rate": 6.201191167628043e-06, "loss": 1.3439, "step": 6527 }, { "epoch": 0.499617327414664, "grad_norm": 0.5507773075616457, "learning_rate": 6.2002009085837275e-06, "loss": 1.1678, "step": 6528 }, { "epoch": 0.4996938619317312, "grad_norm": 0.6204433660491993, "learning_rate": 6.19921063723847e-06, "loss": 1.5199, "step": 6529 }, { "epoch": 0.4997703964487984, "grad_norm": 0.9112128502457527, "learning_rate": 6.198220353653115e-06, "loss": 1.556, "step": 6530 }, { "epoch": 0.49984693096586563, "grad_norm": 0.7654268982754737, "learning_rate": 6.19723005788851e-06, "loss": 1.5966, "step": 6531 }, { "epoch": 0.4999234654829328, "grad_norm": 0.7511948975021416, "learning_rate": 6.196239750005499e-06, "loss": 1.637, "step": 6532 }, { "epoch": 0.5, "grad_norm": 0.5348881532749826, "learning_rate": 6.195249430064934e-06, "loss": 1.1848, "step": 6533 }, { "epoch": 0.5000765345170672, "grad_norm": 0.5697577902210733, "learning_rate": 6.194259098127659e-06, "loss": 1.4003, "step": 6534 }, { "epoch": 0.5001530690341344, "grad_norm": 0.7535187504291407, "learning_rate": 6.193268754254524e-06, "loss": 1.4363, "step": 6535 }, { "epoch": 0.5002296035512016, "grad_norm": 0.5244152599880856, "learning_rate": 6.192278398506378e-06, "loss": 1.1315, "step": 6536 }, { "epoch": 0.5003061380682687, "grad_norm": 0.5727196317942574, "learning_rate": 6.191288030944074e-06, "loss": 1.306, "step": 6537 }, { "epoch": 0.500382672585336, "grad_norm": 0.6014903069669125, "learning_rate": 6.190297651628461e-06, "loss": 1.5399, "step": 6538 }, { "epoch": 0.5004592071024032, "grad_norm": 1.2555895964284858, "learning_rate": 6.1893072606203905e-06, "loss": 1.3107, "step": 6539 }, { "epoch": 0.5005357416194703, "grad_norm": 0.4655815709812628, "learning_rate": 6.188316857980716e-06, "loss": 0.9385, "step": 6540 }, { "epoch": 0.5006122761365376, "grad_norm": 0.5894788093952951, "learning_rate": 6.187326443770293e-06, "loss": 1.2525, "step": 6541 }, { "epoch": 0.5006888106536048, "grad_norm": 0.5478129752040174, "learning_rate": 6.186336018049972e-06, "loss": 1.2875, "step": 6542 }, { "epoch": 0.500765345170672, "grad_norm": 0.6276448000043922, "learning_rate": 6.185345580880608e-06, "loss": 1.2405, "step": 6543 }, { "epoch": 0.5008418796877392, "grad_norm": 0.5111723958104275, "learning_rate": 6.184355132323057e-06, "loss": 1.0772, "step": 6544 }, { "epoch": 0.5009184142048063, "grad_norm": 0.5327115176080556, "learning_rate": 6.183364672438178e-06, "loss": 1.2564, "step": 6545 }, { "epoch": 0.5009949487218736, "grad_norm": 0.5977048233822654, "learning_rate": 6.182374201286825e-06, "loss": 1.3153, "step": 6546 }, { "epoch": 0.5010714832389408, "grad_norm": 0.5158759627365991, "learning_rate": 6.181383718929855e-06, "loss": 1.3196, "step": 6547 }, { "epoch": 0.5011480177560079, "grad_norm": 0.5934378125091014, "learning_rate": 6.1803932254281265e-06, "loss": 1.5366, "step": 6548 }, { "epoch": 0.5012245522730752, "grad_norm": 0.5850790800689372, "learning_rate": 6.1794027208425e-06, "loss": 1.348, "step": 6549 }, { "epoch": 0.5013010867901424, "grad_norm": 0.6381169296082616, "learning_rate": 6.178412205233834e-06, "loss": 1.4217, "step": 6550 }, { "epoch": 0.5013776213072095, "grad_norm": 0.6051256398439716, "learning_rate": 6.177421678662987e-06, "loss": 1.3322, "step": 6551 }, { "epoch": 0.5014541558242768, "grad_norm": 0.5846919201752616, "learning_rate": 6.176431141190824e-06, "loss": 1.3532, "step": 6552 }, { "epoch": 0.5015306903413439, "grad_norm": 0.7590752643303262, "learning_rate": 6.175440592878204e-06, "loss": 1.4394, "step": 6553 }, { "epoch": 0.5016072248584111, "grad_norm": 0.6236423915531367, "learning_rate": 6.174450033785987e-06, "loss": 1.4726, "step": 6554 }, { "epoch": 0.5016837593754784, "grad_norm": 0.48779319606787064, "learning_rate": 6.173459463975041e-06, "loss": 1.1329, "step": 6555 }, { "epoch": 0.5017602938925455, "grad_norm": 0.5461914823888662, "learning_rate": 6.172468883506226e-06, "loss": 1.2908, "step": 6556 }, { "epoch": 0.5018368284096127, "grad_norm": 0.7905478430186069, "learning_rate": 6.1714782924404086e-06, "loss": 1.5638, "step": 6557 }, { "epoch": 0.50191336292668, "grad_norm": 0.6158490670555081, "learning_rate": 6.1704876908384505e-06, "loss": 1.3001, "step": 6558 }, { "epoch": 0.5019898974437471, "grad_norm": 0.6468976699143171, "learning_rate": 6.16949707876122e-06, "loss": 1.4793, "step": 6559 }, { "epoch": 0.5020664319608144, "grad_norm": 0.5532507244433107, "learning_rate": 6.168506456269582e-06, "loss": 1.2793, "step": 6560 }, { "epoch": 0.5021429664778815, "grad_norm": 0.5700148449471195, "learning_rate": 6.1675158234244035e-06, "loss": 1.3153, "step": 6561 }, { "epoch": 0.5022195009949487, "grad_norm": 0.6391573059663124, "learning_rate": 6.166525180286551e-06, "loss": 1.5542, "step": 6562 }, { "epoch": 0.502296035512016, "grad_norm": 0.765079075202618, "learning_rate": 6.165534526916894e-06, "loss": 1.3279, "step": 6563 }, { "epoch": 0.5023725700290831, "grad_norm": 0.58943668469261, "learning_rate": 6.1645438633763024e-06, "loss": 1.3366, "step": 6564 }, { "epoch": 0.5024491045461503, "grad_norm": 0.5514671114062304, "learning_rate": 6.163553189725643e-06, "loss": 1.2075, "step": 6565 }, { "epoch": 0.5025256390632176, "grad_norm": 0.532827306766699, "learning_rate": 6.1625625060257865e-06, "loss": 1.1832, "step": 6566 }, { "epoch": 0.5026021735802847, "grad_norm": 0.5609043876395047, "learning_rate": 6.161571812337605e-06, "loss": 1.3312, "step": 6567 }, { "epoch": 0.5026787080973519, "grad_norm": 0.7136132638218791, "learning_rate": 6.160581108721967e-06, "loss": 1.1847, "step": 6568 }, { "epoch": 0.5027552426144191, "grad_norm": 0.5779292097816333, "learning_rate": 6.159590395239746e-06, "loss": 1.3088, "step": 6569 }, { "epoch": 0.5028317771314863, "grad_norm": 0.520422625029374, "learning_rate": 6.158599671951813e-06, "loss": 1.1572, "step": 6570 }, { "epoch": 0.5029083116485535, "grad_norm": 0.6430284943602966, "learning_rate": 6.157608938919043e-06, "loss": 1.507, "step": 6571 }, { "epoch": 0.5029848461656207, "grad_norm": 0.5402036887577504, "learning_rate": 6.156618196202309e-06, "loss": 1.2267, "step": 6572 }, { "epoch": 0.5030613806826879, "grad_norm": 0.6093423224558566, "learning_rate": 6.155627443862484e-06, "loss": 1.4416, "step": 6573 }, { "epoch": 0.503137915199755, "grad_norm": 0.5958306653301508, "learning_rate": 6.154636681960444e-06, "loss": 1.337, "step": 6574 }, { "epoch": 0.5032144497168223, "grad_norm": 0.5948124498177633, "learning_rate": 6.1536459105570645e-06, "loss": 1.478, "step": 6575 }, { "epoch": 0.5032909842338895, "grad_norm": 0.5963420671854046, "learning_rate": 6.15265512971322e-06, "loss": 1.4834, "step": 6576 }, { "epoch": 0.5033675187509566, "grad_norm": 0.5786239602229357, "learning_rate": 6.151664339489788e-06, "loss": 1.4813, "step": 6577 }, { "epoch": 0.5034440532680239, "grad_norm": 0.6270110494155507, "learning_rate": 6.150673539947647e-06, "loss": 1.3971, "step": 6578 }, { "epoch": 0.5035205877850911, "grad_norm": 0.5850408707217973, "learning_rate": 6.149682731147672e-06, "loss": 1.4234, "step": 6579 }, { "epoch": 0.5035971223021583, "grad_norm": 0.48693969739942017, "learning_rate": 6.148691913150743e-06, "loss": 1.1147, "step": 6580 }, { "epoch": 0.5036736568192255, "grad_norm": 0.5588953246272436, "learning_rate": 6.147701086017736e-06, "loss": 1.4695, "step": 6581 }, { "epoch": 0.5037501913362926, "grad_norm": 1.2001530190058365, "learning_rate": 6.146710249809536e-06, "loss": 1.7808, "step": 6582 }, { "epoch": 0.5038267258533599, "grad_norm": 0.912070606427879, "learning_rate": 6.145719404587019e-06, "loss": 1.3672, "step": 6583 }, { "epoch": 0.5039032603704271, "grad_norm": 0.5547937996226523, "learning_rate": 6.144728550411065e-06, "loss": 1.2908, "step": 6584 }, { "epoch": 0.5039797948874942, "grad_norm": 0.5244504764044479, "learning_rate": 6.143737687342557e-06, "loss": 1.1957, "step": 6585 }, { "epoch": 0.5040563294045615, "grad_norm": 0.5894088360641663, "learning_rate": 6.142746815442376e-06, "loss": 1.5494, "step": 6586 }, { "epoch": 0.5041328639216287, "grad_norm": 0.6814735373841627, "learning_rate": 6.141755934771403e-06, "loss": 1.5505, "step": 6587 }, { "epoch": 0.5042093984386958, "grad_norm": 0.9584758969841821, "learning_rate": 6.1407650453905236e-06, "loss": 1.5551, "step": 6588 }, { "epoch": 0.5042859329557631, "grad_norm": 0.6096994123158842, "learning_rate": 6.139774147360616e-06, "loss": 1.4803, "step": 6589 }, { "epoch": 0.5043624674728302, "grad_norm": 0.534463374091606, "learning_rate": 6.13878324074257e-06, "loss": 1.1263, "step": 6590 }, { "epoch": 0.5044390019898974, "grad_norm": 0.5055685618666045, "learning_rate": 6.137792325597265e-06, "loss": 1.2069, "step": 6591 }, { "epoch": 0.5045155365069647, "grad_norm": 0.5818637278653357, "learning_rate": 6.136801401985588e-06, "loss": 1.194, "step": 6592 }, { "epoch": 0.5045920710240318, "grad_norm": 0.5581276277501754, "learning_rate": 6.135810469968424e-06, "loss": 1.2039, "step": 6593 }, { "epoch": 0.504668605541099, "grad_norm": 0.5921994033635112, "learning_rate": 6.1348195296066595e-06, "loss": 1.4364, "step": 6594 }, { "epoch": 0.5047451400581663, "grad_norm": 0.7571834082713219, "learning_rate": 6.133828580961178e-06, "loss": 1.7534, "step": 6595 }, { "epoch": 0.5048216745752334, "grad_norm": 0.7906788314117786, "learning_rate": 6.13283762409287e-06, "loss": 1.5068, "step": 6596 }, { "epoch": 0.5048982090923007, "grad_norm": 0.6318950286769358, "learning_rate": 6.1318466590626206e-06, "loss": 1.3641, "step": 6597 }, { "epoch": 0.5049747436093678, "grad_norm": 0.5115681036396119, "learning_rate": 6.130855685931319e-06, "loss": 1.2389, "step": 6598 }, { "epoch": 0.505051278126435, "grad_norm": 0.6034871194969839, "learning_rate": 6.1298647047598515e-06, "loss": 1.4968, "step": 6599 }, { "epoch": 0.5051278126435023, "grad_norm": 0.6027822979185262, "learning_rate": 6.128873715609109e-06, "loss": 1.3969, "step": 6600 }, { "epoch": 0.5052043471605694, "grad_norm": 0.6297278350711821, "learning_rate": 6.127882718539981e-06, "loss": 1.5906, "step": 6601 }, { "epoch": 0.5052808816776366, "grad_norm": 0.6259567031093057, "learning_rate": 6.1268917136133565e-06, "loss": 1.4882, "step": 6602 }, { "epoch": 0.5053574161947038, "grad_norm": 0.5779893843279151, "learning_rate": 6.125900700890125e-06, "loss": 1.2466, "step": 6603 }, { "epoch": 0.505433950711771, "grad_norm": 0.5482750210407957, "learning_rate": 6.124909680431178e-06, "loss": 1.1016, "step": 6604 }, { "epoch": 0.5055104852288382, "grad_norm": 0.6827273193096303, "learning_rate": 6.123918652297407e-06, "loss": 1.6687, "step": 6605 }, { "epoch": 0.5055870197459054, "grad_norm": 0.5786315223626604, "learning_rate": 6.122927616549703e-06, "loss": 1.3362, "step": 6606 }, { "epoch": 0.5056635542629726, "grad_norm": 0.9007690602992745, "learning_rate": 6.12193657324896e-06, "loss": 1.3591, "step": 6607 }, { "epoch": 0.5057400887800398, "grad_norm": 1.3138198716743061, "learning_rate": 6.120945522456069e-06, "loss": 1.4062, "step": 6608 }, { "epoch": 0.505816623297107, "grad_norm": 0.6283917015653122, "learning_rate": 6.119954464231923e-06, "loss": 1.4196, "step": 6609 }, { "epoch": 0.5058931578141742, "grad_norm": 0.5315644673424992, "learning_rate": 6.118963398637417e-06, "loss": 1.2375, "step": 6610 }, { "epoch": 0.5059696923312413, "grad_norm": 1.2741799700133525, "learning_rate": 6.1179723257334424e-06, "loss": 1.6141, "step": 6611 }, { "epoch": 0.5060462268483086, "grad_norm": 0.7528829523029056, "learning_rate": 6.116981245580897e-06, "loss": 1.9445, "step": 6612 }, { "epoch": 0.5061227613653758, "grad_norm": 0.7303280234574097, "learning_rate": 6.115990158240675e-06, "loss": 1.3367, "step": 6613 }, { "epoch": 0.506199295882443, "grad_norm": 0.5597772234624355, "learning_rate": 6.114999063773669e-06, "loss": 1.4914, "step": 6614 }, { "epoch": 0.5062758303995102, "grad_norm": 0.5533346678801273, "learning_rate": 6.114007962240778e-06, "loss": 1.3049, "step": 6615 }, { "epoch": 0.5063523649165774, "grad_norm": 0.5842303557048031, "learning_rate": 6.1130168537028965e-06, "loss": 1.3584, "step": 6616 }, { "epoch": 0.5064288994336446, "grad_norm": 0.6239577675601997, "learning_rate": 6.112025738220921e-06, "loss": 1.4598, "step": 6617 }, { "epoch": 0.5065054339507118, "grad_norm": 0.5184752932508198, "learning_rate": 6.11103461585575e-06, "loss": 1.1934, "step": 6618 }, { "epoch": 0.5065819684677789, "grad_norm": 2.9488030555621916, "learning_rate": 6.11004348666828e-06, "loss": 1.9445, "step": 6619 }, { "epoch": 0.5066585029848462, "grad_norm": 0.6358903781526429, "learning_rate": 6.109052350719409e-06, "loss": 1.5157, "step": 6620 }, { "epoch": 0.5067350375019134, "grad_norm": 0.6467819827440077, "learning_rate": 6.108061208070036e-06, "loss": 1.7519, "step": 6621 }, { "epoch": 0.5068115720189805, "grad_norm": 0.67890362896939, "learning_rate": 6.107070058781059e-06, "loss": 1.6931, "step": 6622 }, { "epoch": 0.5068881065360478, "grad_norm": 0.6059117836554732, "learning_rate": 6.1060789029133774e-06, "loss": 1.6081, "step": 6623 }, { "epoch": 0.506964641053115, "grad_norm": 0.7602347749777768, "learning_rate": 6.1050877405278915e-06, "loss": 1.671, "step": 6624 }, { "epoch": 0.5070411755701821, "grad_norm": 0.5954013633199021, "learning_rate": 6.104096571685501e-06, "loss": 1.3868, "step": 6625 }, { "epoch": 0.5071177100872494, "grad_norm": 0.5528136438050846, "learning_rate": 6.103105396447107e-06, "loss": 1.2594, "step": 6626 }, { "epoch": 0.5071942446043165, "grad_norm": 0.6454641716558258, "learning_rate": 6.102114214873607e-06, "loss": 1.6672, "step": 6627 }, { "epoch": 0.5072707791213837, "grad_norm": 0.611465758058366, "learning_rate": 6.101123027025904e-06, "loss": 1.2589, "step": 6628 }, { "epoch": 0.507347313638451, "grad_norm": 0.6693733833716069, "learning_rate": 6.1001318329649015e-06, "loss": 1.4466, "step": 6629 }, { "epoch": 0.5074238481555181, "grad_norm": 0.6059475856035593, "learning_rate": 6.0991406327514976e-06, "loss": 1.4268, "step": 6630 }, { "epoch": 0.5075003826725853, "grad_norm": 0.6587264797014349, "learning_rate": 6.0981494264465994e-06, "loss": 1.5717, "step": 6631 }, { "epoch": 0.5075769171896526, "grad_norm": 0.8132466477403109, "learning_rate": 6.097158214111106e-06, "loss": 1.5129, "step": 6632 }, { "epoch": 0.5076534517067197, "grad_norm": 0.5383468789418546, "learning_rate": 6.096166995805921e-06, "loss": 1.3437, "step": 6633 }, { "epoch": 0.507729986223787, "grad_norm": 0.5368196417230906, "learning_rate": 6.095175771591949e-06, "loss": 1.0939, "step": 6634 }, { "epoch": 0.5078065207408541, "grad_norm": 0.5666232402823768, "learning_rate": 6.094184541530094e-06, "loss": 1.3604, "step": 6635 }, { "epoch": 0.5078830552579213, "grad_norm": 0.6985878165519565, "learning_rate": 6.093193305681258e-06, "loss": 1.5871, "step": 6636 }, { "epoch": 0.5079595897749886, "grad_norm": 0.5222256781084532, "learning_rate": 6.092202064106347e-06, "loss": 1.1662, "step": 6637 }, { "epoch": 0.5080361242920557, "grad_norm": 0.8906720364963353, "learning_rate": 6.091210816866265e-06, "loss": 1.6062, "step": 6638 }, { "epoch": 0.5081126588091229, "grad_norm": 0.6260615395437004, "learning_rate": 6.090219564021918e-06, "loss": 1.4484, "step": 6639 }, { "epoch": 0.5081891933261901, "grad_norm": 0.8182885386998976, "learning_rate": 6.089228305634211e-06, "loss": 1.5197, "step": 6640 }, { "epoch": 0.5082657278432573, "grad_norm": 0.6986675134846437, "learning_rate": 6.088237041764051e-06, "loss": 1.4451, "step": 6641 }, { "epoch": 0.5083422623603245, "grad_norm": 0.5630076132355797, "learning_rate": 6.087245772472343e-06, "loss": 1.2807, "step": 6642 }, { "epoch": 0.5084187968773917, "grad_norm": 0.9023908060116843, "learning_rate": 6.086254497819994e-06, "loss": 1.7562, "step": 6643 }, { "epoch": 0.5084953313944589, "grad_norm": 0.6491658119695171, "learning_rate": 6.0852632178679105e-06, "loss": 1.6946, "step": 6644 }, { "epoch": 0.5085718659115261, "grad_norm": 0.5036979808406, "learning_rate": 6.084271932677e-06, "loss": 1.0158, "step": 6645 }, { "epoch": 0.5086484004285933, "grad_norm": 0.568172862436122, "learning_rate": 6.0832806423081715e-06, "loss": 1.2546, "step": 6646 }, { "epoch": 0.5087249349456605, "grad_norm": 0.6501636697457548, "learning_rate": 6.082289346822329e-06, "loss": 1.6272, "step": 6647 }, { "epoch": 0.5088014694627276, "grad_norm": 1.1351168940564718, "learning_rate": 6.081298046280385e-06, "loss": 1.6058, "step": 6648 }, { "epoch": 0.5088780039797949, "grad_norm": 0.7703678009264266, "learning_rate": 6.080306740743245e-06, "loss": 1.1565, "step": 6649 }, { "epoch": 0.5089545384968621, "grad_norm": 0.5439345680304377, "learning_rate": 6.079315430271819e-06, "loss": 1.1746, "step": 6650 }, { "epoch": 0.5090310730139292, "grad_norm": 0.7099734622793575, "learning_rate": 6.078324114927016e-06, "loss": 1.5957, "step": 6651 }, { "epoch": 0.5091076075309965, "grad_norm": 0.5594161514683984, "learning_rate": 6.077332794769743e-06, "loss": 1.2962, "step": 6652 }, { "epoch": 0.5091841420480637, "grad_norm": 0.5172374838932242, "learning_rate": 6.076341469860916e-06, "loss": 1.0974, "step": 6653 }, { "epoch": 0.5092606765651309, "grad_norm": 0.6242834258004546, "learning_rate": 6.075350140261438e-06, "loss": 1.2799, "step": 6654 }, { "epoch": 0.5093372110821981, "grad_norm": 0.5988902892128238, "learning_rate": 6.0743588060322225e-06, "loss": 1.4726, "step": 6655 }, { "epoch": 0.5094137455992652, "grad_norm": 0.6313361575074843, "learning_rate": 6.07336746723418e-06, "loss": 1.5838, "step": 6656 }, { "epoch": 0.5094902801163325, "grad_norm": 0.6486100273640977, "learning_rate": 6.072376123928221e-06, "loss": 1.3852, "step": 6657 }, { "epoch": 0.5095668146333997, "grad_norm": 0.5122810004593268, "learning_rate": 6.071384776175258e-06, "loss": 1.1878, "step": 6658 }, { "epoch": 0.5096433491504668, "grad_norm": 0.819096773972513, "learning_rate": 6.070393424036198e-06, "loss": 1.4686, "step": 6659 }, { "epoch": 0.5097198836675341, "grad_norm": 0.761251795612864, "learning_rate": 6.069402067571959e-06, "loss": 1.709, "step": 6660 }, { "epoch": 0.5097964181846013, "grad_norm": 1.060789424771458, "learning_rate": 6.0684107068434465e-06, "loss": 1.3549, "step": 6661 }, { "epoch": 0.5098729527016684, "grad_norm": 0.7112870854589309, "learning_rate": 6.067419341911577e-06, "loss": 1.7096, "step": 6662 }, { "epoch": 0.5099494872187357, "grad_norm": 0.8173477654036443, "learning_rate": 6.066427972837261e-06, "loss": 1.5691, "step": 6663 }, { "epoch": 0.5100260217358028, "grad_norm": 0.5774968203497768, "learning_rate": 6.0654365996814125e-06, "loss": 1.3316, "step": 6664 }, { "epoch": 0.51010255625287, "grad_norm": 0.6956712018554594, "learning_rate": 6.064445222504944e-06, "loss": 1.442, "step": 6665 }, { "epoch": 0.5101790907699373, "grad_norm": 0.5372968170811944, "learning_rate": 6.063453841368767e-06, "loss": 1.2511, "step": 6666 }, { "epoch": 0.5102556252870044, "grad_norm": 0.6161342108733896, "learning_rate": 6.062462456333797e-06, "loss": 1.6361, "step": 6667 }, { "epoch": 0.5103321598040717, "grad_norm": 0.7004628224795222, "learning_rate": 6.061471067460948e-06, "loss": 1.8308, "step": 6668 }, { "epoch": 0.5104086943211388, "grad_norm": 0.6000022202746076, "learning_rate": 6.060479674811132e-06, "loss": 1.6319, "step": 6669 }, { "epoch": 0.510485228838206, "grad_norm": 0.5540403684380873, "learning_rate": 6.059488278445263e-06, "loss": 1.1967, "step": 6670 }, { "epoch": 0.5105617633552733, "grad_norm": 0.5965825114503172, "learning_rate": 6.058496878424258e-06, "loss": 1.5907, "step": 6671 }, { "epoch": 0.5106382978723404, "grad_norm": 0.628199977629214, "learning_rate": 6.057505474809029e-06, "loss": 1.4683, "step": 6672 }, { "epoch": 0.5107148323894076, "grad_norm": 0.554910573227594, "learning_rate": 6.0565140676604916e-06, "loss": 1.3747, "step": 6673 }, { "epoch": 0.5107913669064749, "grad_norm": 0.7982897034239975, "learning_rate": 6.055522657039561e-06, "loss": 1.4776, "step": 6674 }, { "epoch": 0.510867901423542, "grad_norm": 0.9461140256540375, "learning_rate": 6.0545312430071525e-06, "loss": 1.8651, "step": 6675 }, { "epoch": 0.5109444359406092, "grad_norm": 1.0328578730382385, "learning_rate": 6.053539825624182e-06, "loss": 1.6943, "step": 6676 }, { "epoch": 0.5110209704576764, "grad_norm": 0.5945596463386106, "learning_rate": 6.0525484049515635e-06, "loss": 1.4578, "step": 6677 }, { "epoch": 0.5110975049747436, "grad_norm": 0.6076246590435431, "learning_rate": 6.051556981050215e-06, "loss": 1.5015, "step": 6678 }, { "epoch": 0.5111740394918108, "grad_norm": 0.5167245757880258, "learning_rate": 6.0505655539810505e-06, "loss": 1.201, "step": 6679 }, { "epoch": 0.511250574008878, "grad_norm": 0.6167735975995975, "learning_rate": 6.049574123804988e-06, "loss": 1.6406, "step": 6680 }, { "epoch": 0.5113271085259452, "grad_norm": 0.6743569632720982, "learning_rate": 6.048582690582942e-06, "loss": 1.6613, "step": 6681 }, { "epoch": 0.5114036430430124, "grad_norm": 0.6355518560873276, "learning_rate": 6.047591254375832e-06, "loss": 1.5742, "step": 6682 }, { "epoch": 0.5114801775600796, "grad_norm": 0.5762394374116705, "learning_rate": 6.046599815244571e-06, "loss": 1.4294, "step": 6683 }, { "epoch": 0.5115567120771468, "grad_norm": 0.5461880233252306, "learning_rate": 6.0456083732500805e-06, "loss": 1.27, "step": 6684 }, { "epoch": 0.5116332465942139, "grad_norm": 1.0566157413290618, "learning_rate": 6.044616928453273e-06, "loss": 1.4126, "step": 6685 }, { "epoch": 0.5117097811112812, "grad_norm": 0.5649331090544735, "learning_rate": 6.0436254809150685e-06, "loss": 1.2869, "step": 6686 }, { "epoch": 0.5117863156283484, "grad_norm": 0.6867750406742705, "learning_rate": 6.0426340306963834e-06, "loss": 1.6003, "step": 6687 }, { "epoch": 0.5118628501454155, "grad_norm": 0.5484460744082255, "learning_rate": 6.041642577858138e-06, "loss": 1.3643, "step": 6688 }, { "epoch": 0.5119393846624828, "grad_norm": 0.475997976539207, "learning_rate": 6.040651122461247e-06, "loss": 1.0592, "step": 6689 }, { "epoch": 0.51201591917955, "grad_norm": 0.7112488683012717, "learning_rate": 6.03965966456663e-06, "loss": 1.3918, "step": 6690 }, { "epoch": 0.5120924536966172, "grad_norm": 0.5188647197746986, "learning_rate": 6.038668204235204e-06, "loss": 1.1294, "step": 6691 }, { "epoch": 0.5121689882136844, "grad_norm": 0.647412298917899, "learning_rate": 6.03767674152789e-06, "loss": 1.6263, "step": 6692 }, { "epoch": 0.5122455227307515, "grad_norm": 0.5622518466638811, "learning_rate": 6.0366852765056004e-06, "loss": 1.2918, "step": 6693 }, { "epoch": 0.5123220572478188, "grad_norm": 0.6495336937247019, "learning_rate": 6.035693809229261e-06, "loss": 1.5508, "step": 6694 }, { "epoch": 0.512398591764886, "grad_norm": 0.5593563829776896, "learning_rate": 6.034702339759786e-06, "loss": 1.2747, "step": 6695 }, { "epoch": 0.5124751262819531, "grad_norm": 0.5785191326625858, "learning_rate": 6.033710868158095e-06, "loss": 1.3592, "step": 6696 }, { "epoch": 0.5125516607990204, "grad_norm": 0.6902338021730978, "learning_rate": 6.032719394485108e-06, "loss": 1.535, "step": 6697 }, { "epoch": 0.5126281953160876, "grad_norm": 0.5295774856229847, "learning_rate": 6.031727918801744e-06, "loss": 1.2174, "step": 6698 }, { "epoch": 0.5127047298331547, "grad_norm": 0.5763562945194641, "learning_rate": 6.0307364411689205e-06, "loss": 1.3653, "step": 6699 }, { "epoch": 0.512781264350222, "grad_norm": 1.0476786290236044, "learning_rate": 6.02974496164756e-06, "loss": 1.3824, "step": 6700 }, { "epoch": 0.5128577988672891, "grad_norm": 0.6156561111352901, "learning_rate": 6.0287534802985795e-06, "loss": 1.548, "step": 6701 }, { "epoch": 0.5129343333843563, "grad_norm": 0.6756196800970038, "learning_rate": 6.027761997182899e-06, "loss": 1.7026, "step": 6702 }, { "epoch": 0.5130108679014236, "grad_norm": 0.5188228001222498, "learning_rate": 6.026770512361439e-06, "loss": 1.1414, "step": 6703 }, { "epoch": 0.5130874024184907, "grad_norm": 0.5711805110772084, "learning_rate": 6.025779025895118e-06, "loss": 1.404, "step": 6704 }, { "epoch": 0.513163936935558, "grad_norm": 1.0619234901221117, "learning_rate": 6.024787537844857e-06, "loss": 1.2967, "step": 6705 }, { "epoch": 0.5132404714526251, "grad_norm": 0.5835971627027375, "learning_rate": 6.023796048271575e-06, "loss": 1.3341, "step": 6706 }, { "epoch": 0.5133170059696923, "grad_norm": 0.5506452860342019, "learning_rate": 6.022804557236192e-06, "loss": 1.3429, "step": 6707 }, { "epoch": 0.5133935404867596, "grad_norm": 0.6034618366862827, "learning_rate": 6.021813064799631e-06, "loss": 1.4563, "step": 6708 }, { "epoch": 0.5134700750038267, "grad_norm": 1.6651301802974505, "learning_rate": 6.020821571022809e-06, "loss": 1.1636, "step": 6709 }, { "epoch": 0.5135466095208939, "grad_norm": 0.6564369410774309, "learning_rate": 6.019830075966646e-06, "loss": 1.6099, "step": 6710 }, { "epoch": 0.5136231440379612, "grad_norm": 0.5807800622758515, "learning_rate": 6.018838579692066e-06, "loss": 1.2486, "step": 6711 }, { "epoch": 0.5136996785550283, "grad_norm": 0.5059602856607033, "learning_rate": 6.017847082259986e-06, "loss": 1.1444, "step": 6712 }, { "epoch": 0.5137762130720955, "grad_norm": 0.6244979148341656, "learning_rate": 6.016855583731329e-06, "loss": 1.7379, "step": 6713 }, { "epoch": 0.5138527475891627, "grad_norm": 0.5937333447871068, "learning_rate": 6.015864084167015e-06, "loss": 1.3683, "step": 6714 }, { "epoch": 0.5139292821062299, "grad_norm": 0.5693940705329472, "learning_rate": 6.014872583627962e-06, "loss": 1.3446, "step": 6715 }, { "epoch": 0.5140058166232971, "grad_norm": 0.8572403534420915, "learning_rate": 6.013881082175095e-06, "loss": 1.7856, "step": 6716 }, { "epoch": 0.5140823511403643, "grad_norm": 0.4993514366907099, "learning_rate": 6.012889579869332e-06, "loss": 0.9627, "step": 6717 }, { "epoch": 0.5141588856574315, "grad_norm": 0.5533903794610752, "learning_rate": 6.011898076771594e-06, "loss": 1.3529, "step": 6718 }, { "epoch": 0.5142354201744987, "grad_norm": 0.6198387632157837, "learning_rate": 6.010906572942803e-06, "loss": 1.6452, "step": 6719 }, { "epoch": 0.5143119546915659, "grad_norm": 0.5464905431699421, "learning_rate": 6.0099150684438815e-06, "loss": 1.2765, "step": 6720 }, { "epoch": 0.5143884892086331, "grad_norm": 0.99330725664422, "learning_rate": 6.0089235633357454e-06, "loss": 1.4866, "step": 6721 }, { "epoch": 0.5144650237257002, "grad_norm": 0.6060353388740696, "learning_rate": 6.007932057679321e-06, "loss": 1.3843, "step": 6722 }, { "epoch": 0.5145415582427675, "grad_norm": 0.5463360963678477, "learning_rate": 6.0069405515355275e-06, "loss": 1.2959, "step": 6723 }, { "epoch": 0.5146180927598347, "grad_norm": 0.6412378688263966, "learning_rate": 6.005949044965285e-06, "loss": 1.5794, "step": 6724 }, { "epoch": 0.5146946272769019, "grad_norm": 0.594024651945046, "learning_rate": 6.004957538029514e-06, "loss": 1.5769, "step": 6725 }, { "epoch": 0.5147711617939691, "grad_norm": 0.6416065769054529, "learning_rate": 6.003966030789139e-06, "loss": 1.1234, "step": 6726 }, { "epoch": 0.5148476963110363, "grad_norm": 0.5919717592102143, "learning_rate": 6.002974523305078e-06, "loss": 1.3774, "step": 6727 }, { "epoch": 0.5149242308281035, "grad_norm": 0.7167887456913021, "learning_rate": 6.001983015638256e-06, "loss": 1.4286, "step": 6728 }, { "epoch": 0.5150007653451707, "grad_norm": 0.614195086587616, "learning_rate": 6.000991507849587e-06, "loss": 1.3896, "step": 6729 }, { "epoch": 0.5150772998622378, "grad_norm": 0.5119088462720337, "learning_rate": 6e-06, "loss": 1.1509, "step": 6730 }, { "epoch": 0.5151538343793051, "grad_norm": 0.4411121327515221, "learning_rate": 5.999008492150413e-06, "loss": 0.8732, "step": 6731 }, { "epoch": 0.5152303688963723, "grad_norm": 0.47684593061286235, "learning_rate": 5.998016984361747e-06, "loss": 1.0319, "step": 6732 }, { "epoch": 0.5153069034134394, "grad_norm": 0.6456577362494517, "learning_rate": 5.997025476694922e-06, "loss": 1.5086, "step": 6733 }, { "epoch": 0.5153834379305067, "grad_norm": 0.5107513982744729, "learning_rate": 5.996033969210862e-06, "loss": 1.0411, "step": 6734 }, { "epoch": 0.5154599724475739, "grad_norm": 0.5306620568828012, "learning_rate": 5.995042461970486e-06, "loss": 1.3229, "step": 6735 }, { "epoch": 0.515536506964641, "grad_norm": 0.6430764482337888, "learning_rate": 5.994050955034717e-06, "loss": 1.7269, "step": 6736 }, { "epoch": 0.5156130414817083, "grad_norm": 0.6132701852487406, "learning_rate": 5.993059448464475e-06, "loss": 1.5103, "step": 6737 }, { "epoch": 0.5156895759987754, "grad_norm": 0.5692959077185625, "learning_rate": 5.992067942320682e-06, "loss": 1.4756, "step": 6738 }, { "epoch": 0.5157661105158426, "grad_norm": 0.6331147018532731, "learning_rate": 5.991076436664255e-06, "loss": 1.4327, "step": 6739 }, { "epoch": 0.5158426450329099, "grad_norm": 0.6098663924877914, "learning_rate": 5.9900849315561204e-06, "loss": 1.4959, "step": 6740 }, { "epoch": 0.515919179549977, "grad_norm": 0.654662312862195, "learning_rate": 5.9890934270571975e-06, "loss": 1.721, "step": 6741 }, { "epoch": 0.5159957140670443, "grad_norm": 0.616821658828822, "learning_rate": 5.988101923228407e-06, "loss": 1.4193, "step": 6742 }, { "epoch": 0.5160722485841114, "grad_norm": 0.6019503393965184, "learning_rate": 5.987110420130671e-06, "loss": 1.2879, "step": 6743 }, { "epoch": 0.5161487831011786, "grad_norm": 0.8749613816686819, "learning_rate": 5.986118917824906e-06, "loss": 1.6055, "step": 6744 }, { "epoch": 0.5162253176182459, "grad_norm": 0.6395327851750977, "learning_rate": 5.985127416372039e-06, "loss": 1.5312, "step": 6745 }, { "epoch": 0.516301852135313, "grad_norm": 0.8134737284739637, "learning_rate": 5.984135915832987e-06, "loss": 1.4249, "step": 6746 }, { "epoch": 0.5163783866523802, "grad_norm": 0.7020213682058082, "learning_rate": 5.983144416268673e-06, "loss": 1.9649, "step": 6747 }, { "epoch": 0.5164549211694475, "grad_norm": 0.6057634758363116, "learning_rate": 5.982152917740015e-06, "loss": 1.3629, "step": 6748 }, { "epoch": 0.5165314556865146, "grad_norm": 0.6310748005992388, "learning_rate": 5.981161420307935e-06, "loss": 1.5436, "step": 6749 }, { "epoch": 0.5166079902035818, "grad_norm": 0.5923387498527354, "learning_rate": 5.980169924033354e-06, "loss": 1.2889, "step": 6750 }, { "epoch": 0.516684524720649, "grad_norm": 0.5422932816504142, "learning_rate": 5.979178428977193e-06, "loss": 1.2102, "step": 6751 }, { "epoch": 0.5167610592377162, "grad_norm": 0.5340961391764271, "learning_rate": 5.978186935200371e-06, "loss": 1.3148, "step": 6752 }, { "epoch": 0.5168375937547834, "grad_norm": 0.6744195340528442, "learning_rate": 5.977195442763809e-06, "loss": 1.7453, "step": 6753 }, { "epoch": 0.5169141282718506, "grad_norm": 0.46476433300971626, "learning_rate": 5.976203951728427e-06, "loss": 0.8562, "step": 6754 }, { "epoch": 0.5169906627889178, "grad_norm": 0.5031582667568975, "learning_rate": 5.975212462155145e-06, "loss": 1.1434, "step": 6755 }, { "epoch": 0.517067197305985, "grad_norm": 0.4839300643629946, "learning_rate": 5.9742209741048826e-06, "loss": 1.0246, "step": 6756 }, { "epoch": 0.5171437318230522, "grad_norm": 0.5703867495888323, "learning_rate": 5.973229487638562e-06, "loss": 1.329, "step": 6757 }, { "epoch": 0.5172202663401194, "grad_norm": 0.5553168377395996, "learning_rate": 5.972238002817102e-06, "loss": 1.247, "step": 6758 }, { "epoch": 0.5172968008571865, "grad_norm": 0.5896579762865988, "learning_rate": 5.971246519701422e-06, "loss": 1.4726, "step": 6759 }, { "epoch": 0.5173733353742538, "grad_norm": 0.559758530409408, "learning_rate": 5.970255038352442e-06, "loss": 1.3297, "step": 6760 }, { "epoch": 0.517449869891321, "grad_norm": 0.6436920988560839, "learning_rate": 5.96926355883108e-06, "loss": 1.6389, "step": 6761 }, { "epoch": 0.5175264044083882, "grad_norm": 0.5835095632711987, "learning_rate": 5.968272081198257e-06, "loss": 1.3456, "step": 6762 }, { "epoch": 0.5176029389254554, "grad_norm": 0.5881518420738437, "learning_rate": 5.967280605514894e-06, "loss": 1.386, "step": 6763 }, { "epoch": 0.5176794734425226, "grad_norm": 0.6899266424333469, "learning_rate": 5.966289131841907e-06, "loss": 1.3479, "step": 6764 }, { "epoch": 0.5177560079595898, "grad_norm": 0.7806293436195391, "learning_rate": 5.965297660240217e-06, "loss": 1.734, "step": 6765 }, { "epoch": 0.517832542476657, "grad_norm": 0.5795800890071661, "learning_rate": 5.96430619077074e-06, "loss": 1.4852, "step": 6766 }, { "epoch": 0.5179090769937241, "grad_norm": 0.5238354078985067, "learning_rate": 5.9633147234944e-06, "loss": 1.2003, "step": 6767 }, { "epoch": 0.5179856115107914, "grad_norm": 0.5618717303164402, "learning_rate": 5.962323258472113e-06, "loss": 1.4208, "step": 6768 }, { "epoch": 0.5180621460278586, "grad_norm": 0.5803884051542053, "learning_rate": 5.961331795764797e-06, "loss": 1.263, "step": 6769 }, { "epoch": 0.5181386805449257, "grad_norm": 0.6841381647670004, "learning_rate": 5.960340335433372e-06, "loss": 1.6091, "step": 6770 }, { "epoch": 0.518215215061993, "grad_norm": 0.5209141499943181, "learning_rate": 5.959348877538755e-06, "loss": 0.9936, "step": 6771 }, { "epoch": 0.5182917495790601, "grad_norm": 0.48578572190906244, "learning_rate": 5.958357422141863e-06, "loss": 1.0714, "step": 6772 }, { "epoch": 0.5183682840961273, "grad_norm": 0.6094387174681662, "learning_rate": 5.957365969303617e-06, "loss": 1.4222, "step": 6773 }, { "epoch": 0.5184448186131946, "grad_norm": 0.5342750158832432, "learning_rate": 5.9563745190849335e-06, "loss": 1.0863, "step": 6774 }, { "epoch": 0.5185213531302617, "grad_norm": 0.6370381600788716, "learning_rate": 5.95538307154673e-06, "loss": 1.6677, "step": 6775 }, { "epoch": 0.518597887647329, "grad_norm": 0.6322831896866445, "learning_rate": 5.954391626749922e-06, "loss": 1.5869, "step": 6776 }, { "epoch": 0.5186744221643962, "grad_norm": 0.5974441063297272, "learning_rate": 5.953400184755429e-06, "loss": 1.3356, "step": 6777 }, { "epoch": 0.5187509566814633, "grad_norm": 0.6563493563347266, "learning_rate": 5.95240874562417e-06, "loss": 1.6997, "step": 6778 }, { "epoch": 0.5188274911985306, "grad_norm": 0.5521807189479451, "learning_rate": 5.951417309417059e-06, "loss": 1.3128, "step": 6779 }, { "epoch": 0.5189040257155977, "grad_norm": 0.639853887562239, "learning_rate": 5.950425876195014e-06, "loss": 1.4521, "step": 6780 }, { "epoch": 0.5189805602326649, "grad_norm": 0.6004532407462277, "learning_rate": 5.9494344460189515e-06, "loss": 1.5093, "step": 6781 }, { "epoch": 0.5190570947497322, "grad_norm": 0.6095115378458406, "learning_rate": 5.948443018949788e-06, "loss": 1.6566, "step": 6782 }, { "epoch": 0.5191336292667993, "grad_norm": 0.7634044198312224, "learning_rate": 5.947451595048438e-06, "loss": 1.6736, "step": 6783 }, { "epoch": 0.5192101637838665, "grad_norm": 0.6112326291234577, "learning_rate": 5.9464601743758195e-06, "loss": 1.5286, "step": 6784 }, { "epoch": 0.5192866983009338, "grad_norm": 0.5663902556567858, "learning_rate": 5.9454687569928495e-06, "loss": 1.333, "step": 6785 }, { "epoch": 0.5193632328180009, "grad_norm": 0.5961483685654058, "learning_rate": 5.944477342960441e-06, "loss": 1.5094, "step": 6786 }, { "epoch": 0.5194397673350681, "grad_norm": 0.5153798316131672, "learning_rate": 5.94348593233951e-06, "loss": 1.1557, "step": 6787 }, { "epoch": 0.5195163018521353, "grad_norm": 1.8957220399254735, "learning_rate": 5.942494525190972e-06, "loss": 1.2671, "step": 6788 }, { "epoch": 0.5195928363692025, "grad_norm": 0.5749367686853983, "learning_rate": 5.941503121575743e-06, "loss": 1.4054, "step": 6789 }, { "epoch": 0.5196693708862697, "grad_norm": 0.8960449376613394, "learning_rate": 5.940511721554738e-06, "loss": 1.615, "step": 6790 }, { "epoch": 0.5197459054033369, "grad_norm": 0.7945346369225635, "learning_rate": 5.93952032518887e-06, "loss": 1.6421, "step": 6791 }, { "epoch": 0.5198224399204041, "grad_norm": 0.6531679141417647, "learning_rate": 5.938528932539054e-06, "loss": 1.4245, "step": 6792 }, { "epoch": 0.5198989744374714, "grad_norm": 0.6835654728074121, "learning_rate": 5.937537543666204e-06, "loss": 1.6914, "step": 6793 }, { "epoch": 0.5199755089545385, "grad_norm": 0.621171153352889, "learning_rate": 5.936546158631233e-06, "loss": 1.4837, "step": 6794 }, { "epoch": 0.5200520434716057, "grad_norm": 0.575329974616244, "learning_rate": 5.935554777495059e-06, "loss": 1.3154, "step": 6795 }, { "epoch": 0.5201285779886728, "grad_norm": 0.5312479138828349, "learning_rate": 5.9345634003185895e-06, "loss": 1.1407, "step": 6796 }, { "epoch": 0.5202051125057401, "grad_norm": 0.5640056799506018, "learning_rate": 5.933572027162741e-06, "loss": 1.338, "step": 6797 }, { "epoch": 0.5202816470228073, "grad_norm": 0.47043873377239825, "learning_rate": 5.932580658088425e-06, "loss": 0.9933, "step": 6798 }, { "epoch": 0.5203581815398745, "grad_norm": 0.5644159345874044, "learning_rate": 5.931589293156555e-06, "loss": 1.331, "step": 6799 }, { "epoch": 0.5204347160569417, "grad_norm": 0.9564106714365258, "learning_rate": 5.930597932428042e-06, "loss": 1.5465, "step": 6800 }, { "epoch": 0.5205112505740089, "grad_norm": 0.5912853964312507, "learning_rate": 5.929606575963802e-06, "loss": 1.2705, "step": 6801 }, { "epoch": 0.5205877850910761, "grad_norm": 0.6514988093043651, "learning_rate": 5.928615223824744e-06, "loss": 1.4381, "step": 6802 }, { "epoch": 0.5206643196081433, "grad_norm": 0.610576653622657, "learning_rate": 5.92762387607178e-06, "loss": 1.4827, "step": 6803 }, { "epoch": 0.5207408541252104, "grad_norm": 0.7105917404911741, "learning_rate": 5.926632532765821e-06, "loss": 1.7096, "step": 6804 }, { "epoch": 0.5208173886422777, "grad_norm": 0.6257414711351698, "learning_rate": 5.925641193967779e-06, "loss": 1.6996, "step": 6805 }, { "epoch": 0.5208939231593449, "grad_norm": 0.5740488748893015, "learning_rate": 5.924649859738564e-06, "loss": 1.284, "step": 6806 }, { "epoch": 0.520970457676412, "grad_norm": 0.56371161392355, "learning_rate": 5.923658530139087e-06, "loss": 1.2616, "step": 6807 }, { "epoch": 0.5210469921934793, "grad_norm": 0.5565273316734681, "learning_rate": 5.922667205230258e-06, "loss": 1.1277, "step": 6808 }, { "epoch": 0.5211235267105464, "grad_norm": 0.6865617296060481, "learning_rate": 5.921675885072987e-06, "loss": 1.6053, "step": 6809 }, { "epoch": 0.5212000612276136, "grad_norm": 0.5515964451864136, "learning_rate": 5.9206845697281825e-06, "loss": 1.2731, "step": 6810 }, { "epoch": 0.5212765957446809, "grad_norm": 0.5542762895104598, "learning_rate": 5.919693259256756e-06, "loss": 1.1565, "step": 6811 }, { "epoch": 0.521353130261748, "grad_norm": 0.670511777644024, "learning_rate": 5.918701953719617e-06, "loss": 1.4088, "step": 6812 }, { "epoch": 0.5214296647788152, "grad_norm": 0.5790998672316066, "learning_rate": 5.917710653177672e-06, "loss": 1.1336, "step": 6813 }, { "epoch": 0.5215061992958825, "grad_norm": 0.558703147951219, "learning_rate": 5.916719357691831e-06, "loss": 1.1017, "step": 6814 }, { "epoch": 0.5215827338129496, "grad_norm": 0.5543734378975521, "learning_rate": 5.915728067323001e-06, "loss": 1.2187, "step": 6815 }, { "epoch": 0.5216592683300169, "grad_norm": 0.6233203722834963, "learning_rate": 5.9147367821320915e-06, "loss": 1.5137, "step": 6816 }, { "epoch": 0.521735802847084, "grad_norm": 0.5367491884860132, "learning_rate": 5.913745502180007e-06, "loss": 1.1264, "step": 6817 }, { "epoch": 0.5218123373641512, "grad_norm": 0.6123800562511175, "learning_rate": 5.912754227527659e-06, "loss": 1.5208, "step": 6818 }, { "epoch": 0.5218888718812185, "grad_norm": 0.6133467002395487, "learning_rate": 5.911762958235951e-06, "loss": 1.6262, "step": 6819 }, { "epoch": 0.5219654063982856, "grad_norm": 0.4985748732789269, "learning_rate": 5.9107716943657914e-06, "loss": 0.8702, "step": 6820 }, { "epoch": 0.5220419409153528, "grad_norm": 1.6861645023122303, "learning_rate": 5.909780435978083e-06, "loss": 1.5754, "step": 6821 }, { "epoch": 0.5221184754324201, "grad_norm": 0.7062130078108144, "learning_rate": 5.908789183133737e-06, "loss": 1.6694, "step": 6822 }, { "epoch": 0.5221950099494872, "grad_norm": 0.5587578260114608, "learning_rate": 5.907797935893655e-06, "loss": 1.1901, "step": 6823 }, { "epoch": 0.5222715444665544, "grad_norm": 0.7529713368631333, "learning_rate": 5.906806694318744e-06, "loss": 1.625, "step": 6824 }, { "epoch": 0.5223480789836216, "grad_norm": 0.5205649844252128, "learning_rate": 5.905815458469908e-06, "loss": 1.1979, "step": 6825 }, { "epoch": 0.5224246135006888, "grad_norm": 0.6376565446464687, "learning_rate": 5.904824228408051e-06, "loss": 1.5073, "step": 6826 }, { "epoch": 0.522501148017756, "grad_norm": 0.6184097264919648, "learning_rate": 5.90383300419408e-06, "loss": 1.539, "step": 6827 }, { "epoch": 0.5225776825348232, "grad_norm": 0.5130931661493505, "learning_rate": 5.902841785888896e-06, "loss": 1.0412, "step": 6828 }, { "epoch": 0.5226542170518904, "grad_norm": 0.7075750864475305, "learning_rate": 5.9018505735534026e-06, "loss": 1.5004, "step": 6829 }, { "epoch": 0.5227307515689577, "grad_norm": 0.7707095546217362, "learning_rate": 5.900859367248504e-06, "loss": 1.7141, "step": 6830 }, { "epoch": 0.5228072860860248, "grad_norm": 0.5416576364360592, "learning_rate": 5.899868167035101e-06, "loss": 1.1223, "step": 6831 }, { "epoch": 0.522883820603092, "grad_norm": 0.9188193487654466, "learning_rate": 5.898876972974097e-06, "loss": 1.4881, "step": 6832 }, { "epoch": 0.5229603551201591, "grad_norm": 0.5939998619782286, "learning_rate": 5.8978857851263946e-06, "loss": 1.4324, "step": 6833 }, { "epoch": 0.5230368896372264, "grad_norm": 0.6122475673696441, "learning_rate": 5.896894603552896e-06, "loss": 1.434, "step": 6834 }, { "epoch": 0.5231134241542936, "grad_norm": 0.5812424523367239, "learning_rate": 5.8959034283145e-06, "loss": 1.3523, "step": 6835 }, { "epoch": 0.5231899586713608, "grad_norm": 0.5681080321789789, "learning_rate": 5.89491225947211e-06, "loss": 1.233, "step": 6836 }, { "epoch": 0.523266493188428, "grad_norm": 0.6110028034536089, "learning_rate": 5.893921097086624e-06, "loss": 1.4048, "step": 6837 }, { "epoch": 0.5233430277054951, "grad_norm": 0.5793741640354275, "learning_rate": 5.892929941218942e-06, "loss": 1.4316, "step": 6838 }, { "epoch": 0.5234195622225624, "grad_norm": 0.5735397364464813, "learning_rate": 5.891938791929965e-06, "loss": 1.4015, "step": 6839 }, { "epoch": 0.5234960967396296, "grad_norm": 0.6084133841677347, "learning_rate": 5.8909476492805925e-06, "loss": 1.2403, "step": 6840 }, { "epoch": 0.5235726312566967, "grad_norm": 0.6408948741047881, "learning_rate": 5.889956513331722e-06, "loss": 1.3771, "step": 6841 }, { "epoch": 0.523649165773764, "grad_norm": 0.5796459360795375, "learning_rate": 5.888965384144253e-06, "loss": 1.3129, "step": 6842 }, { "epoch": 0.5237257002908312, "grad_norm": 0.5744505270636197, "learning_rate": 5.8879742617790805e-06, "loss": 1.2975, "step": 6843 }, { "epoch": 0.5238022348078983, "grad_norm": 0.5781839202879583, "learning_rate": 5.886983146297105e-06, "loss": 1.3521, "step": 6844 }, { "epoch": 0.5238787693249656, "grad_norm": 0.571410481744961, "learning_rate": 5.885992037759224e-06, "loss": 1.2399, "step": 6845 }, { "epoch": 0.5239553038420327, "grad_norm": 0.4744080519886747, "learning_rate": 5.885000936226333e-06, "loss": 0.8973, "step": 6846 }, { "epoch": 0.5240318383590999, "grad_norm": 0.4874723277329196, "learning_rate": 5.884009841759327e-06, "loss": 0.8693, "step": 6847 }, { "epoch": 0.5241083728761672, "grad_norm": 0.6509186806090231, "learning_rate": 5.883018754419105e-06, "loss": 1.6128, "step": 6848 }, { "epoch": 0.5241849073932343, "grad_norm": 0.5346428401538545, "learning_rate": 5.882027674266559e-06, "loss": 1.2522, "step": 6849 }, { "epoch": 0.5242614419103016, "grad_norm": 0.595025516337036, "learning_rate": 5.881036601362585e-06, "loss": 1.4653, "step": 6850 }, { "epoch": 0.5243379764273688, "grad_norm": 0.6579097116512517, "learning_rate": 5.880045535768079e-06, "loss": 1.6707, "step": 6851 }, { "epoch": 0.5244145109444359, "grad_norm": 0.7575081088872548, "learning_rate": 5.8790544775439325e-06, "loss": 1.5528, "step": 6852 }, { "epoch": 0.5244910454615032, "grad_norm": 0.5984933129071256, "learning_rate": 5.878063426751043e-06, "loss": 1.4266, "step": 6853 }, { "epoch": 0.5245675799785703, "grad_norm": 0.5438124995674274, "learning_rate": 5.8770723834502974e-06, "loss": 1.2441, "step": 6854 }, { "epoch": 0.5246441144956375, "grad_norm": 0.9808823815229314, "learning_rate": 5.876081347702595e-06, "loss": 1.3989, "step": 6855 }, { "epoch": 0.5247206490127048, "grad_norm": 0.6683094706041303, "learning_rate": 5.875090319568823e-06, "loss": 1.6093, "step": 6856 }, { "epoch": 0.5247971835297719, "grad_norm": 0.6040906558627831, "learning_rate": 5.874099299109877e-06, "loss": 1.3719, "step": 6857 }, { "epoch": 0.5248737180468391, "grad_norm": 0.6264934770798883, "learning_rate": 5.873108286386646e-06, "loss": 1.3743, "step": 6858 }, { "epoch": 0.5249502525639064, "grad_norm": 0.4844019758977131, "learning_rate": 5.872117281460021e-06, "loss": 0.8992, "step": 6859 }, { "epoch": 0.5250267870809735, "grad_norm": 0.6094937982902678, "learning_rate": 5.871126284390893e-06, "loss": 1.3875, "step": 6860 }, { "epoch": 0.5251033215980407, "grad_norm": 0.4879104933124871, "learning_rate": 5.8701352952401505e-06, "loss": 1.2183, "step": 6861 }, { "epoch": 0.5251798561151079, "grad_norm": 0.732070357178313, "learning_rate": 5.869144314068683e-06, "loss": 1.1614, "step": 6862 }, { "epoch": 0.5252563906321751, "grad_norm": 0.636196111219455, "learning_rate": 5.8681533409373814e-06, "loss": 1.2877, "step": 6863 }, { "epoch": 0.5253329251492423, "grad_norm": 0.766507678111982, "learning_rate": 5.867162375907132e-06, "loss": 1.6439, "step": 6864 }, { "epoch": 0.5254094596663095, "grad_norm": 0.7817076867372611, "learning_rate": 5.866171419038823e-06, "loss": 1.4468, "step": 6865 }, { "epoch": 0.5254859941833767, "grad_norm": 0.6075541971619319, "learning_rate": 5.865180470393343e-06, "loss": 1.4964, "step": 6866 }, { "epoch": 0.525562528700444, "grad_norm": 0.846825350661443, "learning_rate": 5.864189530031578e-06, "loss": 1.7763, "step": 6867 }, { "epoch": 0.5256390632175111, "grad_norm": 0.8775960169034113, "learning_rate": 5.863198598014413e-06, "loss": 1.5404, "step": 6868 }, { "epoch": 0.5257155977345783, "grad_norm": 0.5575889129710185, "learning_rate": 5.862207674402737e-06, "loss": 1.3296, "step": 6869 }, { "epoch": 0.5257921322516454, "grad_norm": 0.5935317388800078, "learning_rate": 5.861216759257432e-06, "loss": 1.4422, "step": 6870 }, { "epoch": 0.5258686667687127, "grad_norm": 0.9147380849979032, "learning_rate": 5.860225852639385e-06, "loss": 1.3443, "step": 6871 }, { "epoch": 0.5259452012857799, "grad_norm": 0.7217380196620243, "learning_rate": 5.8592349546094784e-06, "loss": 1.6728, "step": 6872 }, { "epoch": 0.5260217358028471, "grad_norm": 0.6768902411062496, "learning_rate": 5.858244065228597e-06, "loss": 1.7173, "step": 6873 }, { "epoch": 0.5260982703199143, "grad_norm": 0.5674124741927532, "learning_rate": 5.857253184557627e-06, "loss": 1.4115, "step": 6874 }, { "epoch": 0.5261748048369814, "grad_norm": 0.5940969942084512, "learning_rate": 5.856262312657445e-06, "loss": 1.3387, "step": 6875 }, { "epoch": 0.5262513393540487, "grad_norm": 0.6052826903276248, "learning_rate": 5.855271449588935e-06, "loss": 1.6058, "step": 6876 }, { "epoch": 0.5263278738711159, "grad_norm": 0.5250228015815055, "learning_rate": 5.854280595412982e-06, "loss": 1.1039, "step": 6877 }, { "epoch": 0.526404408388183, "grad_norm": 0.6726316088683609, "learning_rate": 5.853289750190465e-06, "loss": 1.6618, "step": 6878 }, { "epoch": 0.5264809429052503, "grad_norm": 0.5398466190118228, "learning_rate": 5.852298913982264e-06, "loss": 1.0753, "step": 6879 }, { "epoch": 0.5265574774223175, "grad_norm": 0.6745022863289436, "learning_rate": 5.85130808684926e-06, "loss": 1.6912, "step": 6880 }, { "epoch": 0.5266340119393846, "grad_norm": 0.5546982197737154, "learning_rate": 5.85031726885233e-06, "loss": 1.3409, "step": 6881 }, { "epoch": 0.5267105464564519, "grad_norm": 0.9806921682831755, "learning_rate": 5.849326460052356e-06, "loss": 1.2357, "step": 6882 }, { "epoch": 0.526787080973519, "grad_norm": 0.6358213488503824, "learning_rate": 5.848335660510213e-06, "loss": 1.45, "step": 6883 }, { "epoch": 0.5268636154905862, "grad_norm": 0.7312762681569979, "learning_rate": 5.847344870286782e-06, "loss": 1.4227, "step": 6884 }, { "epoch": 0.5269401500076535, "grad_norm": 0.6447098156908712, "learning_rate": 5.846354089442938e-06, "loss": 1.5524, "step": 6885 }, { "epoch": 0.5270166845247206, "grad_norm": 0.5905200272006705, "learning_rate": 5.845363318039558e-06, "loss": 1.5036, "step": 6886 }, { "epoch": 0.5270932190417879, "grad_norm": 0.5799745063540469, "learning_rate": 5.844372556137516e-06, "loss": 1.4871, "step": 6887 }, { "epoch": 0.5271697535588551, "grad_norm": 0.626403303352524, "learning_rate": 5.8433818037976916e-06, "loss": 1.5551, "step": 6888 }, { "epoch": 0.5272462880759222, "grad_norm": 0.5165147018497811, "learning_rate": 5.842391061080958e-06, "loss": 1.1104, "step": 6889 }, { "epoch": 0.5273228225929895, "grad_norm": 0.5047044121228021, "learning_rate": 5.841400328048188e-06, "loss": 1.0137, "step": 6890 }, { "epoch": 0.5273993571100566, "grad_norm": 0.6052860764137923, "learning_rate": 5.840409604760257e-06, "loss": 1.3718, "step": 6891 }, { "epoch": 0.5274758916271238, "grad_norm": 0.5996771563938619, "learning_rate": 5.839418891278035e-06, "loss": 1.4027, "step": 6892 }, { "epoch": 0.5275524261441911, "grad_norm": 0.5660357153146758, "learning_rate": 5.838428187662397e-06, "loss": 1.2165, "step": 6893 }, { "epoch": 0.5276289606612582, "grad_norm": 0.9093391114117577, "learning_rate": 5.837437493974215e-06, "loss": 1.6329, "step": 6894 }, { "epoch": 0.5277054951783254, "grad_norm": 0.6048761517369649, "learning_rate": 5.836446810274359e-06, "loss": 1.2009, "step": 6895 }, { "epoch": 0.5277820296953927, "grad_norm": 0.5241575167041811, "learning_rate": 5.8354561366236996e-06, "loss": 1.2486, "step": 6896 }, { "epoch": 0.5278585642124598, "grad_norm": 0.5911911719109294, "learning_rate": 5.834465473083108e-06, "loss": 1.1777, "step": 6897 }, { "epoch": 0.527935098729527, "grad_norm": 0.5744703222990462, "learning_rate": 5.8334748197134495e-06, "loss": 1.3662, "step": 6898 }, { "epoch": 0.5280116332465942, "grad_norm": 0.6092073260204721, "learning_rate": 5.832484176575598e-06, "loss": 1.2906, "step": 6899 }, { "epoch": 0.5280881677636614, "grad_norm": 1.427747009872901, "learning_rate": 5.83149354373042e-06, "loss": 1.769, "step": 6900 }, { "epoch": 0.5281647022807286, "grad_norm": 0.7202805002277555, "learning_rate": 5.830502921238782e-06, "loss": 1.4859, "step": 6901 }, { "epoch": 0.5282412367977958, "grad_norm": 0.6738228294774556, "learning_rate": 5.8295123091615515e-06, "loss": 1.5118, "step": 6902 }, { "epoch": 0.528317771314863, "grad_norm": 0.6422872096044294, "learning_rate": 5.828521707559593e-06, "loss": 1.5308, "step": 6903 }, { "epoch": 0.5283943058319303, "grad_norm": 0.6596234074990766, "learning_rate": 5.827531116493776e-06, "loss": 1.6026, "step": 6904 }, { "epoch": 0.5284708403489974, "grad_norm": 0.5078566624352795, "learning_rate": 5.826540536024961e-06, "loss": 0.9991, "step": 6905 }, { "epoch": 0.5285473748660646, "grad_norm": 0.5357125886904623, "learning_rate": 5.825549966214014e-06, "loss": 1.0411, "step": 6906 }, { "epoch": 0.5286239093831318, "grad_norm": 0.5289955500228753, "learning_rate": 5.824559407121799e-06, "loss": 1.088, "step": 6907 }, { "epoch": 0.528700443900199, "grad_norm": 0.5621018309733112, "learning_rate": 5.823568858809179e-06, "loss": 1.4261, "step": 6908 }, { "epoch": 0.5287769784172662, "grad_norm": 0.6140791064294252, "learning_rate": 5.822578321337014e-06, "loss": 1.5152, "step": 6909 }, { "epoch": 0.5288535129343334, "grad_norm": 0.6590501129047366, "learning_rate": 5.821587794766168e-06, "loss": 1.6438, "step": 6910 }, { "epoch": 0.5289300474514006, "grad_norm": 0.5896858223714335, "learning_rate": 5.820597279157501e-06, "loss": 1.4321, "step": 6911 }, { "epoch": 0.5290065819684677, "grad_norm": 0.49986591797090707, "learning_rate": 5.8196067745718755e-06, "loss": 1.0763, "step": 6912 }, { "epoch": 0.529083116485535, "grad_norm": 1.742683329467151, "learning_rate": 5.818616281070148e-06, "loss": 1.2398, "step": 6913 }, { "epoch": 0.5291596510026022, "grad_norm": 0.967142545337757, "learning_rate": 5.817625798713177e-06, "loss": 1.6145, "step": 6914 }, { "epoch": 0.5292361855196693, "grad_norm": 0.5887020631292268, "learning_rate": 5.816635327561824e-06, "loss": 1.4413, "step": 6915 }, { "epoch": 0.5293127200367366, "grad_norm": 0.6043478282374377, "learning_rate": 5.815644867676944e-06, "loss": 1.4353, "step": 6916 }, { "epoch": 0.5293892545538038, "grad_norm": 0.5911877588826912, "learning_rate": 5.814654419119395e-06, "loss": 1.4045, "step": 6917 }, { "epoch": 0.5294657890708709, "grad_norm": 0.5611595269470269, "learning_rate": 5.813663981950031e-06, "loss": 1.4072, "step": 6918 }, { "epoch": 0.5295423235879382, "grad_norm": 0.763680236206781, "learning_rate": 5.81267355622971e-06, "loss": 1.5341, "step": 6919 }, { "epoch": 0.5296188581050053, "grad_norm": 0.5394412980100808, "learning_rate": 5.811683142019284e-06, "loss": 1.2232, "step": 6920 }, { "epoch": 0.5296953926220725, "grad_norm": 0.6388489563786701, "learning_rate": 5.810692739379611e-06, "loss": 1.7438, "step": 6921 }, { "epoch": 0.5297719271391398, "grad_norm": 0.7660452887118653, "learning_rate": 5.809702348371541e-06, "loss": 1.838, "step": 6922 }, { "epoch": 0.5298484616562069, "grad_norm": 0.6401001629256122, "learning_rate": 5.8087119690559275e-06, "loss": 1.5117, "step": 6923 }, { "epoch": 0.5299249961732742, "grad_norm": 0.538954776021446, "learning_rate": 5.807721601493623e-06, "loss": 1.3172, "step": 6924 }, { "epoch": 0.5300015306903414, "grad_norm": 0.4437879056411892, "learning_rate": 5.806731245745478e-06, "loss": 0.9512, "step": 6925 }, { "epoch": 0.5300780652074085, "grad_norm": 0.6610408776506714, "learning_rate": 5.805740901872343e-06, "loss": 1.5998, "step": 6926 }, { "epoch": 0.5301545997244758, "grad_norm": 0.6461449007835798, "learning_rate": 5.804750569935068e-06, "loss": 1.6416, "step": 6927 }, { "epoch": 0.5302311342415429, "grad_norm": 0.6214966753967641, "learning_rate": 5.803760249994501e-06, "loss": 1.5874, "step": 6928 }, { "epoch": 0.5303076687586101, "grad_norm": 0.5515070029048602, "learning_rate": 5.802769942111493e-06, "loss": 1.248, "step": 6929 }, { "epoch": 0.5303842032756774, "grad_norm": 0.5590788638765063, "learning_rate": 5.801779646346886e-06, "loss": 1.1625, "step": 6930 }, { "epoch": 0.5304607377927445, "grad_norm": 0.6006671441439255, "learning_rate": 5.800789362761531e-06, "loss": 1.473, "step": 6931 }, { "epoch": 0.5305372723098117, "grad_norm": 0.6094198358854452, "learning_rate": 5.799799091416274e-06, "loss": 1.5088, "step": 6932 }, { "epoch": 0.530613806826879, "grad_norm": 0.7559317130276115, "learning_rate": 5.798808832371959e-06, "loss": 1.6777, "step": 6933 }, { "epoch": 0.5306903413439461, "grad_norm": 0.661642842026003, "learning_rate": 5.797818585689432e-06, "loss": 1.4572, "step": 6934 }, { "epoch": 0.5307668758610133, "grad_norm": 0.6066177496611636, "learning_rate": 5.796828351429534e-06, "loss": 1.3797, "step": 6935 }, { "epoch": 0.5308434103780805, "grad_norm": 0.600759611825186, "learning_rate": 5.79583812965311e-06, "loss": 1.3431, "step": 6936 }, { "epoch": 0.5309199448951477, "grad_norm": 0.6071516592710047, "learning_rate": 5.794847920421001e-06, "loss": 1.4558, "step": 6937 }, { "epoch": 0.530996479412215, "grad_norm": 0.5391906083267588, "learning_rate": 5.79385772379405e-06, "loss": 1.1801, "step": 6938 }, { "epoch": 0.5310730139292821, "grad_norm": 0.5759347813663568, "learning_rate": 5.792867539833096e-06, "loss": 1.3821, "step": 6939 }, { "epoch": 0.5311495484463493, "grad_norm": 0.633085100747952, "learning_rate": 5.791877368598981e-06, "loss": 1.2797, "step": 6940 }, { "epoch": 0.5312260829634164, "grad_norm": 0.5497803218233187, "learning_rate": 5.79088721015254e-06, "loss": 1.1286, "step": 6941 }, { "epoch": 0.5313026174804837, "grad_norm": 0.6417155065165853, "learning_rate": 5.789897064554615e-06, "loss": 1.3439, "step": 6942 }, { "epoch": 0.5313791519975509, "grad_norm": 0.7146989711435892, "learning_rate": 5.7889069318660425e-06, "loss": 1.5525, "step": 6943 }, { "epoch": 0.531455686514618, "grad_norm": 0.5632389973951927, "learning_rate": 5.78791681214766e-06, "loss": 1.341, "step": 6944 }, { "epoch": 0.5315322210316853, "grad_norm": 0.6270206201655126, "learning_rate": 5.7869267054603025e-06, "loss": 1.6276, "step": 6945 }, { "epoch": 0.5316087555487525, "grad_norm": 0.5465798585140048, "learning_rate": 5.785936611864804e-06, "loss": 1.3268, "step": 6946 }, { "epoch": 0.5316852900658197, "grad_norm": 0.5890973412312872, "learning_rate": 5.784946531422e-06, "loss": 1.524, "step": 6947 }, { "epoch": 0.5317618245828869, "grad_norm": 0.5936654828283648, "learning_rate": 5.783956464192726e-06, "loss": 1.4788, "step": 6948 }, { "epoch": 0.531838359099954, "grad_norm": 0.5766524918703411, "learning_rate": 5.7829664102378116e-06, "loss": 1.2943, "step": 6949 }, { "epoch": 0.5319148936170213, "grad_norm": 0.5888437491065756, "learning_rate": 5.78197636961809e-06, "loss": 1.2003, "step": 6950 }, { "epoch": 0.5319914281340885, "grad_norm": 0.6395406630811302, "learning_rate": 5.780986342394391e-06, "loss": 1.1607, "step": 6951 }, { "epoch": 0.5320679626511556, "grad_norm": 0.5590734134055644, "learning_rate": 5.779996328627547e-06, "loss": 1.3589, "step": 6952 }, { "epoch": 0.5321444971682229, "grad_norm": 0.6982864473530588, "learning_rate": 5.779006328378383e-06, "loss": 1.1991, "step": 6953 }, { "epoch": 0.5322210316852901, "grad_norm": 0.5683784329689401, "learning_rate": 5.778016341707735e-06, "loss": 1.204, "step": 6954 }, { "epoch": 0.5322975662023572, "grad_norm": 0.8356712023104723, "learning_rate": 5.777026368676425e-06, "loss": 1.9414, "step": 6955 }, { "epoch": 0.5323741007194245, "grad_norm": 0.877044536243372, "learning_rate": 5.7760364093452825e-06, "loss": 1.4832, "step": 6956 }, { "epoch": 0.5324506352364916, "grad_norm": 0.5604393856449233, "learning_rate": 5.775046463775131e-06, "loss": 1.2029, "step": 6957 }, { "epoch": 0.5325271697535588, "grad_norm": 0.6654221166407481, "learning_rate": 5.774056532026796e-06, "loss": 1.2935, "step": 6958 }, { "epoch": 0.5326037042706261, "grad_norm": 0.6073875061186981, "learning_rate": 5.773066614161106e-06, "loss": 1.4934, "step": 6959 }, { "epoch": 0.5326802387876932, "grad_norm": 0.7428932129139479, "learning_rate": 5.772076710238881e-06, "loss": 1.505, "step": 6960 }, { "epoch": 0.5327567733047605, "grad_norm": 0.6479268847017863, "learning_rate": 5.771086820320944e-06, "loss": 1.5567, "step": 6961 }, { "epoch": 0.5328333078218277, "grad_norm": 0.6616421757201864, "learning_rate": 5.770096944468118e-06, "loss": 1.4867, "step": 6962 }, { "epoch": 0.5329098423388948, "grad_norm": 0.8423033351898773, "learning_rate": 5.7691070827412195e-06, "loss": 1.2871, "step": 6963 }, { "epoch": 0.5329863768559621, "grad_norm": 0.6263852350835862, "learning_rate": 5.768117235201075e-06, "loss": 1.3538, "step": 6964 }, { "epoch": 0.5330629113730292, "grad_norm": 0.5576102468362161, "learning_rate": 5.767127401908501e-06, "loss": 1.3967, "step": 6965 }, { "epoch": 0.5331394458900964, "grad_norm": 0.6784940994693188, "learning_rate": 5.766137582924315e-06, "loss": 1.4319, "step": 6966 }, { "epoch": 0.5332159804071637, "grad_norm": 0.4540958989675296, "learning_rate": 5.765147778309334e-06, "loss": 0.9301, "step": 6967 }, { "epoch": 0.5332925149242308, "grad_norm": 0.6699435598420609, "learning_rate": 5.764157988124378e-06, "loss": 1.8578, "step": 6968 }, { "epoch": 0.533369049441298, "grad_norm": 0.4609446077408554, "learning_rate": 5.763168212430257e-06, "loss": 0.9751, "step": 6969 }, { "epoch": 0.5334455839583653, "grad_norm": 0.5532486372186994, "learning_rate": 5.762178451287791e-06, "loss": 1.2879, "step": 6970 }, { "epoch": 0.5335221184754324, "grad_norm": 0.5859609573126141, "learning_rate": 5.761188704757792e-06, "loss": 1.496, "step": 6971 }, { "epoch": 0.5335986529924996, "grad_norm": 0.5590845451935403, "learning_rate": 5.760198972901072e-06, "loss": 1.4304, "step": 6972 }, { "epoch": 0.5336751875095668, "grad_norm": 0.7079734849697958, "learning_rate": 5.759209255778445e-06, "loss": 1.6332, "step": 6973 }, { "epoch": 0.533751722026634, "grad_norm": 0.5064376481428692, "learning_rate": 5.75821955345072e-06, "loss": 1.1111, "step": 6974 }, { "epoch": 0.5338282565437013, "grad_norm": 0.567886910300809, "learning_rate": 5.757229865978709e-06, "loss": 1.2751, "step": 6975 }, { "epoch": 0.5339047910607684, "grad_norm": 0.6041407475426382, "learning_rate": 5.756240193423221e-06, "loss": 1.4887, "step": 6976 }, { "epoch": 0.5339813255778356, "grad_norm": 0.6989872612693677, "learning_rate": 5.755250535845065e-06, "loss": 1.4468, "step": 6977 }, { "epoch": 0.5340578600949027, "grad_norm": 0.5276820288525393, "learning_rate": 5.754260893305047e-06, "loss": 1.1218, "step": 6978 }, { "epoch": 0.53413439461197, "grad_norm": 0.6370508033341293, "learning_rate": 5.753271265863974e-06, "loss": 1.543, "step": 6979 }, { "epoch": 0.5342109291290372, "grad_norm": 0.6692576955963158, "learning_rate": 5.752281653582651e-06, "loss": 1.7342, "step": 6980 }, { "epoch": 0.5342874636461044, "grad_norm": 0.5409253298585911, "learning_rate": 5.751292056521884e-06, "loss": 1.177, "step": 6981 }, { "epoch": 0.5343639981631716, "grad_norm": 0.6569189099630062, "learning_rate": 5.750302474742479e-06, "loss": 1.6033, "step": 6982 }, { "epoch": 0.5344405326802388, "grad_norm": 0.6898044450304419, "learning_rate": 5.749312908305234e-06, "loss": 1.6428, "step": 6983 }, { "epoch": 0.534517067197306, "grad_norm": 0.6849435527999869, "learning_rate": 5.748323357270955e-06, "loss": 1.775, "step": 6984 }, { "epoch": 0.5345936017143732, "grad_norm": 0.8903121401069126, "learning_rate": 5.74733382170044e-06, "loss": 1.7308, "step": 6985 }, { "epoch": 0.5346701362314403, "grad_norm": 0.6842903312019076, "learning_rate": 5.7463443016544905e-06, "loss": 1.691, "step": 6986 }, { "epoch": 0.5347466707485076, "grad_norm": 0.5451530371751946, "learning_rate": 5.7453547971939065e-06, "loss": 1.2397, "step": 6987 }, { "epoch": 0.5348232052655748, "grad_norm": 0.5645626629125001, "learning_rate": 5.744365308379485e-06, "loss": 1.3354, "step": 6988 }, { "epoch": 0.5348997397826419, "grad_norm": 0.5267366079821532, "learning_rate": 5.743375835272022e-06, "loss": 1.2949, "step": 6989 }, { "epoch": 0.5349762742997092, "grad_norm": 0.5266173505949908, "learning_rate": 5.742386377932315e-06, "loss": 1.1704, "step": 6990 }, { "epoch": 0.5350528088167764, "grad_norm": 0.6444258723115612, "learning_rate": 5.741396936421157e-06, "loss": 1.6689, "step": 6991 }, { "epoch": 0.5351293433338435, "grad_norm": 0.872944682517976, "learning_rate": 5.740407510799347e-06, "loss": 1.5857, "step": 6992 }, { "epoch": 0.5352058778509108, "grad_norm": 2.8297849214037716, "learning_rate": 5.739418101127676e-06, "loss": 1.3953, "step": 6993 }, { "epoch": 0.5352824123679779, "grad_norm": 0.590154025065479, "learning_rate": 5.738428707466935e-06, "loss": 1.3511, "step": 6994 }, { "epoch": 0.5353589468850452, "grad_norm": 0.5558040222587528, "learning_rate": 5.737439329877916e-06, "loss": 1.1426, "step": 6995 }, { "epoch": 0.5354354814021124, "grad_norm": 0.5033617363846822, "learning_rate": 5.736449968421409e-06, "loss": 0.9445, "step": 6996 }, { "epoch": 0.5355120159191795, "grad_norm": 0.5485098145746045, "learning_rate": 5.735460623158206e-06, "loss": 1.3456, "step": 6997 }, { "epoch": 0.5355885504362468, "grad_norm": 0.5957154341055804, "learning_rate": 5.734471294149091e-06, "loss": 1.3779, "step": 6998 }, { "epoch": 0.535665084953314, "grad_norm": 0.6976743259311309, "learning_rate": 5.733481981454853e-06, "loss": 1.3688, "step": 6999 }, { "epoch": 0.5357416194703811, "grad_norm": 0.5117030617570815, "learning_rate": 5.732492685136282e-06, "loss": 1.2433, "step": 7000 }, { "epoch": 0.5358181539874484, "grad_norm": 0.47980759674859186, "learning_rate": 5.731503405254158e-06, "loss": 0.8109, "step": 7001 }, { "epoch": 0.5358946885045155, "grad_norm": 0.5934356421418598, "learning_rate": 5.730514141869266e-06, "loss": 1.4723, "step": 7002 }, { "epoch": 0.5359712230215827, "grad_norm": 0.6617844567366904, "learning_rate": 5.729524895042391e-06, "loss": 1.5607, "step": 7003 }, { "epoch": 0.53604775753865, "grad_norm": 0.5752587613699407, "learning_rate": 5.728535664834317e-06, "loss": 1.255, "step": 7004 }, { "epoch": 0.5361242920557171, "grad_norm": 0.5607294470268326, "learning_rate": 5.727546451305824e-06, "loss": 1.2548, "step": 7005 }, { "epoch": 0.5362008265727843, "grad_norm": 1.315496574789146, "learning_rate": 5.7265572545176884e-06, "loss": 1.6647, "step": 7006 }, { "epoch": 0.5362773610898515, "grad_norm": 0.5230713012963902, "learning_rate": 5.725568074530694e-06, "loss": 1.202, "step": 7007 }, { "epoch": 0.5363538956069187, "grad_norm": 0.5958520645531717, "learning_rate": 5.7245789114056174e-06, "loss": 1.4292, "step": 7008 }, { "epoch": 0.536430430123986, "grad_norm": 0.5946705501003455, "learning_rate": 5.723589765203237e-06, "loss": 1.1251, "step": 7009 }, { "epoch": 0.5365069646410531, "grad_norm": 0.6996921045536402, "learning_rate": 5.722600635984326e-06, "loss": 1.4647, "step": 7010 }, { "epoch": 0.5365834991581203, "grad_norm": 0.5344964213827998, "learning_rate": 5.721611523809663e-06, "loss": 1.2836, "step": 7011 }, { "epoch": 0.5366600336751876, "grad_norm": 0.7307539588541854, "learning_rate": 5.720622428740019e-06, "loss": 1.4981, "step": 7012 }, { "epoch": 0.5367365681922547, "grad_norm": 0.4938705578476642, "learning_rate": 5.719633350836167e-06, "loss": 1.1238, "step": 7013 }, { "epoch": 0.5368131027093219, "grad_norm": 0.6182508602871325, "learning_rate": 5.7186442901588815e-06, "loss": 1.5073, "step": 7014 }, { "epoch": 0.536889637226389, "grad_norm": 0.6157619322623625, "learning_rate": 5.717655246768931e-06, "loss": 1.3883, "step": 7015 }, { "epoch": 0.5369661717434563, "grad_norm": 0.7979742354105934, "learning_rate": 5.716666220727088e-06, "loss": 1.4931, "step": 7016 }, { "epoch": 0.5370427062605235, "grad_norm": 0.626420664285597, "learning_rate": 5.715677212094117e-06, "loss": 1.605, "step": 7017 }, { "epoch": 0.5371192407775907, "grad_norm": 0.5446735811235859, "learning_rate": 5.71468822093079e-06, "loss": 1.2469, "step": 7018 }, { "epoch": 0.5371957752946579, "grad_norm": 0.6290935366556548, "learning_rate": 5.7136992472978705e-06, "loss": 1.4444, "step": 7019 }, { "epoch": 0.5372723098117251, "grad_norm": 0.6814549121674031, "learning_rate": 5.712710291256126e-06, "loss": 1.557, "step": 7020 }, { "epoch": 0.5373488443287923, "grad_norm": 0.5708985278593546, "learning_rate": 5.711721352866319e-06, "loss": 1.2837, "step": 7021 }, { "epoch": 0.5374253788458595, "grad_norm": 0.9792426003673766, "learning_rate": 5.7107324321892154e-06, "loss": 1.4675, "step": 7022 }, { "epoch": 0.5375019133629266, "grad_norm": 0.6046716476614383, "learning_rate": 5.709743529285575e-06, "loss": 1.5064, "step": 7023 }, { "epoch": 0.5375784478799939, "grad_norm": 0.682733047968195, "learning_rate": 5.708754644216159e-06, "loss": 1.7371, "step": 7024 }, { "epoch": 0.5376549823970611, "grad_norm": 0.5929527309638291, "learning_rate": 5.7077657770417296e-06, "loss": 1.3207, "step": 7025 }, { "epoch": 0.5377315169141282, "grad_norm": 0.6139617752427511, "learning_rate": 5.706776927823044e-06, "loss": 1.5938, "step": 7026 }, { "epoch": 0.5378080514311955, "grad_norm": 0.532581700581151, "learning_rate": 5.705788096620861e-06, "loss": 1.0864, "step": 7027 }, { "epoch": 0.5378845859482627, "grad_norm": 0.6649710264305604, "learning_rate": 5.704799283495936e-06, "loss": 1.5947, "step": 7028 }, { "epoch": 0.5379611204653298, "grad_norm": 0.5824517821923549, "learning_rate": 5.703810488509026e-06, "loss": 1.4183, "step": 7029 }, { "epoch": 0.5380376549823971, "grad_norm": 0.509347494358759, "learning_rate": 5.702821711720885e-06, "loss": 1.2167, "step": 7030 }, { "epoch": 0.5381141894994642, "grad_norm": 0.6225421928798237, "learning_rate": 5.701832953192267e-06, "loss": 1.5247, "step": 7031 }, { "epoch": 0.5381907240165315, "grad_norm": 1.2901052575244623, "learning_rate": 5.700844212983922e-06, "loss": 1.7973, "step": 7032 }, { "epoch": 0.5382672585335987, "grad_norm": 0.733525283356439, "learning_rate": 5.699855491156605e-06, "loss": 1.4956, "step": 7033 }, { "epoch": 0.5383437930506658, "grad_norm": 0.7082105968554102, "learning_rate": 5.698866787771063e-06, "loss": 1.5348, "step": 7034 }, { "epoch": 0.5384203275677331, "grad_norm": 0.5848649077111174, "learning_rate": 5.697878102888045e-06, "loss": 1.3672, "step": 7035 }, { "epoch": 0.5384968620848003, "grad_norm": 0.616680375675865, "learning_rate": 5.6968894365683005e-06, "loss": 1.5496, "step": 7036 }, { "epoch": 0.5385733966018674, "grad_norm": 0.597365385243766, "learning_rate": 5.695900788872575e-06, "loss": 1.3171, "step": 7037 }, { "epoch": 0.5386499311189347, "grad_norm": 0.7029531374370451, "learning_rate": 5.694912159861614e-06, "loss": 1.3245, "step": 7038 }, { "epoch": 0.5387264656360018, "grad_norm": 0.6802364110793607, "learning_rate": 5.693923549596162e-06, "loss": 1.8781, "step": 7039 }, { "epoch": 0.538803000153069, "grad_norm": 0.635745994907382, "learning_rate": 5.692934958136961e-06, "loss": 1.5713, "step": 7040 }, { "epoch": 0.5388795346701363, "grad_norm": 0.5779256502640375, "learning_rate": 5.691946385544755e-06, "loss": 1.3257, "step": 7041 }, { "epoch": 0.5389560691872034, "grad_norm": 0.5259916848492984, "learning_rate": 5.690957831880285e-06, "loss": 1.3048, "step": 7042 }, { "epoch": 0.5390326037042706, "grad_norm": 0.6053987586912015, "learning_rate": 5.689969297204289e-06, "loss": 1.49, "step": 7043 }, { "epoch": 0.5391091382213378, "grad_norm": 0.608030065654641, "learning_rate": 5.688980781577507e-06, "loss": 1.3064, "step": 7044 }, { "epoch": 0.539185672738405, "grad_norm": 0.5766418815261221, "learning_rate": 5.687992285060674e-06, "loss": 1.4563, "step": 7045 }, { "epoch": 0.5392622072554722, "grad_norm": 0.5230908188072522, "learning_rate": 5.687003807714526e-06, "loss": 1.1282, "step": 7046 }, { "epoch": 0.5393387417725394, "grad_norm": 0.5474828971407939, "learning_rate": 5.686015349599803e-06, "loss": 1.2073, "step": 7047 }, { "epoch": 0.5394152762896066, "grad_norm": 0.5519806305918041, "learning_rate": 5.685026910777233e-06, "loss": 1.3094, "step": 7048 }, { "epoch": 0.5394918108066739, "grad_norm": 0.677828601433083, "learning_rate": 5.684038491307554e-06, "loss": 1.2403, "step": 7049 }, { "epoch": 0.539568345323741, "grad_norm": 0.5410113174447739, "learning_rate": 5.683050091251494e-06, "loss": 1.0672, "step": 7050 }, { "epoch": 0.5396448798408082, "grad_norm": 0.70050103970787, "learning_rate": 5.682061710669783e-06, "loss": 1.6023, "step": 7051 }, { "epoch": 0.5397214143578754, "grad_norm": 0.9910531919373197, "learning_rate": 5.681073349623151e-06, "loss": 1.6792, "step": 7052 }, { "epoch": 0.5397979488749426, "grad_norm": 0.627064849169574, "learning_rate": 5.680085008172326e-06, "loss": 1.3646, "step": 7053 }, { "epoch": 0.5398744833920098, "grad_norm": 0.5370820300632521, "learning_rate": 5.679096686378035e-06, "loss": 1.3004, "step": 7054 }, { "epoch": 0.539951017909077, "grad_norm": 1.1567420808374185, "learning_rate": 5.678108384301002e-06, "loss": 1.4556, "step": 7055 }, { "epoch": 0.5400275524261442, "grad_norm": 0.5940200273533179, "learning_rate": 5.677120102001953e-06, "loss": 1.3158, "step": 7056 }, { "epoch": 0.5401040869432114, "grad_norm": 0.6601386285586043, "learning_rate": 5.676131839541608e-06, "loss": 1.4273, "step": 7057 }, { "epoch": 0.5401806214602786, "grad_norm": 0.5456926926112424, "learning_rate": 5.675143596980693e-06, "loss": 1.2447, "step": 7058 }, { "epoch": 0.5402571559773458, "grad_norm": 0.6479301248970049, "learning_rate": 5.674155374379926e-06, "loss": 1.6088, "step": 7059 }, { "epoch": 0.5403336904944129, "grad_norm": 0.5901482608660462, "learning_rate": 5.673167171800028e-06, "loss": 1.4729, "step": 7060 }, { "epoch": 0.5404102250114802, "grad_norm": 0.6234296119593412, "learning_rate": 5.672178989301715e-06, "loss": 1.4571, "step": 7061 }, { "epoch": 0.5404867595285474, "grad_norm": 0.5288564542925952, "learning_rate": 5.671190826945705e-06, "loss": 1.1916, "step": 7062 }, { "epoch": 0.5405632940456145, "grad_norm": 0.5705843520121942, "learning_rate": 5.670202684792714e-06, "loss": 1.2943, "step": 7063 }, { "epoch": 0.5406398285626818, "grad_norm": 0.5603106301637567, "learning_rate": 5.669214562903455e-06, "loss": 1.3268, "step": 7064 }, { "epoch": 0.540716363079749, "grad_norm": 0.6172343816510798, "learning_rate": 5.6682264613386425e-06, "loss": 1.3307, "step": 7065 }, { "epoch": 0.5407928975968161, "grad_norm": 0.9673277991706357, "learning_rate": 5.667238380158988e-06, "loss": 1.3162, "step": 7066 }, { "epoch": 0.5408694321138834, "grad_norm": 0.63443534396063, "learning_rate": 5.666250319425203e-06, "loss": 1.4044, "step": 7067 }, { "epoch": 0.5409459666309505, "grad_norm": 0.587073929535788, "learning_rate": 5.665262279197994e-06, "loss": 1.388, "step": 7068 }, { "epoch": 0.5410225011480178, "grad_norm": 0.6476837180114519, "learning_rate": 5.664274259538073e-06, "loss": 1.6001, "step": 7069 }, { "epoch": 0.541099035665085, "grad_norm": 0.6280726784277477, "learning_rate": 5.663286260506145e-06, "loss": 1.4888, "step": 7070 }, { "epoch": 0.5411755701821521, "grad_norm": 0.5467332072083249, "learning_rate": 5.662298282162916e-06, "loss": 1.2976, "step": 7071 }, { "epoch": 0.5412521046992194, "grad_norm": 0.6305583419691401, "learning_rate": 5.661310324569089e-06, "loss": 1.0828, "step": 7072 }, { "epoch": 0.5413286392162866, "grad_norm": 0.5033886018704332, "learning_rate": 5.660322387785369e-06, "loss": 1.2702, "step": 7073 }, { "epoch": 0.5414051737333537, "grad_norm": 0.542431336011318, "learning_rate": 5.659334471872456e-06, "loss": 1.0762, "step": 7074 }, { "epoch": 0.541481708250421, "grad_norm": 0.829373929037622, "learning_rate": 5.6583465768910525e-06, "loss": 1.5786, "step": 7075 }, { "epoch": 0.5415582427674881, "grad_norm": 0.5310900513411608, "learning_rate": 5.657358702901856e-06, "loss": 1.2755, "step": 7076 }, { "epoch": 0.5416347772845553, "grad_norm": 0.9893770898989123, "learning_rate": 5.656370849965567e-06, "loss": 1.6598, "step": 7077 }, { "epoch": 0.5417113118016226, "grad_norm": 0.6653713310590934, "learning_rate": 5.6553830181428785e-06, "loss": 1.6326, "step": 7078 }, { "epoch": 0.5417878463186897, "grad_norm": 0.632468824723258, "learning_rate": 5.654395207494485e-06, "loss": 1.5073, "step": 7079 }, { "epoch": 0.5418643808357569, "grad_norm": 0.5677704209284294, "learning_rate": 5.653407418081087e-06, "loss": 1.3408, "step": 7080 }, { "epoch": 0.5419409153528241, "grad_norm": 0.8411747851760824, "learning_rate": 5.652419649963372e-06, "loss": 1.5195, "step": 7081 }, { "epoch": 0.5420174498698913, "grad_norm": 0.6077293840046104, "learning_rate": 5.6514319032020335e-06, "loss": 1.3935, "step": 7082 }, { "epoch": 0.5420939843869585, "grad_norm": 0.5628165460284312, "learning_rate": 5.65044417785776e-06, "loss": 1.3499, "step": 7083 }, { "epoch": 0.5421705189040257, "grad_norm": 0.6157615955585273, "learning_rate": 5.649456473991242e-06, "loss": 1.5279, "step": 7084 }, { "epoch": 0.5422470534210929, "grad_norm": 0.5809141400966895, "learning_rate": 5.648468791663165e-06, "loss": 1.2895, "step": 7085 }, { "epoch": 0.5423235879381602, "grad_norm": 0.5119216540036274, "learning_rate": 5.647481130934217e-06, "loss": 1.1454, "step": 7086 }, { "epoch": 0.5424001224552273, "grad_norm": 0.5495000710692178, "learning_rate": 5.646493491865083e-06, "loss": 1.2198, "step": 7087 }, { "epoch": 0.5424766569722945, "grad_norm": 0.7711811210228215, "learning_rate": 5.645505874516445e-06, "loss": 1.5727, "step": 7088 }, { "epoch": 0.5425531914893617, "grad_norm": 0.6283801337664605, "learning_rate": 5.644518278948985e-06, "loss": 1.3602, "step": 7089 }, { "epoch": 0.5426297260064289, "grad_norm": 0.5968736831733534, "learning_rate": 5.643530705223383e-06, "loss": 1.4332, "step": 7090 }, { "epoch": 0.5427062605234961, "grad_norm": 0.5760834101604034, "learning_rate": 5.642543153400324e-06, "loss": 1.3163, "step": 7091 }, { "epoch": 0.5427827950405633, "grad_norm": 0.4817706052544167, "learning_rate": 5.64155562354048e-06, "loss": 1.0594, "step": 7092 }, { "epoch": 0.5428593295576305, "grad_norm": 0.6706582341472962, "learning_rate": 5.640568115704531e-06, "loss": 1.49, "step": 7093 }, { "epoch": 0.5429358640746977, "grad_norm": 0.6795689281143169, "learning_rate": 5.639580629953151e-06, "loss": 1.6889, "step": 7094 }, { "epoch": 0.5430123985917649, "grad_norm": 0.5991753229741548, "learning_rate": 5.638593166347014e-06, "loss": 1.3888, "step": 7095 }, { "epoch": 0.5430889331088321, "grad_norm": 0.6118205868962139, "learning_rate": 5.637605724946794e-06, "loss": 1.4917, "step": 7096 }, { "epoch": 0.5431654676258992, "grad_norm": 0.6430699587058515, "learning_rate": 5.636618305813162e-06, "loss": 1.3196, "step": 7097 }, { "epoch": 0.5432420021429665, "grad_norm": 0.4588589023571201, "learning_rate": 5.635630909006787e-06, "loss": 1.041, "step": 7098 }, { "epoch": 0.5433185366600337, "grad_norm": 0.5799631657674996, "learning_rate": 5.634643534588339e-06, "loss": 1.4043, "step": 7099 }, { "epoch": 0.5433950711771008, "grad_norm": 0.6730064290638614, "learning_rate": 5.633656182618484e-06, "loss": 1.4256, "step": 7100 }, { "epoch": 0.5434716056941681, "grad_norm": 0.5764125858743907, "learning_rate": 5.632668853157887e-06, "loss": 1.401, "step": 7101 }, { "epoch": 0.5435481402112353, "grad_norm": 0.6060538267013165, "learning_rate": 5.631681546267214e-06, "loss": 1.5709, "step": 7102 }, { "epoch": 0.5436246747283024, "grad_norm": 0.7793630220719239, "learning_rate": 5.630694262007129e-06, "loss": 1.3477, "step": 7103 }, { "epoch": 0.5437012092453697, "grad_norm": 0.6022950692035545, "learning_rate": 5.629707000438293e-06, "loss": 1.4519, "step": 7104 }, { "epoch": 0.5437777437624368, "grad_norm": 0.5035526238663267, "learning_rate": 5.628719761621365e-06, "loss": 1.0533, "step": 7105 }, { "epoch": 0.543854278279504, "grad_norm": 0.5514864093132814, "learning_rate": 5.627732545617004e-06, "loss": 1.1062, "step": 7106 }, { "epoch": 0.5439308127965713, "grad_norm": 0.5205321509871973, "learning_rate": 5.626745352485869e-06, "loss": 1.1086, "step": 7107 }, { "epoch": 0.5440073473136384, "grad_norm": 0.730559390503948, "learning_rate": 5.625758182288617e-06, "loss": 1.8168, "step": 7108 }, { "epoch": 0.5440838818307057, "grad_norm": 0.8748090651227359, "learning_rate": 5.6247710350858984e-06, "loss": 1.4338, "step": 7109 }, { "epoch": 0.5441604163477728, "grad_norm": 0.619343476458322, "learning_rate": 5.623783910938371e-06, "loss": 1.4698, "step": 7110 }, { "epoch": 0.54423695086484, "grad_norm": 0.5428684643787743, "learning_rate": 5.622796809906682e-06, "loss": 1.1392, "step": 7111 }, { "epoch": 0.5443134853819073, "grad_norm": 0.5811826764103748, "learning_rate": 5.621809732051487e-06, "loss": 1.4972, "step": 7112 }, { "epoch": 0.5443900198989744, "grad_norm": 0.5462747571282006, "learning_rate": 5.620822677433433e-06, "loss": 1.3902, "step": 7113 }, { "epoch": 0.5444665544160416, "grad_norm": 0.6363065757793186, "learning_rate": 5.619835646113167e-06, "loss": 1.39, "step": 7114 }, { "epoch": 0.5445430889331089, "grad_norm": 0.8174601242810242, "learning_rate": 5.618848638151334e-06, "loss": 1.4159, "step": 7115 }, { "epoch": 0.544619623450176, "grad_norm": 0.5317490106608408, "learning_rate": 5.617861653608582e-06, "loss": 1.1276, "step": 7116 }, { "epoch": 0.5446961579672432, "grad_norm": 0.5136975921786542, "learning_rate": 5.616874692545551e-06, "loss": 1.1852, "step": 7117 }, { "epoch": 0.5447726924843104, "grad_norm": 0.47338207326755893, "learning_rate": 5.615887755022884e-06, "loss": 0.8316, "step": 7118 }, { "epoch": 0.5448492270013776, "grad_norm": 0.530688032246276, "learning_rate": 5.614900841101223e-06, "loss": 1.1797, "step": 7119 }, { "epoch": 0.5449257615184449, "grad_norm": 0.5714463702238506, "learning_rate": 5.613913950841204e-06, "loss": 1.2842, "step": 7120 }, { "epoch": 0.545002296035512, "grad_norm": 0.5881481341373508, "learning_rate": 5.612927084303468e-06, "loss": 1.3462, "step": 7121 }, { "epoch": 0.5450788305525792, "grad_norm": 0.5564170656021037, "learning_rate": 5.6119402415486454e-06, "loss": 1.325, "step": 7122 }, { "epoch": 0.5451553650696465, "grad_norm": 0.6082567137491751, "learning_rate": 5.610953422637377e-06, "loss": 1.492, "step": 7123 }, { "epoch": 0.5452318995867136, "grad_norm": 0.6818154087189839, "learning_rate": 5.609966627630295e-06, "loss": 1.6631, "step": 7124 }, { "epoch": 0.5453084341037808, "grad_norm": 0.5890891317445247, "learning_rate": 5.608979856588026e-06, "loss": 1.4496, "step": 7125 }, { "epoch": 0.545384968620848, "grad_norm": 0.6088553080821236, "learning_rate": 5.607993109571206e-06, "loss": 1.5556, "step": 7126 }, { "epoch": 0.5454615031379152, "grad_norm": 0.5910823898881153, "learning_rate": 5.607006386640461e-06, "loss": 1.3446, "step": 7127 }, { "epoch": 0.5455380376549824, "grad_norm": 0.5727919922589223, "learning_rate": 5.6060196878564165e-06, "loss": 1.303, "step": 7128 }, { "epoch": 0.5456145721720496, "grad_norm": 0.7432945602731044, "learning_rate": 5.605033013279701e-06, "loss": 1.5123, "step": 7129 }, { "epoch": 0.5456911066891168, "grad_norm": 0.6132135113704719, "learning_rate": 5.60404636297094e-06, "loss": 1.4209, "step": 7130 }, { "epoch": 0.545767641206184, "grad_norm": 0.49857948850318634, "learning_rate": 5.603059736990753e-06, "loss": 1.2077, "step": 7131 }, { "epoch": 0.5458441757232512, "grad_norm": 0.6642411675068077, "learning_rate": 5.602073135399761e-06, "loss": 1.6, "step": 7132 }, { "epoch": 0.5459207102403184, "grad_norm": 0.6520626603695328, "learning_rate": 5.601086558258585e-06, "loss": 1.4914, "step": 7133 }, { "epoch": 0.5459972447573855, "grad_norm": 0.5664938937929274, "learning_rate": 5.600100005627844e-06, "loss": 1.2092, "step": 7134 }, { "epoch": 0.5460737792744528, "grad_norm": 0.5479370739533223, "learning_rate": 5.599113477568154e-06, "loss": 1.2304, "step": 7135 }, { "epoch": 0.54615031379152, "grad_norm": 0.4892141799346639, "learning_rate": 5.598126974140132e-06, "loss": 1.0216, "step": 7136 }, { "epoch": 0.5462268483085871, "grad_norm": 0.5668712495623169, "learning_rate": 5.59714049540439e-06, "loss": 1.1573, "step": 7137 }, { "epoch": 0.5463033828256544, "grad_norm": 0.5391957212299039, "learning_rate": 5.596154041421539e-06, "loss": 1.2223, "step": 7138 }, { "epoch": 0.5463799173427216, "grad_norm": 2.7185508659119435, "learning_rate": 5.5951676122521906e-06, "loss": 1.3783, "step": 7139 }, { "epoch": 0.5464564518597887, "grad_norm": 0.4610989238123267, "learning_rate": 5.594181207956955e-06, "loss": 0.881, "step": 7140 }, { "epoch": 0.546532986376856, "grad_norm": 0.5757364414276328, "learning_rate": 5.593194828596439e-06, "loss": 1.5295, "step": 7141 }, { "epoch": 0.5466095208939231, "grad_norm": 0.61806657851278, "learning_rate": 5.592208474231249e-06, "loss": 1.4683, "step": 7142 }, { "epoch": 0.5466860554109904, "grad_norm": 0.5302697540920303, "learning_rate": 5.591222144921989e-06, "loss": 1.2907, "step": 7143 }, { "epoch": 0.5467625899280576, "grad_norm": 0.5606010628037759, "learning_rate": 5.590235840729261e-06, "loss": 1.3513, "step": 7144 }, { "epoch": 0.5468391244451247, "grad_norm": 0.5901680496514526, "learning_rate": 5.58924956171367e-06, "loss": 1.484, "step": 7145 }, { "epoch": 0.546915658962192, "grad_norm": 0.9570438235793436, "learning_rate": 5.588263307935814e-06, "loss": 1.5832, "step": 7146 }, { "epoch": 0.5469921934792591, "grad_norm": 0.6455343181952353, "learning_rate": 5.587277079456291e-06, "loss": 1.6129, "step": 7147 }, { "epoch": 0.5470687279963263, "grad_norm": 0.858696534908547, "learning_rate": 5.586290876335698e-06, "loss": 1.4669, "step": 7148 }, { "epoch": 0.5471452625133936, "grad_norm": 0.6373498539917473, "learning_rate": 5.5853046986346306e-06, "loss": 1.6329, "step": 7149 }, { "epoch": 0.5472217970304607, "grad_norm": 0.6700487025989454, "learning_rate": 5.584318546413681e-06, "loss": 1.4029, "step": 7150 }, { "epoch": 0.5472983315475279, "grad_norm": 0.5522472452984684, "learning_rate": 5.583332419733444e-06, "loss": 1.2183, "step": 7151 }, { "epoch": 0.5473748660645952, "grad_norm": 0.5998555677647613, "learning_rate": 5.582346318654508e-06, "loss": 1.5657, "step": 7152 }, { "epoch": 0.5474514005816623, "grad_norm": 0.7450232421575255, "learning_rate": 5.581360243237463e-06, "loss": 1.767, "step": 7153 }, { "epoch": 0.5475279350987295, "grad_norm": 0.6859445998177989, "learning_rate": 5.5803741935428954e-06, "loss": 1.6934, "step": 7154 }, { "epoch": 0.5476044696157967, "grad_norm": 0.6949861442220922, "learning_rate": 5.579388169631391e-06, "loss": 1.6069, "step": 7155 }, { "epoch": 0.5476810041328639, "grad_norm": 0.6568915043031937, "learning_rate": 5.578402171563536e-06, "loss": 1.546, "step": 7156 }, { "epoch": 0.5477575386499312, "grad_norm": 1.0895855995799084, "learning_rate": 5.577416199399912e-06, "loss": 1.2765, "step": 7157 }, { "epoch": 0.5478340731669983, "grad_norm": 0.5930206895864475, "learning_rate": 5.576430253201101e-06, "loss": 1.4272, "step": 7158 }, { "epoch": 0.5479106076840655, "grad_norm": 0.5524221698801104, "learning_rate": 5.57544433302768e-06, "loss": 1.3525, "step": 7159 }, { "epoch": 0.5479871422011328, "grad_norm": 0.5687417145622468, "learning_rate": 5.574458438940229e-06, "loss": 1.2816, "step": 7160 }, { "epoch": 0.5480636767181999, "grad_norm": 0.6357059462583586, "learning_rate": 5.573472570999323e-06, "loss": 1.4198, "step": 7161 }, { "epoch": 0.5481402112352671, "grad_norm": 0.5594731382622757, "learning_rate": 5.572486729265538e-06, "loss": 1.2579, "step": 7162 }, { "epoch": 0.5482167457523343, "grad_norm": 0.5068258314847145, "learning_rate": 5.571500913799447e-06, "loss": 1.0226, "step": 7163 }, { "epoch": 0.5482932802694015, "grad_norm": 0.7667329531757863, "learning_rate": 5.57051512466162e-06, "loss": 1.5371, "step": 7164 }, { "epoch": 0.5483698147864687, "grad_norm": 0.6480857329541286, "learning_rate": 5.569529361912628e-06, "loss": 1.1022, "step": 7165 }, { "epoch": 0.5484463493035359, "grad_norm": 0.553085105722291, "learning_rate": 5.568543625613036e-06, "loss": 1.2764, "step": 7166 }, { "epoch": 0.5485228838206031, "grad_norm": 0.6076957870022958, "learning_rate": 5.567557915823416e-06, "loss": 1.2735, "step": 7167 }, { "epoch": 0.5485994183376703, "grad_norm": 1.7726243831965476, "learning_rate": 5.566572232604331e-06, "loss": 1.3599, "step": 7168 }, { "epoch": 0.5486759528547375, "grad_norm": 0.5258038409029587, "learning_rate": 5.5655865760163445e-06, "loss": 1.1298, "step": 7169 }, { "epoch": 0.5487524873718047, "grad_norm": 0.6178546416122507, "learning_rate": 5.564600946120016e-06, "loss": 1.6182, "step": 7170 }, { "epoch": 0.5488290218888718, "grad_norm": 0.6906956030936071, "learning_rate": 5.563615342975907e-06, "loss": 1.702, "step": 7171 }, { "epoch": 0.5489055564059391, "grad_norm": 0.6448923997897167, "learning_rate": 5.562629766644574e-06, "loss": 1.6104, "step": 7172 }, { "epoch": 0.5489820909230063, "grad_norm": 0.6543961905406279, "learning_rate": 5.561644217186579e-06, "loss": 1.3588, "step": 7173 }, { "epoch": 0.5490586254400734, "grad_norm": 0.8417109623303073, "learning_rate": 5.5606586946624715e-06, "loss": 1.6233, "step": 7174 }, { "epoch": 0.5491351599571407, "grad_norm": 0.7348296778256996, "learning_rate": 5.559673199132809e-06, "loss": 1.6362, "step": 7175 }, { "epoch": 0.5492116944742079, "grad_norm": 0.983656120633759, "learning_rate": 5.55868773065814e-06, "loss": 1.8041, "step": 7176 }, { "epoch": 0.549288228991275, "grad_norm": 0.6379925504162541, "learning_rate": 5.557702289299015e-06, "loss": 1.5997, "step": 7177 }, { "epoch": 0.5493647635083423, "grad_norm": 0.6517808066189364, "learning_rate": 5.556716875115985e-06, "loss": 1.4128, "step": 7178 }, { "epoch": 0.5494412980254094, "grad_norm": 0.6186123213535972, "learning_rate": 5.555731488169595e-06, "loss": 1.4455, "step": 7179 }, { "epoch": 0.5495178325424767, "grad_norm": 0.6756500803484268, "learning_rate": 5.554746128520392e-06, "loss": 1.6109, "step": 7180 }, { "epoch": 0.5495943670595439, "grad_norm": 0.48147241401605556, "learning_rate": 5.553760796228917e-06, "loss": 0.9988, "step": 7181 }, { "epoch": 0.549670901576611, "grad_norm": 0.8128559053569989, "learning_rate": 5.552775491355713e-06, "loss": 1.7456, "step": 7182 }, { "epoch": 0.5497474360936783, "grad_norm": 0.6068405481809168, "learning_rate": 5.551790213961317e-06, "loss": 1.4208, "step": 7183 }, { "epoch": 0.5498239706107454, "grad_norm": 0.5693899259473436, "learning_rate": 5.550804964106273e-06, "loss": 1.3039, "step": 7184 }, { "epoch": 0.5499005051278126, "grad_norm": 0.615591889107001, "learning_rate": 5.549819741851115e-06, "loss": 1.4796, "step": 7185 }, { "epoch": 0.5499770396448799, "grad_norm": 0.5868507450009331, "learning_rate": 5.548834547256378e-06, "loss": 1.465, "step": 7186 }, { "epoch": 0.550053574161947, "grad_norm": 0.6339235775108899, "learning_rate": 5.547849380382594e-06, "loss": 1.4333, "step": 7187 }, { "epoch": 0.5501301086790142, "grad_norm": 0.5185601313833822, "learning_rate": 5.546864241290294e-06, "loss": 1.1335, "step": 7188 }, { "epoch": 0.5502066431960815, "grad_norm": 0.6121905449195478, "learning_rate": 5.545879130040013e-06, "loss": 1.4141, "step": 7189 }, { "epoch": 0.5502831777131486, "grad_norm": 0.8556300276409792, "learning_rate": 5.544894046692274e-06, "loss": 1.6647, "step": 7190 }, { "epoch": 0.5503597122302158, "grad_norm": 0.7573006618862769, "learning_rate": 5.543908991307605e-06, "loss": 1.5189, "step": 7191 }, { "epoch": 0.550436246747283, "grad_norm": 0.5449675305284709, "learning_rate": 5.542923963946533e-06, "loss": 1.2061, "step": 7192 }, { "epoch": 0.5505127812643502, "grad_norm": 0.8750856564302413, "learning_rate": 5.5419389646695776e-06, "loss": 1.3025, "step": 7193 }, { "epoch": 0.5505893157814175, "grad_norm": 0.5447396421125505, "learning_rate": 5.5409539935372606e-06, "loss": 1.2377, "step": 7194 }, { "epoch": 0.5506658502984846, "grad_norm": 0.5914243436488816, "learning_rate": 5.539969050610103e-06, "loss": 1.435, "step": 7195 }, { "epoch": 0.5507423848155518, "grad_norm": 0.66422113212769, "learning_rate": 5.538984135948623e-06, "loss": 1.6085, "step": 7196 }, { "epoch": 0.5508189193326191, "grad_norm": 0.5606624469424943, "learning_rate": 5.537999249613336e-06, "loss": 1.2104, "step": 7197 }, { "epoch": 0.5508954538496862, "grad_norm": 0.6291197040617266, "learning_rate": 5.5370143916647564e-06, "loss": 1.6358, "step": 7198 }, { "epoch": 0.5509719883667534, "grad_norm": 0.6779092378522781, "learning_rate": 5.5360295621633955e-06, "loss": 1.6549, "step": 7199 }, { "epoch": 0.5510485228838206, "grad_norm": 0.5959981135321122, "learning_rate": 5.535044761169765e-06, "loss": 1.3935, "step": 7200 }, { "epoch": 0.5511250574008878, "grad_norm": 0.5823339984681287, "learning_rate": 5.534059988744376e-06, "loss": 1.2954, "step": 7201 }, { "epoch": 0.551201591917955, "grad_norm": 0.5332486789491298, "learning_rate": 5.533075244947734e-06, "loss": 1.2046, "step": 7202 }, { "epoch": 0.5512781264350222, "grad_norm": 0.49841578634213646, "learning_rate": 5.5320905298403446e-06, "loss": 1.153, "step": 7203 }, { "epoch": 0.5513546609520894, "grad_norm": 0.9831628625902057, "learning_rate": 5.531105843482712e-06, "loss": 1.8269, "step": 7204 }, { "epoch": 0.5514311954691566, "grad_norm": 0.7798460810058794, "learning_rate": 5.530121185935335e-06, "loss": 1.2706, "step": 7205 }, { "epoch": 0.5515077299862238, "grad_norm": 1.210683268404452, "learning_rate": 5.529136557258721e-06, "loss": 1.2062, "step": 7206 }, { "epoch": 0.551584264503291, "grad_norm": 0.8655856236606579, "learning_rate": 5.528151957513362e-06, "loss": 1.5089, "step": 7207 }, { "epoch": 0.5516607990203581, "grad_norm": 0.6591841691986567, "learning_rate": 5.52716738675976e-06, "loss": 1.465, "step": 7208 }, { "epoch": 0.5517373335374254, "grad_norm": 0.5548837350197222, "learning_rate": 5.5261828450584055e-06, "loss": 1.3116, "step": 7209 }, { "epoch": 0.5518138680544926, "grad_norm": 0.5087813209033489, "learning_rate": 5.525198332469792e-06, "loss": 1.0156, "step": 7210 }, { "epoch": 0.5518904025715597, "grad_norm": 0.5764192307875204, "learning_rate": 5.5242138490544135e-06, "loss": 1.2797, "step": 7211 }, { "epoch": 0.551966937088627, "grad_norm": 0.5108619765090031, "learning_rate": 5.52322939487276e-06, "loss": 1.1728, "step": 7212 }, { "epoch": 0.5520434716056941, "grad_norm": 0.5732458893304754, "learning_rate": 5.522244969985317e-06, "loss": 1.1653, "step": 7213 }, { "epoch": 0.5521200061227614, "grad_norm": 0.6129252328167154, "learning_rate": 5.521260574452572e-06, "loss": 1.3911, "step": 7214 }, { "epoch": 0.5521965406398286, "grad_norm": 0.6452306354777468, "learning_rate": 5.520276208335007e-06, "loss": 1.4508, "step": 7215 }, { "epoch": 0.5522730751568957, "grad_norm": 0.6665476698979015, "learning_rate": 5.519291871693105e-06, "loss": 1.4945, "step": 7216 }, { "epoch": 0.552349609673963, "grad_norm": 0.6991734527645604, "learning_rate": 5.518307564587349e-06, "loss": 1.6249, "step": 7217 }, { "epoch": 0.5524261441910302, "grad_norm": 0.8786285843927848, "learning_rate": 5.517323287078216e-06, "loss": 1.4016, "step": 7218 }, { "epoch": 0.5525026787080973, "grad_norm": 0.5945438943188571, "learning_rate": 5.516339039226183e-06, "loss": 1.304, "step": 7219 }, { "epoch": 0.5525792132251646, "grad_norm": 0.6023791447463209, "learning_rate": 5.515354821091725e-06, "loss": 1.5505, "step": 7220 }, { "epoch": 0.5526557477422317, "grad_norm": 0.6573557037410593, "learning_rate": 5.5143706327353155e-06, "loss": 1.3764, "step": 7221 }, { "epoch": 0.5527322822592989, "grad_norm": 0.5407761391002713, "learning_rate": 5.513386474217427e-06, "loss": 1.2042, "step": 7222 }, { "epoch": 0.5528088167763662, "grad_norm": 0.5907059744944753, "learning_rate": 5.512402345598527e-06, "loss": 1.3466, "step": 7223 }, { "epoch": 0.5528853512934333, "grad_norm": 0.588280162363694, "learning_rate": 5.5114182469390866e-06, "loss": 1.3192, "step": 7224 }, { "epoch": 0.5529618858105005, "grad_norm": 0.8531301477164407, "learning_rate": 5.510434178299568e-06, "loss": 1.5893, "step": 7225 }, { "epoch": 0.5530384203275678, "grad_norm": 0.5119111139761892, "learning_rate": 5.509450139740437e-06, "loss": 1.2013, "step": 7226 }, { "epoch": 0.5531149548446349, "grad_norm": 0.9429624806298401, "learning_rate": 5.508466131322155e-06, "loss": 1.3646, "step": 7227 }, { "epoch": 0.5531914893617021, "grad_norm": 0.5034537507016209, "learning_rate": 5.507482153105185e-06, "loss": 0.9762, "step": 7228 }, { "epoch": 0.5532680238787693, "grad_norm": 0.5764360201289992, "learning_rate": 5.506498205149984e-06, "loss": 1.418, "step": 7229 }, { "epoch": 0.5533445583958365, "grad_norm": 0.62365999461287, "learning_rate": 5.505514287517008e-06, "loss": 1.2796, "step": 7230 }, { "epoch": 0.5534210929129038, "grad_norm": 0.5750649106261986, "learning_rate": 5.504530400266712e-06, "loss": 1.2709, "step": 7231 }, { "epoch": 0.5534976274299709, "grad_norm": 0.6955743480999261, "learning_rate": 5.503546543459549e-06, "loss": 1.6052, "step": 7232 }, { "epoch": 0.5535741619470381, "grad_norm": 0.7459028314006986, "learning_rate": 5.5025627171559725e-06, "loss": 1.3847, "step": 7233 }, { "epoch": 0.5536506964641054, "grad_norm": 0.5559317381891987, "learning_rate": 5.5015789214164286e-06, "loss": 1.3305, "step": 7234 }, { "epoch": 0.5537272309811725, "grad_norm": 0.6402910979412209, "learning_rate": 5.500595156301365e-06, "loss": 1.4094, "step": 7235 }, { "epoch": 0.5538037654982397, "grad_norm": 0.6024455323306248, "learning_rate": 5.499611421871229e-06, "loss": 1.2845, "step": 7236 }, { "epoch": 0.5538803000153069, "grad_norm": 0.6358811176509265, "learning_rate": 5.4986277181864634e-06, "loss": 1.4382, "step": 7237 }, { "epoch": 0.5539568345323741, "grad_norm": 1.8188102636871966, "learning_rate": 5.497644045307506e-06, "loss": 1.6085, "step": 7238 }, { "epoch": 0.5540333690494413, "grad_norm": 0.5924332769391603, "learning_rate": 5.496660403294805e-06, "loss": 1.2166, "step": 7239 }, { "epoch": 0.5541099035665085, "grad_norm": 0.5895778040973235, "learning_rate": 5.4956767922087905e-06, "loss": 1.1129, "step": 7240 }, { "epoch": 0.5541864380835757, "grad_norm": 0.5592199999540044, "learning_rate": 5.494693212109904e-06, "loss": 1.2637, "step": 7241 }, { "epoch": 0.554262972600643, "grad_norm": 0.5585625253685416, "learning_rate": 5.493709663058575e-06, "loss": 1.182, "step": 7242 }, { "epoch": 0.5543395071177101, "grad_norm": 0.48972550968234496, "learning_rate": 5.492726145115239e-06, "loss": 1.194, "step": 7243 }, { "epoch": 0.5544160416347773, "grad_norm": 0.5130564741669047, "learning_rate": 5.491742658340324e-06, "loss": 1.1408, "step": 7244 }, { "epoch": 0.5544925761518444, "grad_norm": 0.5770383308192241, "learning_rate": 5.49075920279426e-06, "loss": 1.3683, "step": 7245 }, { "epoch": 0.5545691106689117, "grad_norm": 0.635791920168869, "learning_rate": 5.489775778537474e-06, "loss": 1.0616, "step": 7246 }, { "epoch": 0.5546456451859789, "grad_norm": 0.5623374796068473, "learning_rate": 5.488792385630388e-06, "loss": 1.3436, "step": 7247 }, { "epoch": 0.554722179703046, "grad_norm": 0.6891043923431526, "learning_rate": 5.487809024133428e-06, "loss": 1.4296, "step": 7248 }, { "epoch": 0.5547987142201133, "grad_norm": 0.52539111906368, "learning_rate": 5.48682569410701e-06, "loss": 1.3469, "step": 7249 }, { "epoch": 0.5548752487371804, "grad_norm": 0.6709987304583083, "learning_rate": 5.485842395611556e-06, "loss": 1.313, "step": 7250 }, { "epoch": 0.5549517832542477, "grad_norm": 0.5401014338404119, "learning_rate": 5.484859128707484e-06, "loss": 1.2461, "step": 7251 }, { "epoch": 0.5550283177713149, "grad_norm": 0.6153108373048539, "learning_rate": 5.483875893455206e-06, "loss": 1.4967, "step": 7252 }, { "epoch": 0.555104852288382, "grad_norm": 0.5997318632466113, "learning_rate": 5.482892689915136e-06, "loss": 1.4802, "step": 7253 }, { "epoch": 0.5551813868054493, "grad_norm": 0.5644454220348095, "learning_rate": 5.481909518147684e-06, "loss": 1.5314, "step": 7254 }, { "epoch": 0.5552579213225165, "grad_norm": 0.6047354670345, "learning_rate": 5.480926378213263e-06, "loss": 1.5921, "step": 7255 }, { "epoch": 0.5553344558395836, "grad_norm": 0.5615946711957082, "learning_rate": 5.479943270172274e-06, "loss": 1.3206, "step": 7256 }, { "epoch": 0.5554109903566509, "grad_norm": 1.1317378623718441, "learning_rate": 5.478960194085127e-06, "loss": 1.4313, "step": 7257 }, { "epoch": 0.555487524873718, "grad_norm": 0.8298504241807747, "learning_rate": 5.477977150012223e-06, "loss": 1.4489, "step": 7258 }, { "epoch": 0.5555640593907852, "grad_norm": 0.564155256015017, "learning_rate": 5.476994138013962e-06, "loss": 1.1957, "step": 7259 }, { "epoch": 0.5556405939078525, "grad_norm": 0.574414693511207, "learning_rate": 5.4760111581507445e-06, "loss": 1.3702, "step": 7260 }, { "epoch": 0.5557171284249196, "grad_norm": 0.744171680909794, "learning_rate": 5.47502821048297e-06, "loss": 1.6841, "step": 7261 }, { "epoch": 0.5557936629419868, "grad_norm": 0.515285322319866, "learning_rate": 5.474045295071031e-06, "loss": 1.2641, "step": 7262 }, { "epoch": 0.5558701974590541, "grad_norm": 0.9142895102068634, "learning_rate": 5.4730624119753215e-06, "loss": 1.8883, "step": 7263 }, { "epoch": 0.5559467319761212, "grad_norm": 0.571159142710414, "learning_rate": 5.472079561256233e-06, "loss": 1.3713, "step": 7264 }, { "epoch": 0.5560232664931885, "grad_norm": 0.530358776474261, "learning_rate": 5.4710967429741525e-06, "loss": 1.1289, "step": 7265 }, { "epoch": 0.5560998010102556, "grad_norm": 0.5895501895869845, "learning_rate": 5.470113957189471e-06, "loss": 1.5124, "step": 7266 }, { "epoch": 0.5561763355273228, "grad_norm": 0.7656214429449675, "learning_rate": 5.469131203962571e-06, "loss": 1.2398, "step": 7267 }, { "epoch": 0.5562528700443901, "grad_norm": 0.46862756286189744, "learning_rate": 5.468148483353838e-06, "loss": 1.0364, "step": 7268 }, { "epoch": 0.5563294045614572, "grad_norm": 0.612760847126111, "learning_rate": 5.467165795423652e-06, "loss": 1.4396, "step": 7269 }, { "epoch": 0.5564059390785244, "grad_norm": 0.536695049849297, "learning_rate": 5.466183140232389e-06, "loss": 1.0863, "step": 7270 }, { "epoch": 0.5564824735955917, "grad_norm": 0.6999568013961223, "learning_rate": 5.4652005178404336e-06, "loss": 1.8813, "step": 7271 }, { "epoch": 0.5565590081126588, "grad_norm": 0.5456689342748429, "learning_rate": 5.464217928308156e-06, "loss": 1.2507, "step": 7272 }, { "epoch": 0.556635542629726, "grad_norm": 0.5963766496228443, "learning_rate": 5.463235371695931e-06, "loss": 1.3067, "step": 7273 }, { "epoch": 0.5567120771467932, "grad_norm": 0.6134852393197532, "learning_rate": 5.462252848064129e-06, "loss": 1.5532, "step": 7274 }, { "epoch": 0.5567886116638604, "grad_norm": 0.5714341492655192, "learning_rate": 5.461270357473119e-06, "loss": 1.4638, "step": 7275 }, { "epoch": 0.5568651461809276, "grad_norm": 0.6487289345399753, "learning_rate": 5.460287899983268e-06, "loss": 1.4708, "step": 7276 }, { "epoch": 0.5569416806979948, "grad_norm": 0.644165798779875, "learning_rate": 5.4593054756549445e-06, "loss": 1.631, "step": 7277 }, { "epoch": 0.557018215215062, "grad_norm": 0.5717046523966083, "learning_rate": 5.4583230845485066e-06, "loss": 1.2741, "step": 7278 }, { "epoch": 0.5570947497321291, "grad_norm": 0.5223780550679659, "learning_rate": 5.45734072672432e-06, "loss": 1.1113, "step": 7279 }, { "epoch": 0.5571712842491964, "grad_norm": 0.5263834556427179, "learning_rate": 5.45635840224274e-06, "loss": 1.3132, "step": 7280 }, { "epoch": 0.5572478187662636, "grad_norm": 0.7442121700310022, "learning_rate": 5.455376111164123e-06, "loss": 1.8128, "step": 7281 }, { "epoch": 0.5573243532833307, "grad_norm": 0.5353328064486014, "learning_rate": 5.4543938535488285e-06, "loss": 1.181, "step": 7282 }, { "epoch": 0.557400887800398, "grad_norm": 0.6302051945164284, "learning_rate": 5.453411629457206e-06, "loss": 1.499, "step": 7283 }, { "epoch": 0.5574774223174652, "grad_norm": 1.2073081831336059, "learning_rate": 5.452429438949607e-06, "loss": 1.3834, "step": 7284 }, { "epoch": 0.5575539568345323, "grad_norm": 0.5815294757527563, "learning_rate": 5.451447282086381e-06, "loss": 1.4509, "step": 7285 }, { "epoch": 0.5576304913515996, "grad_norm": 0.5617043607618453, "learning_rate": 5.4504651589278735e-06, "loss": 1.2436, "step": 7286 }, { "epoch": 0.5577070258686667, "grad_norm": 0.5178848368397325, "learning_rate": 5.449483069534428e-06, "loss": 1.2406, "step": 7287 }, { "epoch": 0.557783560385734, "grad_norm": 0.6051247103753106, "learning_rate": 5.44850101396639e-06, "loss": 1.3982, "step": 7288 }, { "epoch": 0.5578600949028012, "grad_norm": 0.6297412812824426, "learning_rate": 5.447518992284098e-06, "loss": 1.6373, "step": 7289 }, { "epoch": 0.5579366294198683, "grad_norm": 0.7674257131851819, "learning_rate": 5.446537004547891e-06, "loss": 1.7416, "step": 7290 }, { "epoch": 0.5580131639369356, "grad_norm": 0.5477679717433814, "learning_rate": 5.445555050818104e-06, "loss": 1.166, "step": 7291 }, { "epoch": 0.5580896984540028, "grad_norm": 0.5066579209734174, "learning_rate": 5.444573131155071e-06, "loss": 1.0077, "step": 7292 }, { "epoch": 0.5581662329710699, "grad_norm": 0.7465922308844984, "learning_rate": 5.443591245619126e-06, "loss": 1.5302, "step": 7293 }, { "epoch": 0.5582427674881372, "grad_norm": 0.6262140142573134, "learning_rate": 5.4426093942706e-06, "loss": 1.4964, "step": 7294 }, { "epoch": 0.5583193020052043, "grad_norm": 0.6017384565172436, "learning_rate": 5.441627577169819e-06, "loss": 1.3091, "step": 7295 }, { "epoch": 0.5583958365222715, "grad_norm": 0.6861502686322948, "learning_rate": 5.440645794377108e-06, "loss": 1.3606, "step": 7296 }, { "epoch": 0.5584723710393388, "grad_norm": 0.7163339426303145, "learning_rate": 5.4396640459527915e-06, "loss": 1.6593, "step": 7297 }, { "epoch": 0.5585489055564059, "grad_norm": 0.7296250348881259, "learning_rate": 5.43868233195719e-06, "loss": 1.3191, "step": 7298 }, { "epoch": 0.5586254400734731, "grad_norm": 0.5468441908709905, "learning_rate": 5.437700652450626e-06, "loss": 1.3448, "step": 7299 }, { "epoch": 0.5587019745905404, "grad_norm": 0.6413586575924701, "learning_rate": 5.436719007493414e-06, "loss": 1.6733, "step": 7300 }, { "epoch": 0.5587785091076075, "grad_norm": 0.5644429206626186, "learning_rate": 5.4357373971458704e-06, "loss": 1.2799, "step": 7301 }, { "epoch": 0.5588550436246748, "grad_norm": 0.5995992653479227, "learning_rate": 5.434755821468308e-06, "loss": 1.4471, "step": 7302 }, { "epoch": 0.5589315781417419, "grad_norm": 0.5979567224932013, "learning_rate": 5.433774280521036e-06, "loss": 1.398, "step": 7303 }, { "epoch": 0.5590081126588091, "grad_norm": 0.5747644372570578, "learning_rate": 5.4327927743643675e-06, "loss": 1.3735, "step": 7304 }, { "epoch": 0.5590846471758764, "grad_norm": 0.6928629471944806, "learning_rate": 5.431811303058605e-06, "loss": 1.3718, "step": 7305 }, { "epoch": 0.5591611816929435, "grad_norm": 0.58455136941194, "learning_rate": 5.430829866664055e-06, "loss": 1.3417, "step": 7306 }, { "epoch": 0.5592377162100107, "grad_norm": 0.6895084560478691, "learning_rate": 5.42984846524102e-06, "loss": 1.3444, "step": 7307 }, { "epoch": 0.559314250727078, "grad_norm": 0.5425067711141487, "learning_rate": 5.428867098849799e-06, "loss": 1.0391, "step": 7308 }, { "epoch": 0.5593907852441451, "grad_norm": 0.6336935765485421, "learning_rate": 5.42788576755069e-06, "loss": 1.606, "step": 7309 }, { "epoch": 0.5594673197612123, "grad_norm": 0.5874301056950543, "learning_rate": 5.426904471403991e-06, "loss": 1.4844, "step": 7310 }, { "epoch": 0.5595438542782795, "grad_norm": 0.48310662004423643, "learning_rate": 5.425923210469994e-06, "loss": 0.9877, "step": 7311 }, { "epoch": 0.5596203887953467, "grad_norm": 0.596361228418462, "learning_rate": 5.424941984808992e-06, "loss": 1.2649, "step": 7312 }, { "epoch": 0.5596969233124139, "grad_norm": 0.5561059978074951, "learning_rate": 5.423960794481273e-06, "loss": 1.3439, "step": 7313 }, { "epoch": 0.5597734578294811, "grad_norm": 0.582366163198552, "learning_rate": 5.422979639547123e-06, "loss": 1.2057, "step": 7314 }, { "epoch": 0.5598499923465483, "grad_norm": 0.5118376573491226, "learning_rate": 5.421998520066831e-06, "loss": 1.0823, "step": 7315 }, { "epoch": 0.5599265268636154, "grad_norm": 0.6428097347756186, "learning_rate": 5.421017436100679e-06, "loss": 1.601, "step": 7316 }, { "epoch": 0.5600030613806827, "grad_norm": 0.5168179252607228, "learning_rate": 5.420036387708944e-06, "loss": 1.2314, "step": 7317 }, { "epoch": 0.5600795958977499, "grad_norm": 0.5302088043736972, "learning_rate": 5.41905537495191e-06, "loss": 1.2588, "step": 7318 }, { "epoch": 0.560156130414817, "grad_norm": 0.52718758442433, "learning_rate": 5.418074397889848e-06, "loss": 1.0478, "step": 7319 }, { "epoch": 0.5602326649318843, "grad_norm": 0.5718305272796216, "learning_rate": 5.417093456583033e-06, "loss": 1.4547, "step": 7320 }, { "epoch": 0.5603091994489515, "grad_norm": 0.5530404951113006, "learning_rate": 5.416112551091742e-06, "loss": 1.391, "step": 7321 }, { "epoch": 0.5603857339660187, "grad_norm": 0.6433641811456622, "learning_rate": 5.41513168147624e-06, "loss": 1.4971, "step": 7322 }, { "epoch": 0.5604622684830859, "grad_norm": 0.5527011541246024, "learning_rate": 5.414150847796796e-06, "loss": 1.4102, "step": 7323 }, { "epoch": 0.560538803000153, "grad_norm": 0.7345649152946021, "learning_rate": 5.413170050113677e-06, "loss": 1.3938, "step": 7324 }, { "epoch": 0.5606153375172203, "grad_norm": 0.6446281665646657, "learning_rate": 5.412189288487139e-06, "loss": 1.4676, "step": 7325 }, { "epoch": 0.5606918720342875, "grad_norm": 0.5739269248629395, "learning_rate": 5.4112085629774535e-06, "loss": 1.2746, "step": 7326 }, { "epoch": 0.5607684065513546, "grad_norm": 0.5510525252621635, "learning_rate": 5.410227873644873e-06, "loss": 1.3057, "step": 7327 }, { "epoch": 0.5608449410684219, "grad_norm": 0.5581777425200886, "learning_rate": 5.4092472205496545e-06, "loss": 1.3781, "step": 7328 }, { "epoch": 0.5609214755854891, "grad_norm": 0.9209169744938022, "learning_rate": 5.408266603752052e-06, "loss": 1.4634, "step": 7329 }, { "epoch": 0.5609980101025562, "grad_norm": 0.6159220766411844, "learning_rate": 5.40728602331232e-06, "loss": 1.4877, "step": 7330 }, { "epoch": 0.5610745446196235, "grad_norm": 0.639785314209327, "learning_rate": 5.406305479290705e-06, "loss": 1.6195, "step": 7331 }, { "epoch": 0.5611510791366906, "grad_norm": 0.6374490876693338, "learning_rate": 5.405324971747457e-06, "loss": 1.5505, "step": 7332 }, { "epoch": 0.5612276136537578, "grad_norm": 0.6097309121150658, "learning_rate": 5.404344500742822e-06, "loss": 1.4007, "step": 7333 }, { "epoch": 0.5613041481708251, "grad_norm": 0.565799670180353, "learning_rate": 5.40336406633704e-06, "loss": 1.3183, "step": 7334 }, { "epoch": 0.5613806826878922, "grad_norm": 0.5590534015980441, "learning_rate": 5.402383668590353e-06, "loss": 1.3467, "step": 7335 }, { "epoch": 0.5614572172049594, "grad_norm": 0.7789993031610957, "learning_rate": 5.401403307562999e-06, "loss": 1.5255, "step": 7336 }, { "epoch": 0.5615337517220267, "grad_norm": 0.8186679449278351, "learning_rate": 5.4004229833152166e-06, "loss": 1.5066, "step": 7337 }, { "epoch": 0.5616102862390938, "grad_norm": 0.6074615870599392, "learning_rate": 5.399442695907239e-06, "loss": 1.2733, "step": 7338 }, { "epoch": 0.561686820756161, "grad_norm": 0.6040490482422982, "learning_rate": 5.398462445399297e-06, "loss": 1.3097, "step": 7339 }, { "epoch": 0.5617633552732282, "grad_norm": 0.7175937716100903, "learning_rate": 5.3974822318516205e-06, "loss": 1.7376, "step": 7340 }, { "epoch": 0.5618398897902954, "grad_norm": 0.5561004787546789, "learning_rate": 5.396502055324438e-06, "loss": 1.1217, "step": 7341 }, { "epoch": 0.5619164243073627, "grad_norm": 0.6384358463659737, "learning_rate": 5.39552191587797e-06, "loss": 1.5469, "step": 7342 }, { "epoch": 0.5619929588244298, "grad_norm": 0.5125648855647684, "learning_rate": 5.394541813572446e-06, "loss": 1.0781, "step": 7343 }, { "epoch": 0.562069493341497, "grad_norm": 0.5544090079637224, "learning_rate": 5.3935617484680815e-06, "loss": 1.3556, "step": 7344 }, { "epoch": 0.5621460278585643, "grad_norm": 0.5628139651249103, "learning_rate": 5.392581720625096e-06, "loss": 1.3524, "step": 7345 }, { "epoch": 0.5622225623756314, "grad_norm": 0.497110364353968, "learning_rate": 5.391601730103707e-06, "loss": 1.162, "step": 7346 }, { "epoch": 0.5622990968926986, "grad_norm": 0.580947458827946, "learning_rate": 5.390621776964123e-06, "loss": 1.4351, "step": 7347 }, { "epoch": 0.5623756314097658, "grad_norm": 0.6251451642753958, "learning_rate": 5.3896418612665614e-06, "loss": 1.5198, "step": 7348 }, { "epoch": 0.562452165926833, "grad_norm": 0.44478492130453323, "learning_rate": 5.388661983071228e-06, "loss": 0.841, "step": 7349 }, { "epoch": 0.5625287004439002, "grad_norm": 0.910192340896635, "learning_rate": 5.387682142438331e-06, "loss": 1.7424, "step": 7350 }, { "epoch": 0.5626052349609674, "grad_norm": 0.6199688067169217, "learning_rate": 5.3867023394280715e-06, "loss": 1.5098, "step": 7351 }, { "epoch": 0.5626817694780346, "grad_norm": 0.6181469261607835, "learning_rate": 5.385722574100656e-06, "loss": 1.5373, "step": 7352 }, { "epoch": 0.5627583039951017, "grad_norm": 0.5367748977673646, "learning_rate": 5.38474284651628e-06, "loss": 1.1011, "step": 7353 }, { "epoch": 0.562834838512169, "grad_norm": 1.1539219407129937, "learning_rate": 5.383763156735143e-06, "loss": 1.5255, "step": 7354 }, { "epoch": 0.5629113730292362, "grad_norm": 0.62739561685277, "learning_rate": 5.38278350481744e-06, "loss": 1.4611, "step": 7355 }, { "epoch": 0.5629879075463033, "grad_norm": 0.6579974756615694, "learning_rate": 5.381803890823365e-06, "loss": 1.467, "step": 7356 }, { "epoch": 0.5630644420633706, "grad_norm": 0.4874256762464496, "learning_rate": 5.380824314813106e-06, "loss": 1.0643, "step": 7357 }, { "epoch": 0.5631409765804378, "grad_norm": 0.5443455098432574, "learning_rate": 5.3798447768468495e-06, "loss": 1.2431, "step": 7358 }, { "epoch": 0.563217511097505, "grad_norm": 0.5830295224560824, "learning_rate": 5.378865276984787e-06, "loss": 1.441, "step": 7359 }, { "epoch": 0.5632940456145722, "grad_norm": 0.5113919442011648, "learning_rate": 5.377885815287098e-06, "loss": 0.891, "step": 7360 }, { "epoch": 0.5633705801316393, "grad_norm": 0.9382365210924857, "learning_rate": 5.376906391813964e-06, "loss": 1.2434, "step": 7361 }, { "epoch": 0.5634471146487066, "grad_norm": 0.7399108737158215, "learning_rate": 5.375927006625566e-06, "loss": 1.7318, "step": 7362 }, { "epoch": 0.5635236491657738, "grad_norm": 0.5484288681607646, "learning_rate": 5.374947659782075e-06, "loss": 1.1666, "step": 7363 }, { "epoch": 0.5636001836828409, "grad_norm": 0.6041086210676246, "learning_rate": 5.37396835134367e-06, "loss": 1.4204, "step": 7364 }, { "epoch": 0.5636767181999082, "grad_norm": 0.5326521744187596, "learning_rate": 5.372989081370521e-06, "loss": 0.9789, "step": 7365 }, { "epoch": 0.5637532527169754, "grad_norm": 0.7798078776350489, "learning_rate": 5.372009849922797e-06, "loss": 1.7748, "step": 7366 }, { "epoch": 0.5638297872340425, "grad_norm": 0.630714256064873, "learning_rate": 5.371030657060665e-06, "loss": 1.4182, "step": 7367 }, { "epoch": 0.5639063217511098, "grad_norm": 0.8213009665818584, "learning_rate": 5.370051502844291e-06, "loss": 1.6117, "step": 7368 }, { "epoch": 0.5639828562681769, "grad_norm": 0.49864772268980445, "learning_rate": 5.369072387333833e-06, "loss": 1.1487, "step": 7369 }, { "epoch": 0.5640593907852441, "grad_norm": 0.6727109715343056, "learning_rate": 5.368093310589455e-06, "loss": 1.4116, "step": 7370 }, { "epoch": 0.5641359253023114, "grad_norm": 0.6363680423206626, "learning_rate": 5.367114272671314e-06, "loss": 1.3149, "step": 7371 }, { "epoch": 0.5642124598193785, "grad_norm": 0.6764696483651738, "learning_rate": 5.366135273639563e-06, "loss": 1.5189, "step": 7372 }, { "epoch": 0.5642889943364457, "grad_norm": 0.5310184617368326, "learning_rate": 5.365156313554357e-06, "loss": 1.2738, "step": 7373 }, { "epoch": 0.564365528853513, "grad_norm": 0.5551469935315873, "learning_rate": 5.364177392475845e-06, "loss": 1.1962, "step": 7374 }, { "epoch": 0.5644420633705801, "grad_norm": 0.548511384655289, "learning_rate": 5.363198510464173e-06, "loss": 1.2148, "step": 7375 }, { "epoch": 0.5645185978876474, "grad_norm": 0.569320960805095, "learning_rate": 5.362219667579489e-06, "loss": 1.1832, "step": 7376 }, { "epoch": 0.5645951324047145, "grad_norm": 0.5335317148112537, "learning_rate": 5.361240863881936e-06, "loss": 1.2812, "step": 7377 }, { "epoch": 0.5646716669217817, "grad_norm": 0.5326633143019757, "learning_rate": 5.360262099431652e-06, "loss": 1.2039, "step": 7378 }, { "epoch": 0.564748201438849, "grad_norm": 0.6536006344489428, "learning_rate": 5.3592833742887795e-06, "loss": 1.785, "step": 7379 }, { "epoch": 0.5648247359559161, "grad_norm": 0.5931756042801924, "learning_rate": 5.3583046885134494e-06, "loss": 1.4899, "step": 7380 }, { "epoch": 0.5649012704729833, "grad_norm": 0.5940865427685464, "learning_rate": 5.357326042165798e-06, "loss": 1.4448, "step": 7381 }, { "epoch": 0.5649778049900505, "grad_norm": 0.5428390442473897, "learning_rate": 5.356347435305958e-06, "loss": 1.1386, "step": 7382 }, { "epoch": 0.5650543395071177, "grad_norm": 0.6079603871015908, "learning_rate": 5.355368867994053e-06, "loss": 1.504, "step": 7383 }, { "epoch": 0.5651308740241849, "grad_norm": 0.6073033393472013, "learning_rate": 5.354390340290214e-06, "loss": 1.4712, "step": 7384 }, { "epoch": 0.5652074085412521, "grad_norm": 0.7713514257146608, "learning_rate": 5.353411852254562e-06, "loss": 1.4199, "step": 7385 }, { "epoch": 0.5652839430583193, "grad_norm": 0.6116405286316736, "learning_rate": 5.352433403947217e-06, "loss": 1.3218, "step": 7386 }, { "epoch": 0.5653604775753865, "grad_norm": 0.6483805085979822, "learning_rate": 5.351454995428301e-06, "loss": 1.712, "step": 7387 }, { "epoch": 0.5654370120924537, "grad_norm": 0.5388264342559427, "learning_rate": 5.350476626757929e-06, "loss": 1.252, "step": 7388 }, { "epoch": 0.5655135466095209, "grad_norm": 0.6182319444357066, "learning_rate": 5.349498297996214e-06, "loss": 1.2899, "step": 7389 }, { "epoch": 0.565590081126588, "grad_norm": 0.622818378433151, "learning_rate": 5.34852000920327e-06, "loss": 1.2162, "step": 7390 }, { "epoch": 0.5656666156436553, "grad_norm": 0.5913465990346344, "learning_rate": 5.347541760439201e-06, "loss": 1.1917, "step": 7391 }, { "epoch": 0.5657431501607225, "grad_norm": 0.9494385047197448, "learning_rate": 5.346563551764118e-06, "loss": 1.6367, "step": 7392 }, { "epoch": 0.5658196846777896, "grad_norm": 0.8617975300045364, "learning_rate": 5.345585383238126e-06, "loss": 1.3725, "step": 7393 }, { "epoch": 0.5658962191948569, "grad_norm": 0.6611792920921397, "learning_rate": 5.344607254921321e-06, "loss": 1.7166, "step": 7394 }, { "epoch": 0.5659727537119241, "grad_norm": 0.6041281233636656, "learning_rate": 5.343629166873809e-06, "loss": 1.6112, "step": 7395 }, { "epoch": 0.5660492882289913, "grad_norm": 0.7331674198463694, "learning_rate": 5.34265111915568e-06, "loss": 1.3996, "step": 7396 }, { "epoch": 0.5661258227460585, "grad_norm": 0.5516763968922147, "learning_rate": 5.341673111827032e-06, "loss": 1.1262, "step": 7397 }, { "epoch": 0.5662023572631256, "grad_norm": 0.5706024548651454, "learning_rate": 5.340695144947958e-06, "loss": 1.3343, "step": 7398 }, { "epoch": 0.5662788917801929, "grad_norm": 0.5531323946218956, "learning_rate": 5.339717218578544e-06, "loss": 1.4629, "step": 7399 }, { "epoch": 0.5663554262972601, "grad_norm": 0.5997952106695824, "learning_rate": 5.338739332778878e-06, "loss": 1.3746, "step": 7400 }, { "epoch": 0.5664319608143272, "grad_norm": 0.4923376889191621, "learning_rate": 5.337761487609044e-06, "loss": 1.1624, "step": 7401 }, { "epoch": 0.5665084953313945, "grad_norm": 0.7100146997010194, "learning_rate": 5.336783683129123e-06, "loss": 1.5516, "step": 7402 }, { "epoch": 0.5665850298484617, "grad_norm": 0.557927492758687, "learning_rate": 5.335805919399196e-06, "loss": 1.2411, "step": 7403 }, { "epoch": 0.5666615643655288, "grad_norm": 0.5132709023894791, "learning_rate": 5.3348281964793405e-06, "loss": 1.2117, "step": 7404 }, { "epoch": 0.5667380988825961, "grad_norm": 0.612492896077706, "learning_rate": 5.333850514429627e-06, "loss": 1.4783, "step": 7405 }, { "epoch": 0.5668146333996632, "grad_norm": 0.5147163506484934, "learning_rate": 5.33287287331013e-06, "loss": 1.0517, "step": 7406 }, { "epoch": 0.5668911679167304, "grad_norm": 0.6556159222294985, "learning_rate": 5.3318952731809184e-06, "loss": 1.5618, "step": 7407 }, { "epoch": 0.5669677024337977, "grad_norm": 0.5974647863290915, "learning_rate": 5.330917714102056e-06, "loss": 1.5706, "step": 7408 }, { "epoch": 0.5670442369508648, "grad_norm": 0.599819879516954, "learning_rate": 5.329940196133612e-06, "loss": 1.4893, "step": 7409 }, { "epoch": 0.567120771467932, "grad_norm": 0.7193846725727373, "learning_rate": 5.328962719335646e-06, "loss": 1.6367, "step": 7410 }, { "epoch": 0.5671973059849993, "grad_norm": 0.653246778216848, "learning_rate": 5.327985283768215e-06, "loss": 1.4743, "step": 7411 }, { "epoch": 0.5672738405020664, "grad_norm": 0.8849656707680714, "learning_rate": 5.327007889491378e-06, "loss": 1.5535, "step": 7412 }, { "epoch": 0.5673503750191337, "grad_norm": 0.7496068847123043, "learning_rate": 5.326030536565186e-06, "loss": 1.6219, "step": 7413 }, { "epoch": 0.5674269095362008, "grad_norm": 0.5522281175047019, "learning_rate": 5.325053225049695e-06, "loss": 1.2863, "step": 7414 }, { "epoch": 0.567503444053268, "grad_norm": 0.6043450308729771, "learning_rate": 5.324075955004951e-06, "loss": 1.3236, "step": 7415 }, { "epoch": 0.5675799785703353, "grad_norm": 0.5721719715656003, "learning_rate": 5.323098726491e-06, "loss": 1.2778, "step": 7416 }, { "epoch": 0.5676565130874024, "grad_norm": 0.6316291966584785, "learning_rate": 5.322121539567888e-06, "loss": 1.5965, "step": 7417 }, { "epoch": 0.5677330476044696, "grad_norm": 0.5485106937890937, "learning_rate": 5.321144394295654e-06, "loss": 1.352, "step": 7418 }, { "epoch": 0.5678095821215368, "grad_norm": 0.8021828128184225, "learning_rate": 5.320167290734339e-06, "loss": 1.5746, "step": 7419 }, { "epoch": 0.567886116638604, "grad_norm": 0.5953778805884257, "learning_rate": 5.319190228943978e-06, "loss": 1.3752, "step": 7420 }, { "epoch": 0.5679626511556712, "grad_norm": 0.5256042705924505, "learning_rate": 5.318213208984604e-06, "loss": 1.0723, "step": 7421 }, { "epoch": 0.5680391856727384, "grad_norm": 0.6314545901850485, "learning_rate": 5.317236230916248e-06, "loss": 1.459, "step": 7422 }, { "epoch": 0.5681157201898056, "grad_norm": 0.6567031396387909, "learning_rate": 5.31625929479894e-06, "loss": 1.5797, "step": 7423 }, { "epoch": 0.5681922547068728, "grad_norm": 0.627848331713443, "learning_rate": 5.315282400692704e-06, "loss": 1.2706, "step": 7424 }, { "epoch": 0.56826878922394, "grad_norm": 0.5368888595083259, "learning_rate": 5.3143055486575655e-06, "loss": 1.2263, "step": 7425 }, { "epoch": 0.5683453237410072, "grad_norm": 0.6102849969644957, "learning_rate": 5.313328738753543e-06, "loss": 1.5725, "step": 7426 }, { "epoch": 0.5684218582580743, "grad_norm": 0.8802428241510191, "learning_rate": 5.3123519710406556e-06, "loss": 1.4504, "step": 7427 }, { "epoch": 0.5684983927751416, "grad_norm": 0.5399071240383676, "learning_rate": 5.3113752455789195e-06, "loss": 1.2173, "step": 7428 }, { "epoch": 0.5685749272922088, "grad_norm": 0.6180891465095667, "learning_rate": 5.310398562428345e-06, "loss": 1.2642, "step": 7429 }, { "epoch": 0.568651461809276, "grad_norm": 0.6225096039018979, "learning_rate": 5.309421921648945e-06, "loss": 1.4299, "step": 7430 }, { "epoch": 0.5687279963263432, "grad_norm": 0.5934862121906789, "learning_rate": 5.308445323300728e-06, "loss": 1.3719, "step": 7431 }, { "epoch": 0.5688045308434104, "grad_norm": 0.5650616407551255, "learning_rate": 5.307468767443697e-06, "loss": 1.3496, "step": 7432 }, { "epoch": 0.5688810653604776, "grad_norm": 0.5563213872454531, "learning_rate": 5.306492254137856e-06, "loss": 1.2506, "step": 7433 }, { "epoch": 0.5689575998775448, "grad_norm": 0.63295329554446, "learning_rate": 5.305515783443204e-06, "loss": 1.5316, "step": 7434 }, { "epoch": 0.5690341343946119, "grad_norm": 1.0773311483000516, "learning_rate": 5.304539355419736e-06, "loss": 1.6091, "step": 7435 }, { "epoch": 0.5691106689116792, "grad_norm": 0.6478628670603426, "learning_rate": 5.303562970127452e-06, "loss": 1.4421, "step": 7436 }, { "epoch": 0.5691872034287464, "grad_norm": 0.7194679311166988, "learning_rate": 5.302586627626339e-06, "loss": 1.4586, "step": 7437 }, { "epoch": 0.5692637379458135, "grad_norm": 0.5762584663738772, "learning_rate": 5.301610327976389e-06, "loss": 1.1317, "step": 7438 }, { "epoch": 0.5693402724628808, "grad_norm": 0.5991888167727987, "learning_rate": 5.3006340712375895e-06, "loss": 1.2707, "step": 7439 }, { "epoch": 0.569416806979948, "grad_norm": 1.3005231549073117, "learning_rate": 5.29965785746992e-06, "loss": 1.6721, "step": 7440 }, { "epoch": 0.5694933414970151, "grad_norm": 0.556784446849206, "learning_rate": 5.2986816867333675e-06, "loss": 1.2994, "step": 7441 }, { "epoch": 0.5695698760140824, "grad_norm": 0.48818539079427076, "learning_rate": 5.29770555908791e-06, "loss": 1.1535, "step": 7442 }, { "epoch": 0.5696464105311495, "grad_norm": 0.6468636094973036, "learning_rate": 5.2967294745935215e-06, "loss": 1.5167, "step": 7443 }, { "epoch": 0.5697229450482167, "grad_norm": 0.5445693789895144, "learning_rate": 5.295753433310176e-06, "loss": 1.2116, "step": 7444 }, { "epoch": 0.569799479565284, "grad_norm": 0.5836127482235298, "learning_rate": 5.294777435297846e-06, "loss": 1.1739, "step": 7445 }, { "epoch": 0.5698760140823511, "grad_norm": 0.61308269882652, "learning_rate": 5.293801480616496e-06, "loss": 1.1793, "step": 7446 }, { "epoch": 0.5699525485994184, "grad_norm": 0.5434720406383522, "learning_rate": 5.292825569326097e-06, "loss": 1.3152, "step": 7447 }, { "epoch": 0.5700290831164855, "grad_norm": 0.558517715786013, "learning_rate": 5.291849701486607e-06, "loss": 1.4255, "step": 7448 }, { "epoch": 0.5701056176335527, "grad_norm": 0.5728839812057293, "learning_rate": 5.290873877157991e-06, "loss": 1.3037, "step": 7449 }, { "epoch": 0.57018215215062, "grad_norm": 0.6482323921903186, "learning_rate": 5.289898096400201e-06, "loss": 1.4023, "step": 7450 }, { "epoch": 0.5702586866676871, "grad_norm": 0.667130012106038, "learning_rate": 5.288922359273195e-06, "loss": 1.3736, "step": 7451 }, { "epoch": 0.5703352211847543, "grad_norm": 0.5671026918288753, "learning_rate": 5.287946665836926e-06, "loss": 1.3604, "step": 7452 }, { "epoch": 0.5704117557018216, "grad_norm": 0.7566037681791742, "learning_rate": 5.286971016151342e-06, "loss": 1.8317, "step": 7453 }, { "epoch": 0.5704882902188887, "grad_norm": 0.6572815337585004, "learning_rate": 5.285995410276391e-06, "loss": 1.5697, "step": 7454 }, { "epoch": 0.5705648247359559, "grad_norm": 0.5186737067913401, "learning_rate": 5.285019848272017e-06, "loss": 1.0045, "step": 7455 }, { "epoch": 0.5706413592530231, "grad_norm": 0.5629994933590513, "learning_rate": 5.284044330198159e-06, "loss": 1.1839, "step": 7456 }, { "epoch": 0.5707178937700903, "grad_norm": 0.5741803651052305, "learning_rate": 5.283068856114759e-06, "loss": 1.2797, "step": 7457 }, { "epoch": 0.5707944282871575, "grad_norm": 0.8685264576044692, "learning_rate": 5.282093426081752e-06, "loss": 1.5535, "step": 7458 }, { "epoch": 0.5708709628042247, "grad_norm": 0.6256591608484303, "learning_rate": 5.28111804015907e-06, "loss": 1.3882, "step": 7459 }, { "epoch": 0.5709474973212919, "grad_norm": 0.5847471502908692, "learning_rate": 5.2801426984066454e-06, "loss": 1.2327, "step": 7460 }, { "epoch": 0.5710240318383591, "grad_norm": 0.5364899579633207, "learning_rate": 5.2791674008844054e-06, "loss": 1.2137, "step": 7461 }, { "epoch": 0.5711005663554263, "grad_norm": 0.614083989261316, "learning_rate": 5.278192147652272e-06, "loss": 1.4781, "step": 7462 }, { "epoch": 0.5711771008724935, "grad_norm": 0.6232076368903305, "learning_rate": 5.277216938770174e-06, "loss": 1.3899, "step": 7463 }, { "epoch": 0.5712536353895606, "grad_norm": 0.6603016232918585, "learning_rate": 5.276241774298029e-06, "loss": 1.66, "step": 7464 }, { "epoch": 0.5713301699066279, "grad_norm": 0.5589497924350019, "learning_rate": 5.275266654295752e-06, "loss": 1.1913, "step": 7465 }, { "epoch": 0.5714067044236951, "grad_norm": 0.5938037561770821, "learning_rate": 5.274291578823258e-06, "loss": 1.3798, "step": 7466 }, { "epoch": 0.5714832389407623, "grad_norm": 0.6621048298477487, "learning_rate": 5.273316547940459e-06, "loss": 1.5469, "step": 7467 }, { "epoch": 0.5715597734578295, "grad_norm": 0.709849961755571, "learning_rate": 5.272341561707263e-06, "loss": 1.7007, "step": 7468 }, { "epoch": 0.5716363079748967, "grad_norm": 0.6084578877520856, "learning_rate": 5.271366620183577e-06, "loss": 1.4067, "step": 7469 }, { "epoch": 0.5717128424919639, "grad_norm": 0.5727222467856052, "learning_rate": 5.270391723429304e-06, "loss": 1.3771, "step": 7470 }, { "epoch": 0.5717893770090311, "grad_norm": 0.5591796447827069, "learning_rate": 5.269416871504344e-06, "loss": 1.4636, "step": 7471 }, { "epoch": 0.5718659115260982, "grad_norm": 0.49984403266748645, "learning_rate": 5.268442064468598e-06, "loss": 1.0223, "step": 7472 }, { "epoch": 0.5719424460431655, "grad_norm": 0.5596798104141408, "learning_rate": 5.267467302381954e-06, "loss": 1.2865, "step": 7473 }, { "epoch": 0.5720189805602327, "grad_norm": 0.5610268014801872, "learning_rate": 5.266492585304313e-06, "loss": 1.3118, "step": 7474 }, { "epoch": 0.5720955150772998, "grad_norm": 0.6211885187463191, "learning_rate": 5.265517913295559e-06, "loss": 1.4983, "step": 7475 }, { "epoch": 0.5721720495943671, "grad_norm": 0.5904343718210232, "learning_rate": 5.26454328641558e-06, "loss": 1.4125, "step": 7476 }, { "epoch": 0.5722485841114343, "grad_norm": 0.4644129575278201, "learning_rate": 5.2635687047242605e-06, "loss": 0.9016, "step": 7477 }, { "epoch": 0.5723251186285014, "grad_norm": 0.6335638893541959, "learning_rate": 5.26259416828148e-06, "loss": 1.5663, "step": 7478 }, { "epoch": 0.5724016531455687, "grad_norm": 0.5842751091797483, "learning_rate": 5.261619677147119e-06, "loss": 1.259, "step": 7479 }, { "epoch": 0.5724781876626358, "grad_norm": 1.20793989252507, "learning_rate": 5.260645231381053e-06, "loss": 1.6552, "step": 7480 }, { "epoch": 0.572554722179703, "grad_norm": 0.6450138352254963, "learning_rate": 5.259670831043154e-06, "loss": 1.1336, "step": 7481 }, { "epoch": 0.5726312566967703, "grad_norm": 0.5495595648466371, "learning_rate": 5.2586964761932926e-06, "loss": 1.3434, "step": 7482 }, { "epoch": 0.5727077912138374, "grad_norm": 0.6878570720584983, "learning_rate": 5.257722166891335e-06, "loss": 1.6841, "step": 7483 }, { "epoch": 0.5727843257309047, "grad_norm": 0.8130279900509971, "learning_rate": 5.256747903197146e-06, "loss": 1.5362, "step": 7484 }, { "epoch": 0.5728608602479718, "grad_norm": 0.6904358426734424, "learning_rate": 5.255773685170587e-06, "loss": 1.6781, "step": 7485 }, { "epoch": 0.572937394765039, "grad_norm": 0.5541067365565276, "learning_rate": 5.25479951287152e-06, "loss": 1.3476, "step": 7486 }, { "epoch": 0.5730139292821063, "grad_norm": 0.6411684226451315, "learning_rate": 5.2538253863597975e-06, "loss": 1.6355, "step": 7487 }, { "epoch": 0.5730904637991734, "grad_norm": 0.5484875322715291, "learning_rate": 5.252851305695274e-06, "loss": 1.1691, "step": 7488 }, { "epoch": 0.5731669983162406, "grad_norm": 0.7013423549240555, "learning_rate": 5.2518772709378006e-06, "loss": 1.8374, "step": 7489 }, { "epoch": 0.5732435328333079, "grad_norm": 0.6579055555931066, "learning_rate": 5.250903282147223e-06, "loss": 1.5512, "step": 7490 }, { "epoch": 0.573320067350375, "grad_norm": 0.8993409552740871, "learning_rate": 5.249929339383388e-06, "loss": 1.2602, "step": 7491 }, { "epoch": 0.5733966018674422, "grad_norm": 0.5557039524950167, "learning_rate": 5.248955442706137e-06, "loss": 1.2202, "step": 7492 }, { "epoch": 0.5734731363845094, "grad_norm": 0.5582366763583752, "learning_rate": 5.2479815921753076e-06, "loss": 1.1311, "step": 7493 }, { "epoch": 0.5735496709015766, "grad_norm": 0.5858831451912936, "learning_rate": 5.247007787850737e-06, "loss": 1.2641, "step": 7494 }, { "epoch": 0.5736262054186438, "grad_norm": 0.659731584524955, "learning_rate": 5.246034029792259e-06, "loss": 1.4233, "step": 7495 }, { "epoch": 0.573702739935711, "grad_norm": 0.592181013374354, "learning_rate": 5.245060318059705e-06, "loss": 1.3035, "step": 7496 }, { "epoch": 0.5737792744527782, "grad_norm": 0.5318212594763627, "learning_rate": 5.244086652712901e-06, "loss": 1.1128, "step": 7497 }, { "epoch": 0.5738558089698454, "grad_norm": 0.5372532978103027, "learning_rate": 5.243113033811674e-06, "loss": 1.1849, "step": 7498 }, { "epoch": 0.5739323434869126, "grad_norm": 0.633009481614199, "learning_rate": 5.242139461415845e-06, "loss": 1.42, "step": 7499 }, { "epoch": 0.5740088780039798, "grad_norm": 0.5800424592169029, "learning_rate": 5.2411659355852325e-06, "loss": 1.1717, "step": 7500 }, { "epoch": 0.574085412521047, "grad_norm": 0.5641038750567002, "learning_rate": 5.240192456379653e-06, "loss": 1.3166, "step": 7501 }, { "epoch": 0.5741619470381142, "grad_norm": 0.8805043440264742, "learning_rate": 5.23921902385892e-06, "loss": 1.4695, "step": 7502 }, { "epoch": 0.5742384815551814, "grad_norm": 0.6559409224982984, "learning_rate": 5.2382456380828464e-06, "loss": 1.5293, "step": 7503 }, { "epoch": 0.5743150160722486, "grad_norm": 0.6139049315579832, "learning_rate": 5.237272299111236e-06, "loss": 1.5443, "step": 7504 }, { "epoch": 0.5743915505893158, "grad_norm": 0.6137823153482745, "learning_rate": 5.236299007003897e-06, "loss": 1.4853, "step": 7505 }, { "epoch": 0.574468085106383, "grad_norm": 0.8102897292658884, "learning_rate": 5.235325761820628e-06, "loss": 1.5613, "step": 7506 }, { "epoch": 0.5745446196234502, "grad_norm": 0.5678161910257115, "learning_rate": 5.234352563621231e-06, "loss": 1.4158, "step": 7507 }, { "epoch": 0.5746211541405174, "grad_norm": 0.579051955196461, "learning_rate": 5.233379412465502e-06, "loss": 1.5351, "step": 7508 }, { "epoch": 0.5746976886575845, "grad_norm": 0.6533726532200299, "learning_rate": 5.232406308413234e-06, "loss": 1.624, "step": 7509 }, { "epoch": 0.5747742231746518, "grad_norm": 0.5452813534013152, "learning_rate": 5.231433251524217e-06, "loss": 1.1281, "step": 7510 }, { "epoch": 0.574850757691719, "grad_norm": 0.561489393712904, "learning_rate": 5.230460241858239e-06, "loss": 1.3551, "step": 7511 }, { "epoch": 0.5749272922087861, "grad_norm": 0.5456302305692307, "learning_rate": 5.229487279475083e-06, "loss": 1.0772, "step": 7512 }, { "epoch": 0.5750038267258534, "grad_norm": 0.5907249729259348, "learning_rate": 5.228514364434531e-06, "loss": 1.5967, "step": 7513 }, { "epoch": 0.5750803612429206, "grad_norm": 0.5794149661563963, "learning_rate": 5.227541496796365e-06, "loss": 1.429, "step": 7514 }, { "epoch": 0.5751568957599877, "grad_norm": 0.8478618310707635, "learning_rate": 5.226568676620358e-06, "loss": 1.8501, "step": 7515 }, { "epoch": 0.575233430277055, "grad_norm": 0.5956391314017447, "learning_rate": 5.225595903966284e-06, "loss": 1.339, "step": 7516 }, { "epoch": 0.5753099647941221, "grad_norm": 1.1220668606875446, "learning_rate": 5.22462317889391e-06, "loss": 1.4336, "step": 7517 }, { "epoch": 0.5753864993111893, "grad_norm": 0.5736344366373719, "learning_rate": 5.223650501463009e-06, "loss": 1.3784, "step": 7518 }, { "epoch": 0.5754630338282566, "grad_norm": 0.9067763099902502, "learning_rate": 5.2226778717333415e-06, "loss": 1.6471, "step": 7519 }, { "epoch": 0.5755395683453237, "grad_norm": 0.6976468590033162, "learning_rate": 5.2217052897646695e-06, "loss": 1.7248, "step": 7520 }, { "epoch": 0.575616102862391, "grad_norm": 0.6743292864752666, "learning_rate": 5.2207327556167494e-06, "loss": 1.5824, "step": 7521 }, { "epoch": 0.5756926373794581, "grad_norm": 0.5735206055722017, "learning_rate": 5.21976026934934e-06, "loss": 1.2436, "step": 7522 }, { "epoch": 0.5757691718965253, "grad_norm": 0.6099484539991992, "learning_rate": 5.218787831022191e-06, "loss": 1.481, "step": 7523 }, { "epoch": 0.5758457064135926, "grad_norm": 0.5840783579962229, "learning_rate": 5.2178154406950536e-06, "loss": 1.4198, "step": 7524 }, { "epoch": 0.5759222409306597, "grad_norm": 0.5521402186004705, "learning_rate": 5.216843098427674e-06, "loss": 1.1401, "step": 7525 }, { "epoch": 0.5759987754477269, "grad_norm": 0.6008886875299132, "learning_rate": 5.215870804279796e-06, "loss": 1.4017, "step": 7526 }, { "epoch": 0.5760753099647942, "grad_norm": 0.6272145732786911, "learning_rate": 5.214898558311159e-06, "loss": 1.4848, "step": 7527 }, { "epoch": 0.5761518444818613, "grad_norm": 0.5716947725648946, "learning_rate": 5.2139263605815e-06, "loss": 1.4056, "step": 7528 }, { "epoch": 0.5762283789989285, "grad_norm": 0.8129893668838477, "learning_rate": 5.212954211150558e-06, "loss": 1.3664, "step": 7529 }, { "epoch": 0.5763049135159957, "grad_norm": 0.6983428273968955, "learning_rate": 5.2119821100780625e-06, "loss": 1.5852, "step": 7530 }, { "epoch": 0.5763814480330629, "grad_norm": 0.5507798227926923, "learning_rate": 5.21101005742374e-06, "loss": 1.286, "step": 7531 }, { "epoch": 0.5764579825501301, "grad_norm": 0.644918467172006, "learning_rate": 5.210038053247319e-06, "loss": 1.5055, "step": 7532 }, { "epoch": 0.5765345170671973, "grad_norm": 0.8102903210989217, "learning_rate": 5.209066097608522e-06, "loss": 1.4629, "step": 7533 }, { "epoch": 0.5766110515842645, "grad_norm": 0.5188049005254265, "learning_rate": 5.208094190567067e-06, "loss": 1.1612, "step": 7534 }, { "epoch": 0.5766875861013318, "grad_norm": 0.6431429380701736, "learning_rate": 5.207122332182673e-06, "loss": 1.4738, "step": 7535 }, { "epoch": 0.5767641206183989, "grad_norm": 0.6018791308267974, "learning_rate": 5.206150522515055e-06, "loss": 1.3771, "step": 7536 }, { "epoch": 0.5768406551354661, "grad_norm": 0.5415996092833202, "learning_rate": 5.2051787616239195e-06, "loss": 1.2254, "step": 7537 }, { "epoch": 0.5769171896525332, "grad_norm": 0.6327400259696428, "learning_rate": 5.204207049568978e-06, "loss": 1.4156, "step": 7538 }, { "epoch": 0.5769937241696005, "grad_norm": 0.6095424875108957, "learning_rate": 5.2032353864099324e-06, "loss": 1.4648, "step": 7539 }, { "epoch": 0.5770702586866677, "grad_norm": 0.5742027845939759, "learning_rate": 5.2022637722064884e-06, "loss": 1.3791, "step": 7540 }, { "epoch": 0.5771467932037349, "grad_norm": 0.5946105753215427, "learning_rate": 5.201292207018342e-06, "loss": 1.3811, "step": 7541 }, { "epoch": 0.5772233277208021, "grad_norm": 0.6426612829263066, "learning_rate": 5.200320690905191e-06, "loss": 1.713, "step": 7542 }, { "epoch": 0.5772998622378693, "grad_norm": 0.6335764068361116, "learning_rate": 5.1993492239267284e-06, "loss": 1.4971, "step": 7543 }, { "epoch": 0.5773763967549365, "grad_norm": 0.5347187946677463, "learning_rate": 5.198377806142642e-06, "loss": 1.3131, "step": 7544 }, { "epoch": 0.5774529312720037, "grad_norm": 0.6100479646545665, "learning_rate": 5.19740643761262e-06, "loss": 1.3264, "step": 7545 }, { "epoch": 0.5775294657890708, "grad_norm": 1.2155029788957592, "learning_rate": 5.196435118396345e-06, "loss": 1.4656, "step": 7546 }, { "epoch": 0.5776060003061381, "grad_norm": 0.6208389520434388, "learning_rate": 5.195463848553498e-06, "loss": 1.5636, "step": 7547 }, { "epoch": 0.5776825348232053, "grad_norm": 0.7922088088431846, "learning_rate": 5.194492628143759e-06, "loss": 1.2115, "step": 7548 }, { "epoch": 0.5777590693402724, "grad_norm": 0.5177296471781067, "learning_rate": 5.1935214572268e-06, "loss": 1.2556, "step": 7549 }, { "epoch": 0.5778356038573397, "grad_norm": 0.5428994131417102, "learning_rate": 5.192550335862292e-06, "loss": 1.228, "step": 7550 }, { "epoch": 0.5779121383744068, "grad_norm": 0.5642414520641078, "learning_rate": 5.191579264109907e-06, "loss": 1.0916, "step": 7551 }, { "epoch": 0.577988672891474, "grad_norm": 0.5592386960391521, "learning_rate": 5.190608242029308e-06, "loss": 1.4147, "step": 7552 }, { "epoch": 0.5780652074085413, "grad_norm": 0.6078632187860319, "learning_rate": 5.189637269680159e-06, "loss": 1.4997, "step": 7553 }, { "epoch": 0.5781417419256084, "grad_norm": 0.5582089896517972, "learning_rate": 5.188666347122118e-06, "loss": 1.2593, "step": 7554 }, { "epoch": 0.5782182764426756, "grad_norm": 0.5062879240098265, "learning_rate": 5.1876954744148435e-06, "loss": 1.1412, "step": 7555 }, { "epoch": 0.5782948109597429, "grad_norm": 0.9781255717708752, "learning_rate": 5.1867246516179855e-06, "loss": 1.5066, "step": 7556 }, { "epoch": 0.57837134547681, "grad_norm": 0.5991843432444744, "learning_rate": 5.1857538787911975e-06, "loss": 1.3733, "step": 7557 }, { "epoch": 0.5784478799938773, "grad_norm": 0.584567472322272, "learning_rate": 5.184783155994124e-06, "loss": 1.4256, "step": 7558 }, { "epoch": 0.5785244145109444, "grad_norm": 0.8594912775284209, "learning_rate": 5.183812483286412e-06, "loss": 1.8486, "step": 7559 }, { "epoch": 0.5786009490280116, "grad_norm": 0.617899020267035, "learning_rate": 5.1828418607277e-06, "loss": 1.4567, "step": 7560 }, { "epoch": 0.5786774835450789, "grad_norm": 0.6680382429861766, "learning_rate": 5.181871288377625e-06, "loss": 1.7576, "step": 7561 }, { "epoch": 0.578754018062146, "grad_norm": 0.6731332884793055, "learning_rate": 5.1809007662958265e-06, "loss": 1.7448, "step": 7562 }, { "epoch": 0.5788305525792132, "grad_norm": 0.5267305972959142, "learning_rate": 5.179930294541933e-06, "loss": 1.0313, "step": 7563 }, { "epoch": 0.5789070870962805, "grad_norm": 1.0323535422961099, "learning_rate": 5.178959873175574e-06, "loss": 1.5825, "step": 7564 }, { "epoch": 0.5789836216133476, "grad_norm": 0.5186405231448915, "learning_rate": 5.177989502256375e-06, "loss": 1.0487, "step": 7565 }, { "epoch": 0.5790601561304148, "grad_norm": 0.6143511055540379, "learning_rate": 5.177019181843956e-06, "loss": 1.4275, "step": 7566 }, { "epoch": 0.579136690647482, "grad_norm": 0.5304449155576175, "learning_rate": 5.176048911997942e-06, "loss": 1.2097, "step": 7567 }, { "epoch": 0.5792132251645492, "grad_norm": 0.7217991542295914, "learning_rate": 5.175078692777944e-06, "loss": 1.4499, "step": 7568 }, { "epoch": 0.5792897596816164, "grad_norm": 0.620130171023785, "learning_rate": 5.174108524243579e-06, "loss": 1.55, "step": 7569 }, { "epoch": 0.5793662941986836, "grad_norm": 0.5959333428655237, "learning_rate": 5.173138406454453e-06, "loss": 1.3851, "step": 7570 }, { "epoch": 0.5794428287157508, "grad_norm": 0.5179927040044242, "learning_rate": 5.172168339470176e-06, "loss": 1.1394, "step": 7571 }, { "epoch": 0.579519363232818, "grad_norm": 0.6620041651742684, "learning_rate": 5.17119832335035e-06, "loss": 1.6823, "step": 7572 }, { "epoch": 0.5795958977498852, "grad_norm": 0.6928378719456726, "learning_rate": 5.170228358154577e-06, "loss": 1.586, "step": 7573 }, { "epoch": 0.5796724322669524, "grad_norm": 0.6360242824456563, "learning_rate": 5.169258443942455e-06, "loss": 1.6108, "step": 7574 }, { "epoch": 0.5797489667840195, "grad_norm": 0.5759313519212844, "learning_rate": 5.168288580773578e-06, "loss": 1.4044, "step": 7575 }, { "epoch": 0.5798255013010868, "grad_norm": 0.572752967014888, "learning_rate": 5.167318768707537e-06, "loss": 1.3608, "step": 7576 }, { "epoch": 0.579902035818154, "grad_norm": 0.5454327089202101, "learning_rate": 5.166349007803919e-06, "loss": 1.1856, "step": 7577 }, { "epoch": 0.5799785703352212, "grad_norm": 0.6005542619763543, "learning_rate": 5.1653792981223105e-06, "loss": 1.5603, "step": 7578 }, { "epoch": 0.5800551048522884, "grad_norm": 0.6199224006807587, "learning_rate": 5.164409639722293e-06, "loss": 1.0246, "step": 7579 }, { "epoch": 0.5801316393693556, "grad_norm": 0.581675421499051, "learning_rate": 5.163440032663446e-06, "loss": 1.3425, "step": 7580 }, { "epoch": 0.5802081738864228, "grad_norm": 0.5962468543566489, "learning_rate": 5.162470477005344e-06, "loss": 1.4858, "step": 7581 }, { "epoch": 0.58028470840349, "grad_norm": 0.526919474432129, "learning_rate": 5.16150097280756e-06, "loss": 1.0667, "step": 7582 }, { "epoch": 0.5803612429205571, "grad_norm": 0.6514580636331296, "learning_rate": 5.160531520129659e-06, "loss": 1.5602, "step": 7583 }, { "epoch": 0.5804377774376244, "grad_norm": 1.1934827776963293, "learning_rate": 5.159562119031215e-06, "loss": 1.477, "step": 7584 }, { "epoch": 0.5805143119546916, "grad_norm": 0.6039082454154227, "learning_rate": 5.158592769571787e-06, "loss": 1.4836, "step": 7585 }, { "epoch": 0.5805908464717587, "grad_norm": 0.7861550160617874, "learning_rate": 5.157623471810932e-06, "loss": 1.4294, "step": 7586 }, { "epoch": 0.580667380988826, "grad_norm": 1.1286525990582876, "learning_rate": 5.1566542258082104e-06, "loss": 1.6197, "step": 7587 }, { "epoch": 0.5807439155058931, "grad_norm": 0.6492036763621796, "learning_rate": 5.155685031623174e-06, "loss": 1.5847, "step": 7588 }, { "epoch": 0.5808204500229603, "grad_norm": 0.6199445719872384, "learning_rate": 5.1547158893153736e-06, "loss": 1.1386, "step": 7589 }, { "epoch": 0.5808969845400276, "grad_norm": 0.6538467457504485, "learning_rate": 5.153746798944357e-06, "loss": 1.6278, "step": 7590 }, { "epoch": 0.5809735190570947, "grad_norm": 0.5110222947821821, "learning_rate": 5.152777760569666e-06, "loss": 1.1034, "step": 7591 }, { "epoch": 0.581050053574162, "grad_norm": 0.5253228746383742, "learning_rate": 5.151808774250841e-06, "loss": 1.0911, "step": 7592 }, { "epoch": 0.5811265880912292, "grad_norm": 0.6494957197539079, "learning_rate": 5.150839840047422e-06, "loss": 1.6636, "step": 7593 }, { "epoch": 0.5812031226082963, "grad_norm": 0.6724711765919793, "learning_rate": 5.14987095801894e-06, "loss": 1.2899, "step": 7594 }, { "epoch": 0.5812796571253636, "grad_norm": 0.5473605661728805, "learning_rate": 5.14890212822493e-06, "loss": 1.2535, "step": 7595 }, { "epoch": 0.5813561916424307, "grad_norm": 0.8831018599608814, "learning_rate": 5.1479333507249166e-06, "loss": 1.7863, "step": 7596 }, { "epoch": 0.5814327261594979, "grad_norm": 0.5540267963061573, "learning_rate": 5.146964625578424e-06, "loss": 1.2285, "step": 7597 }, { "epoch": 0.5815092606765652, "grad_norm": 0.7598963849982246, "learning_rate": 5.1459959528449764e-06, "loss": 1.6051, "step": 7598 }, { "epoch": 0.5815857951936323, "grad_norm": 0.6402443287885337, "learning_rate": 5.14502733258409e-06, "loss": 1.622, "step": 7599 }, { "epoch": 0.5816623297106995, "grad_norm": 0.7836067241899082, "learning_rate": 5.14405876485528e-06, "loss": 1.7543, "step": 7600 }, { "epoch": 0.5817388642277668, "grad_norm": 0.5689218032620599, "learning_rate": 5.143090249718059e-06, "loss": 1.3528, "step": 7601 }, { "epoch": 0.5818153987448339, "grad_norm": 0.5275887141600393, "learning_rate": 5.142121787231935e-06, "loss": 1.1055, "step": 7602 }, { "epoch": 0.5818919332619011, "grad_norm": 0.6542149649243961, "learning_rate": 5.141153377456412e-06, "loss": 1.717, "step": 7603 }, { "epoch": 0.5819684677789683, "grad_norm": 0.6217219250924297, "learning_rate": 5.140185020450993e-06, "loss": 1.4706, "step": 7604 }, { "epoch": 0.5820450022960355, "grad_norm": 0.7462747892571904, "learning_rate": 5.139216716275176e-06, "loss": 1.6869, "step": 7605 }, { "epoch": 0.5821215368131027, "grad_norm": 0.7365209615733868, "learning_rate": 5.138248464988458e-06, "loss": 1.2875, "step": 7606 }, { "epoch": 0.5821980713301699, "grad_norm": 0.5874228860751765, "learning_rate": 5.137280266650331e-06, "loss": 1.4369, "step": 7607 }, { "epoch": 0.5822746058472371, "grad_norm": 1.6488387258745505, "learning_rate": 5.136312121320282e-06, "loss": 1.3237, "step": 7608 }, { "epoch": 0.5823511403643044, "grad_norm": 0.4850890206063963, "learning_rate": 5.135344029057801e-06, "loss": 1.0446, "step": 7609 }, { "epoch": 0.5824276748813715, "grad_norm": 0.5235370457938038, "learning_rate": 5.134375989922364e-06, "loss": 1.1718, "step": 7610 }, { "epoch": 0.5825042093984387, "grad_norm": 0.688166177490619, "learning_rate": 5.133408003973457e-06, "loss": 1.5502, "step": 7611 }, { "epoch": 0.5825807439155058, "grad_norm": 0.5065481728606908, "learning_rate": 5.132440071270551e-06, "loss": 1.1455, "step": 7612 }, { "epoch": 0.5826572784325731, "grad_norm": 0.5629727987372507, "learning_rate": 5.131472191873121e-06, "loss": 1.3083, "step": 7613 }, { "epoch": 0.5827338129496403, "grad_norm": 0.5510952956791565, "learning_rate": 5.130504365840637e-06, "loss": 1.2336, "step": 7614 }, { "epoch": 0.5828103474667075, "grad_norm": 0.9438831691738437, "learning_rate": 5.129536593232564e-06, "loss": 1.9712, "step": 7615 }, { "epoch": 0.5828868819837747, "grad_norm": 0.8617197994085364, "learning_rate": 5.128568874108362e-06, "loss": 1.5788, "step": 7616 }, { "epoch": 0.5829634165008418, "grad_norm": 0.5980499051938959, "learning_rate": 5.127601208527495e-06, "loss": 1.4796, "step": 7617 }, { "epoch": 0.5830399510179091, "grad_norm": 0.5853992533579474, "learning_rate": 5.12663359654942e-06, "loss": 1.4853, "step": 7618 }, { "epoch": 0.5831164855349763, "grad_norm": 0.570128376444344, "learning_rate": 5.125666038233585e-06, "loss": 1.3337, "step": 7619 }, { "epoch": 0.5831930200520434, "grad_norm": 0.683600499953042, "learning_rate": 5.124698533639444e-06, "loss": 1.5218, "step": 7620 }, { "epoch": 0.5832695545691107, "grad_norm": 0.5842660832927019, "learning_rate": 5.1237310828264395e-06, "loss": 1.3039, "step": 7621 }, { "epoch": 0.5833460890861779, "grad_norm": 0.7517749166801716, "learning_rate": 5.12276368585402e-06, "loss": 1.4038, "step": 7622 }, { "epoch": 0.583422623603245, "grad_norm": 0.5643122865005656, "learning_rate": 5.1217963427816194e-06, "loss": 1.3402, "step": 7623 }, { "epoch": 0.5834991581203123, "grad_norm": 0.5143002838867893, "learning_rate": 5.120829053668679e-06, "loss": 1.1376, "step": 7624 }, { "epoch": 0.5835756926373794, "grad_norm": 0.6904880492840243, "learning_rate": 5.119861818574628e-06, "loss": 1.133, "step": 7625 }, { "epoch": 0.5836522271544466, "grad_norm": 0.7771931945201835, "learning_rate": 5.1188946375589e-06, "loss": 1.3907, "step": 7626 }, { "epoch": 0.5837287616715139, "grad_norm": 0.5407448828107511, "learning_rate": 5.117927510680916e-06, "loss": 1.3862, "step": 7627 }, { "epoch": 0.583805296188581, "grad_norm": 0.9093420753808041, "learning_rate": 5.116960438000105e-06, "loss": 1.4855, "step": 7628 }, { "epoch": 0.5838818307056483, "grad_norm": 0.6035420240940177, "learning_rate": 5.115993419575884e-06, "loss": 1.5604, "step": 7629 }, { "epoch": 0.5839583652227155, "grad_norm": 0.4854455840351377, "learning_rate": 5.115026455467671e-06, "loss": 1.057, "step": 7630 }, { "epoch": 0.5840348997397826, "grad_norm": 0.6441533207175293, "learning_rate": 5.114059545734877e-06, "loss": 1.3892, "step": 7631 }, { "epoch": 0.5841114342568499, "grad_norm": 0.6752269093907023, "learning_rate": 5.113092690436911e-06, "loss": 1.5209, "step": 7632 }, { "epoch": 0.584187968773917, "grad_norm": 0.5991660056853192, "learning_rate": 5.112125889633185e-06, "loss": 1.3942, "step": 7633 }, { "epoch": 0.5842645032909842, "grad_norm": 0.8043651613861401, "learning_rate": 5.111159143383098e-06, "loss": 1.7291, "step": 7634 }, { "epoch": 0.5843410378080515, "grad_norm": 0.6336319660925052, "learning_rate": 5.11019245174605e-06, "loss": 1.5936, "step": 7635 }, { "epoch": 0.5844175723251186, "grad_norm": 0.4911885124087999, "learning_rate": 5.109225814781438e-06, "loss": 0.8988, "step": 7636 }, { "epoch": 0.5844941068421858, "grad_norm": 0.6975381522739186, "learning_rate": 5.108259232548655e-06, "loss": 1.5628, "step": 7637 }, { "epoch": 0.5845706413592531, "grad_norm": 2.575226615262767, "learning_rate": 5.107292705107087e-06, "loss": 1.6743, "step": 7638 }, { "epoch": 0.5846471758763202, "grad_norm": 0.5042128453616789, "learning_rate": 5.106326232516128e-06, "loss": 1.0094, "step": 7639 }, { "epoch": 0.5847237103933874, "grad_norm": 0.5808512402517141, "learning_rate": 5.105359814835155e-06, "loss": 1.3178, "step": 7640 }, { "epoch": 0.5848002449104546, "grad_norm": 0.5310451670615516, "learning_rate": 5.104393452123551e-06, "loss": 1.176, "step": 7641 }, { "epoch": 0.5848767794275218, "grad_norm": 0.6060017608512729, "learning_rate": 5.10342714444069e-06, "loss": 1.3497, "step": 7642 }, { "epoch": 0.584953313944589, "grad_norm": 0.5585305402718677, "learning_rate": 5.102460891845944e-06, "loss": 1.2498, "step": 7643 }, { "epoch": 0.5850298484616562, "grad_norm": 0.5613002320565922, "learning_rate": 5.1014946943986855e-06, "loss": 1.302, "step": 7644 }, { "epoch": 0.5851063829787234, "grad_norm": 0.534385316546142, "learning_rate": 5.100528552158278e-06, "loss": 1.2458, "step": 7645 }, { "epoch": 0.5851829174957907, "grad_norm": 0.5668269221006383, "learning_rate": 5.099562465184086e-06, "loss": 1.2292, "step": 7646 }, { "epoch": 0.5852594520128578, "grad_norm": 0.5811610467567605, "learning_rate": 5.098596433535468e-06, "loss": 1.3881, "step": 7647 }, { "epoch": 0.585335986529925, "grad_norm": 0.5676501707567803, "learning_rate": 5.097630457271781e-06, "loss": 1.3284, "step": 7648 }, { "epoch": 0.5854125210469922, "grad_norm": 1.4717742429703635, "learning_rate": 5.096664536452372e-06, "loss": 1.1447, "step": 7649 }, { "epoch": 0.5854890555640594, "grad_norm": 0.556225686836562, "learning_rate": 5.095698671136599e-06, "loss": 1.3785, "step": 7650 }, { "epoch": 0.5855655900811266, "grad_norm": 0.6047675015831194, "learning_rate": 5.094732861383801e-06, "loss": 1.5417, "step": 7651 }, { "epoch": 0.5856421245981938, "grad_norm": 0.8550825780665994, "learning_rate": 5.093767107253323e-06, "loss": 1.4143, "step": 7652 }, { "epoch": 0.585718659115261, "grad_norm": 0.7635376282373724, "learning_rate": 5.0928014088045015e-06, "loss": 1.401, "step": 7653 }, { "epoch": 0.5857951936323281, "grad_norm": 0.4864996308300847, "learning_rate": 5.091835766096673e-06, "loss": 1.1611, "step": 7654 }, { "epoch": 0.5858717281493954, "grad_norm": 0.5224962224798284, "learning_rate": 5.0908701791891714e-06, "loss": 1.1024, "step": 7655 }, { "epoch": 0.5859482626664626, "grad_norm": 0.7756025087934983, "learning_rate": 5.089904648141324e-06, "loss": 1.5796, "step": 7656 }, { "epoch": 0.5860247971835297, "grad_norm": 0.5427407234364822, "learning_rate": 5.088939173012456e-06, "loss": 1.2525, "step": 7657 }, { "epoch": 0.586101331700597, "grad_norm": 0.5889790586542157, "learning_rate": 5.087973753861888e-06, "loss": 1.5469, "step": 7658 }, { "epoch": 0.5861778662176642, "grad_norm": 0.7607989227969214, "learning_rate": 5.08700839074894e-06, "loss": 1.6635, "step": 7659 }, { "epoch": 0.5862544007347313, "grad_norm": 0.6277633055749293, "learning_rate": 5.086043083732923e-06, "loss": 1.5744, "step": 7660 }, { "epoch": 0.5863309352517986, "grad_norm": 0.6637145772439277, "learning_rate": 5.0850778328731545e-06, "loss": 1.6922, "step": 7661 }, { "epoch": 0.5864074697688657, "grad_norm": 0.43342072097079953, "learning_rate": 5.084112638228937e-06, "loss": 0.9499, "step": 7662 }, { "epoch": 0.586484004285933, "grad_norm": 0.775118143714623, "learning_rate": 5.083147499859578e-06, "loss": 1.3591, "step": 7663 }, { "epoch": 0.5865605388030002, "grad_norm": 0.8260884762783264, "learning_rate": 5.082182417824378e-06, "loss": 1.7288, "step": 7664 }, { "epoch": 0.5866370733200673, "grad_norm": 0.6208450394207262, "learning_rate": 5.0812173921826315e-06, "loss": 1.5349, "step": 7665 }, { "epoch": 0.5867136078371346, "grad_norm": 0.5991450695482086, "learning_rate": 5.080252422993636e-06, "loss": 1.2975, "step": 7666 }, { "epoch": 0.5867901423542018, "grad_norm": 0.5918511793969116, "learning_rate": 5.0792875103166815e-06, "loss": 1.4493, "step": 7667 }, { "epoch": 0.5868666768712689, "grad_norm": 0.5655156559801799, "learning_rate": 5.0783226542110564e-06, "loss": 1.025, "step": 7668 }, { "epoch": 0.5869432113883362, "grad_norm": 0.6265715733304393, "learning_rate": 5.077357854736042e-06, "loss": 1.4413, "step": 7669 }, { "epoch": 0.5870197459054033, "grad_norm": 0.5658598373007736, "learning_rate": 5.076393111950918e-06, "loss": 1.3751, "step": 7670 }, { "epoch": 0.5870962804224705, "grad_norm": 0.584184560101888, "learning_rate": 5.075428425914962e-06, "loss": 1.5453, "step": 7671 }, { "epoch": 0.5871728149395378, "grad_norm": 0.6737157187553268, "learning_rate": 5.074463796687449e-06, "loss": 1.5861, "step": 7672 }, { "epoch": 0.5872493494566049, "grad_norm": 0.6200669947429437, "learning_rate": 5.073499224327646e-06, "loss": 1.6796, "step": 7673 }, { "epoch": 0.5873258839736721, "grad_norm": 0.5443170938269125, "learning_rate": 5.072534708894821e-06, "loss": 1.3112, "step": 7674 }, { "epoch": 0.5874024184907394, "grad_norm": 0.5156541899661528, "learning_rate": 5.0715702504482364e-06, "loss": 1.1411, "step": 7675 }, { "epoch": 0.5874789530078065, "grad_norm": 0.47399406808618116, "learning_rate": 5.070605849047149e-06, "loss": 0.9722, "step": 7676 }, { "epoch": 0.5875554875248737, "grad_norm": 0.6031742148123456, "learning_rate": 5.069641504750819e-06, "loss": 1.5859, "step": 7677 }, { "epoch": 0.5876320220419409, "grad_norm": 0.8602861567618084, "learning_rate": 5.068677217618495e-06, "loss": 1.4754, "step": 7678 }, { "epoch": 0.5877085565590081, "grad_norm": 0.7470679128438643, "learning_rate": 5.067712987709427e-06, "loss": 1.4453, "step": 7679 }, { "epoch": 0.5877850910760754, "grad_norm": 0.6797257722971315, "learning_rate": 5.06674881508286e-06, "loss": 1.5555, "step": 7680 }, { "epoch": 0.5878616255931425, "grad_norm": 0.7357122112031714, "learning_rate": 5.065784699798037e-06, "loss": 1.6882, "step": 7681 }, { "epoch": 0.5879381601102097, "grad_norm": 0.5693267957596497, "learning_rate": 5.064820641914193e-06, "loss": 1.3446, "step": 7682 }, { "epoch": 0.588014694627277, "grad_norm": 0.6128028926880416, "learning_rate": 5.0638566414905656e-06, "loss": 1.333, "step": 7683 }, { "epoch": 0.5880912291443441, "grad_norm": 0.552291361938563, "learning_rate": 5.062892698586384e-06, "loss": 1.334, "step": 7684 }, { "epoch": 0.5881677636614113, "grad_norm": 0.7970730490149107, "learning_rate": 5.061928813260877e-06, "loss": 1.9742, "step": 7685 }, { "epoch": 0.5882442981784785, "grad_norm": 1.7460084093095862, "learning_rate": 5.0609649855732675e-06, "loss": 1.6862, "step": 7686 }, { "epoch": 0.5883208326955457, "grad_norm": 0.48196809364258486, "learning_rate": 5.060001215582776e-06, "loss": 1.0681, "step": 7687 }, { "epoch": 0.5883973672126129, "grad_norm": 0.9272954148406508, "learning_rate": 5.0590375033486205e-06, "loss": 1.4601, "step": 7688 }, { "epoch": 0.5884739017296801, "grad_norm": 0.509013435205675, "learning_rate": 5.058073848930014e-06, "loss": 1.1146, "step": 7689 }, { "epoch": 0.5885504362467473, "grad_norm": 1.1231118454476163, "learning_rate": 5.057110252386166e-06, "loss": 1.5183, "step": 7690 }, { "epoch": 0.5886269707638144, "grad_norm": 0.6572892393358626, "learning_rate": 5.056146713776282e-06, "loss": 1.4811, "step": 7691 }, { "epoch": 0.5887035052808817, "grad_norm": 0.5921038285940068, "learning_rate": 5.055183233159567e-06, "loss": 1.4077, "step": 7692 }, { "epoch": 0.5887800397979489, "grad_norm": 0.6277838628119229, "learning_rate": 5.054219810595216e-06, "loss": 1.6163, "step": 7693 }, { "epoch": 0.588856574315016, "grad_norm": 0.5568121879488912, "learning_rate": 5.0532564461424304e-06, "loss": 1.1312, "step": 7694 }, { "epoch": 0.5889331088320833, "grad_norm": 0.4984139963191408, "learning_rate": 5.052293139860398e-06, "loss": 1.0655, "step": 7695 }, { "epoch": 0.5890096433491505, "grad_norm": 0.9144900881393433, "learning_rate": 5.051329891808308e-06, "loss": 1.9567, "step": 7696 }, { "epoch": 0.5890861778662176, "grad_norm": 0.6375503855658681, "learning_rate": 5.050366702045344e-06, "loss": 1.5643, "step": 7697 }, { "epoch": 0.5891627123832849, "grad_norm": 0.6169618379425271, "learning_rate": 5.049403570630688e-06, "loss": 1.5529, "step": 7698 }, { "epoch": 0.589239246900352, "grad_norm": 0.5317038968130947, "learning_rate": 5.048440497623521e-06, "loss": 1.2626, "step": 7699 }, { "epoch": 0.5893157814174192, "grad_norm": 0.70892358808841, "learning_rate": 5.0474774830830144e-06, "loss": 1.4782, "step": 7700 }, { "epoch": 0.5893923159344865, "grad_norm": 0.5203678311041152, "learning_rate": 5.046514527068338e-06, "loss": 1.0018, "step": 7701 }, { "epoch": 0.5894688504515536, "grad_norm": 0.7239912420847698, "learning_rate": 5.04555162963866e-06, "loss": 1.3944, "step": 7702 }, { "epoch": 0.5895453849686209, "grad_norm": 0.5307745378363301, "learning_rate": 5.044588790853142e-06, "loss": 1.0871, "step": 7703 }, { "epoch": 0.5896219194856881, "grad_norm": 0.700659070046867, "learning_rate": 5.043626010770944e-06, "loss": 1.4598, "step": 7704 }, { "epoch": 0.5896984540027552, "grad_norm": 0.5615081562275491, "learning_rate": 5.042663289451225e-06, "loss": 1.3762, "step": 7705 }, { "epoch": 0.5897749885198225, "grad_norm": 0.7730540977748944, "learning_rate": 5.0417006269531356e-06, "loss": 1.8459, "step": 7706 }, { "epoch": 0.5898515230368896, "grad_norm": 0.5604337895974821, "learning_rate": 5.0407380233358226e-06, "loss": 1.2674, "step": 7707 }, { "epoch": 0.5899280575539568, "grad_norm": 0.48353370104667154, "learning_rate": 5.039775478658435e-06, "loss": 1.0002, "step": 7708 }, { "epoch": 0.5900045920710241, "grad_norm": 0.5106006752558604, "learning_rate": 5.03881299298011e-06, "loss": 1.1022, "step": 7709 }, { "epoch": 0.5900811265880912, "grad_norm": 0.6031364453213033, "learning_rate": 5.03785056635999e-06, "loss": 1.2196, "step": 7710 }, { "epoch": 0.5901576611051584, "grad_norm": 0.6144476426803218, "learning_rate": 5.036888198857207e-06, "loss": 1.3408, "step": 7711 }, { "epoch": 0.5902341956222257, "grad_norm": 0.6716686508710618, "learning_rate": 5.035925890530892e-06, "loss": 1.6264, "step": 7712 }, { "epoch": 0.5903107301392928, "grad_norm": 0.5638866584935888, "learning_rate": 5.034963641440174e-06, "loss": 1.1246, "step": 7713 }, { "epoch": 0.59038726465636, "grad_norm": 0.4842732819465424, "learning_rate": 5.034001451644173e-06, "loss": 1.0472, "step": 7714 }, { "epoch": 0.5904637991734272, "grad_norm": 0.6023473563141183, "learning_rate": 5.03303932120201e-06, "loss": 1.4591, "step": 7715 }, { "epoch": 0.5905403336904944, "grad_norm": 0.49967547578534866, "learning_rate": 5.032077250172803e-06, "loss": 1.1614, "step": 7716 }, { "epoch": 0.5906168682075617, "grad_norm": 0.6450959464859238, "learning_rate": 5.031115238615664e-06, "loss": 1.52, "step": 7717 }, { "epoch": 0.5906934027246288, "grad_norm": 0.7666179368173613, "learning_rate": 5.0301532865897e-06, "loss": 1.2917, "step": 7718 }, { "epoch": 0.590769937241696, "grad_norm": 0.6113841003490282, "learning_rate": 5.0291913941540185e-06, "loss": 1.2643, "step": 7719 }, { "epoch": 0.5908464717587631, "grad_norm": 0.4994544776477872, "learning_rate": 5.028229561367717e-06, "loss": 1.1242, "step": 7720 }, { "epoch": 0.5909230062758304, "grad_norm": 0.5488303589056458, "learning_rate": 5.027267788289898e-06, "loss": 1.3353, "step": 7721 }, { "epoch": 0.5909995407928976, "grad_norm": 0.6589393277684261, "learning_rate": 5.026306074979654e-06, "loss": 1.277, "step": 7722 }, { "epoch": 0.5910760753099648, "grad_norm": 0.6490140303642672, "learning_rate": 5.025344421496077e-06, "loss": 1.5096, "step": 7723 }, { "epoch": 0.591152609827032, "grad_norm": 0.5236857025360463, "learning_rate": 5.024382827898251e-06, "loss": 1.3036, "step": 7724 }, { "epoch": 0.5912291443440992, "grad_norm": 0.5980657550188658, "learning_rate": 5.0234212942452605e-06, "loss": 1.4585, "step": 7725 }, { "epoch": 0.5913056788611664, "grad_norm": 0.5717137844150071, "learning_rate": 5.022459820596186e-06, "loss": 1.41, "step": 7726 }, { "epoch": 0.5913822133782336, "grad_norm": 0.5789887408251333, "learning_rate": 5.021498407010102e-06, "loss": 1.459, "step": 7727 }, { "epoch": 0.5914587478953007, "grad_norm": 0.5412881370868409, "learning_rate": 5.0205370535460805e-06, "loss": 1.4695, "step": 7728 }, { "epoch": 0.591535282412368, "grad_norm": 0.6808913678520232, "learning_rate": 5.019575760263191e-06, "loss": 1.3761, "step": 7729 }, { "epoch": 0.5916118169294352, "grad_norm": 0.6552474922565825, "learning_rate": 5.0186145272204975e-06, "loss": 1.5844, "step": 7730 }, { "epoch": 0.5916883514465023, "grad_norm": 0.6081840859463931, "learning_rate": 5.01765335447706e-06, "loss": 1.4861, "step": 7731 }, { "epoch": 0.5917648859635696, "grad_norm": 0.5844848774655566, "learning_rate": 5.016692242091938e-06, "loss": 1.2764, "step": 7732 }, { "epoch": 0.5918414204806368, "grad_norm": 0.560217468580114, "learning_rate": 5.015731190124186e-06, "loss": 1.2482, "step": 7733 }, { "epoch": 0.5919179549977039, "grad_norm": 0.5121027608757281, "learning_rate": 5.01477019863285e-06, "loss": 0.9852, "step": 7734 }, { "epoch": 0.5919944895147712, "grad_norm": 0.6443142616536819, "learning_rate": 5.013809267676979e-06, "loss": 1.6381, "step": 7735 }, { "epoch": 0.5920710240318383, "grad_norm": 0.5840567463548032, "learning_rate": 5.012848397315614e-06, "loss": 1.3501, "step": 7736 }, { "epoch": 0.5921475585489056, "grad_norm": 0.9252773362034589, "learning_rate": 5.0118875876077955e-06, "loss": 1.5066, "step": 7737 }, { "epoch": 0.5922240930659728, "grad_norm": 0.5929089138343491, "learning_rate": 5.010926838612557e-06, "loss": 1.3059, "step": 7738 }, { "epoch": 0.5923006275830399, "grad_norm": 0.7100834049110792, "learning_rate": 5.009966150388931e-06, "loss": 1.6287, "step": 7739 }, { "epoch": 0.5923771621001072, "grad_norm": 0.609649815827981, "learning_rate": 5.009005522995944e-06, "loss": 1.6364, "step": 7740 }, { "epoch": 0.5924536966171744, "grad_norm": 0.7817867456897802, "learning_rate": 5.008044956492619e-06, "loss": 1.8716, "step": 7741 }, { "epoch": 0.5925302311342415, "grad_norm": 0.5463108601906634, "learning_rate": 5.007084450937978e-06, "loss": 1.355, "step": 7742 }, { "epoch": 0.5926067656513088, "grad_norm": 0.5812692505839214, "learning_rate": 5.0061240063910364e-06, "loss": 1.4013, "step": 7743 }, { "epoch": 0.5926833001683759, "grad_norm": 0.9954499044651913, "learning_rate": 5.005163622910808e-06, "loss": 1.6209, "step": 7744 }, { "epoch": 0.5927598346854431, "grad_norm": 0.5866019734476434, "learning_rate": 5.0042033005563e-06, "loss": 1.2992, "step": 7745 }, { "epoch": 0.5928363692025104, "grad_norm": 0.5600552205129485, "learning_rate": 5.003243039386519e-06, "loss": 1.3699, "step": 7746 }, { "epoch": 0.5929129037195775, "grad_norm": 0.5883214578729423, "learning_rate": 5.002282839460463e-06, "loss": 1.3329, "step": 7747 }, { "epoch": 0.5929894382366447, "grad_norm": 0.6866107321952394, "learning_rate": 5.001322700837133e-06, "loss": 1.6289, "step": 7748 }, { "epoch": 0.593065972753712, "grad_norm": 0.7332960040783544, "learning_rate": 5.000362623575523e-06, "loss": 1.8874, "step": 7749 }, { "epoch": 0.5931425072707791, "grad_norm": 0.4934504897486304, "learning_rate": 4.999402607734621e-06, "loss": 1.1112, "step": 7750 }, { "epoch": 0.5932190417878463, "grad_norm": 0.5118652835402622, "learning_rate": 4.998442653373414e-06, "loss": 1.116, "step": 7751 }, { "epoch": 0.5932955763049135, "grad_norm": 0.598593725978676, "learning_rate": 4.997482760550885e-06, "loss": 1.4575, "step": 7752 }, { "epoch": 0.5933721108219807, "grad_norm": 0.6097546605017562, "learning_rate": 4.996522929326009e-06, "loss": 1.4446, "step": 7753 }, { "epoch": 0.593448645339048, "grad_norm": 0.6863045226330511, "learning_rate": 4.995563159757767e-06, "loss": 1.5021, "step": 7754 }, { "epoch": 0.5935251798561151, "grad_norm": 0.5717869996006005, "learning_rate": 4.994603451905127e-06, "loss": 1.4271, "step": 7755 }, { "epoch": 0.5936017143731823, "grad_norm": 0.5489956018633049, "learning_rate": 4.993643805827057e-06, "loss": 1.3188, "step": 7756 }, { "epoch": 0.5936782488902494, "grad_norm": 0.49567088026142564, "learning_rate": 4.992684221582519e-06, "loss": 0.975, "step": 7757 }, { "epoch": 0.5937547834073167, "grad_norm": 0.5215627082311453, "learning_rate": 4.991724699230473e-06, "loss": 1.3291, "step": 7758 }, { "epoch": 0.5938313179243839, "grad_norm": 0.663500972551561, "learning_rate": 4.990765238829876e-06, "loss": 1.4032, "step": 7759 }, { "epoch": 0.5939078524414511, "grad_norm": 0.5830404915008566, "learning_rate": 4.989805840439681e-06, "loss": 1.2652, "step": 7760 }, { "epoch": 0.5939843869585183, "grad_norm": 0.5988727584985623, "learning_rate": 4.988846504118835e-06, "loss": 1.5368, "step": 7761 }, { "epoch": 0.5940609214755855, "grad_norm": 0.7029451247147118, "learning_rate": 4.9878872299262825e-06, "loss": 1.5125, "step": 7762 }, { "epoch": 0.5941374559926527, "grad_norm": 0.9389390180822479, "learning_rate": 4.986928017920964e-06, "loss": 1.6493, "step": 7763 }, { "epoch": 0.5942139905097199, "grad_norm": 0.6208550366032527, "learning_rate": 4.985968868161815e-06, "loss": 1.5308, "step": 7764 }, { "epoch": 0.594290525026787, "grad_norm": 0.6587957743740387, "learning_rate": 4.985009780707771e-06, "loss": 1.7496, "step": 7765 }, { "epoch": 0.5943670595438543, "grad_norm": 0.5505447156272937, "learning_rate": 4.984050755617761e-06, "loss": 1.3636, "step": 7766 }, { "epoch": 0.5944435940609215, "grad_norm": 0.6069229348392926, "learning_rate": 4.9830917929507094e-06, "loss": 1.4889, "step": 7767 }, { "epoch": 0.5945201285779886, "grad_norm": 0.5327913468875836, "learning_rate": 4.982132892765539e-06, "loss": 1.2207, "step": 7768 }, { "epoch": 0.5945966630950559, "grad_norm": 1.1829711949611679, "learning_rate": 4.981174055121164e-06, "loss": 1.6987, "step": 7769 }, { "epoch": 0.5946731976121231, "grad_norm": 0.633886605218558, "learning_rate": 4.980215280076502e-06, "loss": 1.498, "step": 7770 }, { "epoch": 0.5947497321291902, "grad_norm": 1.1232177349264658, "learning_rate": 4.979256567690463e-06, "loss": 1.306, "step": 7771 }, { "epoch": 0.5948262666462575, "grad_norm": 0.476805400148577, "learning_rate": 4.9782979180219506e-06, "loss": 1.1263, "step": 7772 }, { "epoch": 0.5949028011633246, "grad_norm": 0.5404671527934516, "learning_rate": 4.977339331129869e-06, "loss": 1.207, "step": 7773 }, { "epoch": 0.5949793356803919, "grad_norm": 0.49787882577881065, "learning_rate": 4.976380807073116e-06, "loss": 1.0039, "step": 7774 }, { "epoch": 0.5950558701974591, "grad_norm": 0.7781376432006805, "learning_rate": 4.975422345910586e-06, "loss": 1.4504, "step": 7775 }, { "epoch": 0.5951324047145262, "grad_norm": 0.5737223949900352, "learning_rate": 4.974463947701169e-06, "loss": 1.4649, "step": 7776 }, { "epoch": 0.5952089392315935, "grad_norm": 0.5804238495136277, "learning_rate": 4.973505612503754e-06, "loss": 1.3845, "step": 7777 }, { "epoch": 0.5952854737486607, "grad_norm": 0.5022839947573172, "learning_rate": 4.972547340377223e-06, "loss": 1.0505, "step": 7778 }, { "epoch": 0.5953620082657278, "grad_norm": 0.5805993827723949, "learning_rate": 4.9715891313804555e-06, "loss": 1.2048, "step": 7779 }, { "epoch": 0.5954385427827951, "grad_norm": 0.539939185456657, "learning_rate": 4.9706309855723244e-06, "loss": 1.4789, "step": 7780 }, { "epoch": 0.5955150772998622, "grad_norm": 0.5826714257621498, "learning_rate": 4.969672903011703e-06, "loss": 1.2229, "step": 7781 }, { "epoch": 0.5955916118169294, "grad_norm": 0.5678085589996927, "learning_rate": 4.9687148837574595e-06, "loss": 1.3367, "step": 7782 }, { "epoch": 0.5956681463339967, "grad_norm": 0.5270565505236362, "learning_rate": 4.967756927868456e-06, "loss": 1.094, "step": 7783 }, { "epoch": 0.5957446808510638, "grad_norm": 0.6374976641993154, "learning_rate": 4.966799035403553e-06, "loss": 1.4936, "step": 7784 }, { "epoch": 0.595821215368131, "grad_norm": 0.8450675362969419, "learning_rate": 4.965841206421607e-06, "loss": 1.4479, "step": 7785 }, { "epoch": 0.5958977498851982, "grad_norm": 0.5822195515365494, "learning_rate": 4.964883440981465e-06, "loss": 1.2201, "step": 7786 }, { "epoch": 0.5959742844022654, "grad_norm": 0.8879794863181841, "learning_rate": 4.9639257391419825e-06, "loss": 1.5806, "step": 7787 }, { "epoch": 0.5960508189193326, "grad_norm": 0.5920854174676616, "learning_rate": 4.962968100961999e-06, "loss": 1.0874, "step": 7788 }, { "epoch": 0.5961273534363998, "grad_norm": 0.553188080221915, "learning_rate": 4.962010526500355e-06, "loss": 1.3739, "step": 7789 }, { "epoch": 0.596203887953467, "grad_norm": 0.5713070983578294, "learning_rate": 4.961053015815888e-06, "loss": 1.2101, "step": 7790 }, { "epoch": 0.5962804224705343, "grad_norm": 0.5362720088250142, "learning_rate": 4.960095568967428e-06, "loss": 1.1308, "step": 7791 }, { "epoch": 0.5963569569876014, "grad_norm": 0.5568206471634017, "learning_rate": 4.9591381860138065e-06, "loss": 1.3475, "step": 7792 }, { "epoch": 0.5964334915046686, "grad_norm": 0.5274354734897261, "learning_rate": 4.958180867013847e-06, "loss": 1.195, "step": 7793 }, { "epoch": 0.5965100260217358, "grad_norm": 0.7442357118724234, "learning_rate": 4.9572236120263675e-06, "loss": 1.3506, "step": 7794 }, { "epoch": 0.596586560538803, "grad_norm": 0.6246619873532642, "learning_rate": 4.956266421110188e-06, "loss": 1.6586, "step": 7795 }, { "epoch": 0.5966630950558702, "grad_norm": 0.8429131627489771, "learning_rate": 4.9553092943241195e-06, "loss": 1.9646, "step": 7796 }, { "epoch": 0.5967396295729374, "grad_norm": 0.6128626867440681, "learning_rate": 4.95435223172697e-06, "loss": 1.4093, "step": 7797 }, { "epoch": 0.5968161640900046, "grad_norm": 0.5181837434385023, "learning_rate": 4.953395233377546e-06, "loss": 1.1195, "step": 7798 }, { "epoch": 0.5968926986070718, "grad_norm": 0.6227561927211488, "learning_rate": 4.952438299334648e-06, "loss": 1.4814, "step": 7799 }, { "epoch": 0.596969233124139, "grad_norm": 0.5930053089908593, "learning_rate": 4.9514814296570725e-06, "loss": 1.4312, "step": 7800 }, { "epoch": 0.5970457676412062, "grad_norm": 0.7106377426244009, "learning_rate": 4.950524624403612e-06, "loss": 1.386, "step": 7801 }, { "epoch": 0.5971223021582733, "grad_norm": 0.6665138014238112, "learning_rate": 4.949567883633055e-06, "loss": 1.1367, "step": 7802 }, { "epoch": 0.5971988366753406, "grad_norm": 0.5668132261897494, "learning_rate": 4.948611207404189e-06, "loss": 1.3816, "step": 7803 }, { "epoch": 0.5972753711924078, "grad_norm": 0.6770491456434946, "learning_rate": 4.947654595775793e-06, "loss": 1.549, "step": 7804 }, { "epoch": 0.5973519057094749, "grad_norm": 0.528851385322461, "learning_rate": 4.946698048806644e-06, "loss": 1.1676, "step": 7805 }, { "epoch": 0.5974284402265422, "grad_norm": 0.6549488377021316, "learning_rate": 4.9457415665555165e-06, "loss": 1.4336, "step": 7806 }, { "epoch": 0.5975049747436094, "grad_norm": 0.6138906245367166, "learning_rate": 4.9447851490811795e-06, "loss": 1.5478, "step": 7807 }, { "epoch": 0.5975815092606765, "grad_norm": 0.5770428296867882, "learning_rate": 4.943828796442393e-06, "loss": 1.4212, "step": 7808 }, { "epoch": 0.5976580437777438, "grad_norm": 0.5819041500715445, "learning_rate": 4.942872508697927e-06, "loss": 1.4538, "step": 7809 }, { "epoch": 0.5977345782948109, "grad_norm": 0.543552852133287, "learning_rate": 4.941916285906534e-06, "loss": 1.1809, "step": 7810 }, { "epoch": 0.5978111128118782, "grad_norm": 0.691046205262209, "learning_rate": 4.940960128126967e-06, "loss": 1.773, "step": 7811 }, { "epoch": 0.5978876473289454, "grad_norm": 0.5753617313583896, "learning_rate": 4.940004035417975e-06, "loss": 1.3633, "step": 7812 }, { "epoch": 0.5979641818460125, "grad_norm": 0.622803296626733, "learning_rate": 4.939048007838303e-06, "loss": 1.4632, "step": 7813 }, { "epoch": 0.5980407163630798, "grad_norm": 0.46492288620242717, "learning_rate": 4.9380920454466956e-06, "loss": 0.975, "step": 7814 }, { "epoch": 0.598117250880147, "grad_norm": 0.5625286274789084, "learning_rate": 4.937136148301885e-06, "loss": 1.2995, "step": 7815 }, { "epoch": 0.5981937853972141, "grad_norm": 0.5632150035839734, "learning_rate": 4.936180316462608e-06, "loss": 1.3584, "step": 7816 }, { "epoch": 0.5982703199142814, "grad_norm": 0.5496487779673802, "learning_rate": 4.935224549987594e-06, "loss": 1.2917, "step": 7817 }, { "epoch": 0.5983468544313485, "grad_norm": 0.6105390955118433, "learning_rate": 4.934268848935565e-06, "loss": 1.4931, "step": 7818 }, { "epoch": 0.5984233889484157, "grad_norm": 0.5958215783776536, "learning_rate": 4.9333132133652425e-06, "loss": 1.4727, "step": 7819 }, { "epoch": 0.598499923465483, "grad_norm": 0.5145735960877842, "learning_rate": 4.932357643335346e-06, "loss": 1.3136, "step": 7820 }, { "epoch": 0.5985764579825501, "grad_norm": 0.580478842042107, "learning_rate": 4.931402138904589e-06, "loss": 1.3925, "step": 7821 }, { "epoch": 0.5986529924996173, "grad_norm": 0.6734643676866241, "learning_rate": 4.930446700131678e-06, "loss": 1.4789, "step": 7822 }, { "epoch": 0.5987295270166845, "grad_norm": 0.5769797935013511, "learning_rate": 4.9294913270753186e-06, "loss": 1.202, "step": 7823 }, { "epoch": 0.5988060615337517, "grad_norm": 0.6363333742153037, "learning_rate": 4.928536019794213e-06, "loss": 1.2644, "step": 7824 }, { "epoch": 0.598882596050819, "grad_norm": 0.6264130370046083, "learning_rate": 4.927580778347058e-06, "loss": 1.5155, "step": 7825 }, { "epoch": 0.5989591305678861, "grad_norm": 0.4991166850866014, "learning_rate": 4.926625602792545e-06, "loss": 1.2476, "step": 7826 }, { "epoch": 0.5990356650849533, "grad_norm": 0.5756278244880537, "learning_rate": 4.925670493189364e-06, "loss": 1.277, "step": 7827 }, { "epoch": 0.5991121996020206, "grad_norm": 0.5328304038738375, "learning_rate": 4.9247154495961995e-06, "loss": 1.1789, "step": 7828 }, { "epoch": 0.5991887341190877, "grad_norm": 0.8182018789361926, "learning_rate": 4.923760472071733e-06, "loss": 1.3554, "step": 7829 }, { "epoch": 0.5992652686361549, "grad_norm": 0.6868591908796311, "learning_rate": 4.9228055606746376e-06, "loss": 1.3364, "step": 7830 }, { "epoch": 0.599341803153222, "grad_norm": 0.6881618260159438, "learning_rate": 4.921850715463591e-06, "loss": 1.6958, "step": 7831 }, { "epoch": 0.5994183376702893, "grad_norm": 0.5478255502165934, "learning_rate": 4.9208959364972594e-06, "loss": 1.2159, "step": 7832 }, { "epoch": 0.5994948721873565, "grad_norm": 0.770880613114982, "learning_rate": 4.9199412238343065e-06, "loss": 1.5165, "step": 7833 }, { "epoch": 0.5995714067044237, "grad_norm": 0.6046546912507464, "learning_rate": 4.918986577533396e-06, "loss": 1.3446, "step": 7834 }, { "epoch": 0.5996479412214909, "grad_norm": 0.6248156533057365, "learning_rate": 4.9180319976531785e-06, "loss": 1.4764, "step": 7835 }, { "epoch": 0.5997244757385581, "grad_norm": 0.581769444021748, "learning_rate": 4.917077484252312e-06, "loss": 1.3633, "step": 7836 }, { "epoch": 0.5998010102556253, "grad_norm": 0.5380758080696342, "learning_rate": 4.916123037389441e-06, "loss": 1.2303, "step": 7837 }, { "epoch": 0.5998775447726925, "grad_norm": 0.9884269455553673, "learning_rate": 4.915168657123211e-06, "loss": 1.63, "step": 7838 }, { "epoch": 0.5999540792897596, "grad_norm": 0.5622602906681662, "learning_rate": 4.914214343512262e-06, "loss": 1.2503, "step": 7839 }, { "epoch": 0.6000306138068269, "grad_norm": 0.5292099568718543, "learning_rate": 4.91326009661523e-06, "loss": 1.1889, "step": 7840 }, { "epoch": 0.6001071483238941, "grad_norm": 0.6953846323169837, "learning_rate": 4.912305916490744e-06, "loss": 1.6317, "step": 7841 }, { "epoch": 0.6001836828409612, "grad_norm": 0.6595077558039754, "learning_rate": 4.911351803197437e-06, "loss": 1.645, "step": 7842 }, { "epoch": 0.6002602173580285, "grad_norm": 0.6130767225131277, "learning_rate": 4.910397756793928e-06, "loss": 1.3458, "step": 7843 }, { "epoch": 0.6003367518750957, "grad_norm": 0.5760874430631874, "learning_rate": 4.909443777338839e-06, "loss": 1.5037, "step": 7844 }, { "epoch": 0.6004132863921628, "grad_norm": 1.0530041932387746, "learning_rate": 4.908489864890784e-06, "loss": 1.1111, "step": 7845 }, { "epoch": 0.6004898209092301, "grad_norm": 0.7113975545509965, "learning_rate": 4.907536019508374e-06, "loss": 1.3861, "step": 7846 }, { "epoch": 0.6005663554262972, "grad_norm": 0.5564745838380035, "learning_rate": 4.906582241250219e-06, "loss": 1.338, "step": 7847 }, { "epoch": 0.6006428899433645, "grad_norm": 0.5437546772645397, "learning_rate": 4.905628530174918e-06, "loss": 1.269, "step": 7848 }, { "epoch": 0.6007194244604317, "grad_norm": 0.5189432686452319, "learning_rate": 4.904674886341072e-06, "loss": 1.1887, "step": 7849 }, { "epoch": 0.6007959589774988, "grad_norm": 0.524859577723372, "learning_rate": 4.903721309807277e-06, "loss": 1.1103, "step": 7850 }, { "epoch": 0.6008724934945661, "grad_norm": 0.6042422134910767, "learning_rate": 4.90276780063212e-06, "loss": 1.3797, "step": 7851 }, { "epoch": 0.6009490280116333, "grad_norm": 0.5913087485070301, "learning_rate": 4.901814358874189e-06, "loss": 1.4146, "step": 7852 }, { "epoch": 0.6010255625287004, "grad_norm": 0.6021322791759156, "learning_rate": 4.900860984592067e-06, "loss": 1.3646, "step": 7853 }, { "epoch": 0.6011020970457677, "grad_norm": 0.6315187333681511, "learning_rate": 4.8999076778443335e-06, "loss": 1.4181, "step": 7854 }, { "epoch": 0.6011786315628348, "grad_norm": 0.5871411016750318, "learning_rate": 4.8989544386895595e-06, "loss": 1.3754, "step": 7855 }, { "epoch": 0.601255166079902, "grad_norm": 0.5145969071475407, "learning_rate": 4.898001267186317e-06, "loss": 1.0429, "step": 7856 }, { "epoch": 0.6013317005969693, "grad_norm": 0.6186985343229331, "learning_rate": 4.897048163393169e-06, "loss": 1.4866, "step": 7857 }, { "epoch": 0.6014082351140364, "grad_norm": 0.5754079282863855, "learning_rate": 4.896095127368681e-06, "loss": 1.4535, "step": 7858 }, { "epoch": 0.6014847696311036, "grad_norm": 0.6348742870738423, "learning_rate": 4.895142159171409e-06, "loss": 1.3428, "step": 7859 }, { "epoch": 0.6015613041481708, "grad_norm": 0.5779627418665833, "learning_rate": 4.894189258859905e-06, "loss": 1.3755, "step": 7860 }, { "epoch": 0.601637838665238, "grad_norm": 0.6379792525921346, "learning_rate": 4.893236426492719e-06, "loss": 1.4747, "step": 7861 }, { "epoch": 0.6017143731823053, "grad_norm": 0.6511185490754109, "learning_rate": 4.892283662128393e-06, "loss": 1.5453, "step": 7862 }, { "epoch": 0.6017909076993724, "grad_norm": 0.9811497270251379, "learning_rate": 4.891330965825474e-06, "loss": 1.5782, "step": 7863 }, { "epoch": 0.6018674422164396, "grad_norm": 0.5712486368559268, "learning_rate": 4.890378337642492e-06, "loss": 1.2823, "step": 7864 }, { "epoch": 0.6019439767335069, "grad_norm": 0.9115526373548095, "learning_rate": 4.8894257776379835e-06, "loss": 1.427, "step": 7865 }, { "epoch": 0.602020511250574, "grad_norm": 0.6857612826764892, "learning_rate": 4.888473285870476e-06, "loss": 1.4662, "step": 7866 }, { "epoch": 0.6020970457676412, "grad_norm": 0.6230247641138293, "learning_rate": 4.887520862398492e-06, "loss": 1.5291, "step": 7867 }, { "epoch": 0.6021735802847084, "grad_norm": 0.5031902804655612, "learning_rate": 4.886568507280549e-06, "loss": 1.2113, "step": 7868 }, { "epoch": 0.6022501148017756, "grad_norm": 0.6392306134058772, "learning_rate": 4.885616220575168e-06, "loss": 1.477, "step": 7869 }, { "epoch": 0.6023266493188428, "grad_norm": 0.6379583815295262, "learning_rate": 4.8846640023408576e-06, "loss": 1.6257, "step": 7870 }, { "epoch": 0.60240318383591, "grad_norm": 0.5608303628933702, "learning_rate": 4.883711852636125e-06, "loss": 1.3103, "step": 7871 }, { "epoch": 0.6024797183529772, "grad_norm": 0.582575187608017, "learning_rate": 4.882759771519474e-06, "loss": 1.3754, "step": 7872 }, { "epoch": 0.6025562528700444, "grad_norm": 0.49145904395107454, "learning_rate": 4.881807759049401e-06, "loss": 0.9492, "step": 7873 }, { "epoch": 0.6026327873871116, "grad_norm": 0.6802966550079442, "learning_rate": 4.880855815284403e-06, "loss": 1.9364, "step": 7874 }, { "epoch": 0.6027093219041788, "grad_norm": 0.6390413114760689, "learning_rate": 4.87990394028297e-06, "loss": 1.7299, "step": 7875 }, { "epoch": 0.6027858564212459, "grad_norm": 0.614795942996442, "learning_rate": 4.878952134103586e-06, "loss": 1.4247, "step": 7876 }, { "epoch": 0.6028623909383132, "grad_norm": 0.5093275199745689, "learning_rate": 4.878000396804735e-06, "loss": 1.2002, "step": 7877 }, { "epoch": 0.6029389254553804, "grad_norm": 0.5523600695845301, "learning_rate": 4.8770487284448945e-06, "loss": 1.1604, "step": 7878 }, { "epoch": 0.6030154599724475, "grad_norm": 0.5203762251623434, "learning_rate": 4.876097129082534e-06, "loss": 1.1857, "step": 7879 }, { "epoch": 0.6030919944895148, "grad_norm": 0.6339293886466555, "learning_rate": 4.875145598776128e-06, "loss": 1.5281, "step": 7880 }, { "epoch": 0.603168529006582, "grad_norm": 0.6979635550723229, "learning_rate": 4.874194137584139e-06, "loss": 1.5522, "step": 7881 }, { "epoch": 0.6032450635236491, "grad_norm": 0.5272762992575213, "learning_rate": 4.8732427455650274e-06, "loss": 1.0736, "step": 7882 }, { "epoch": 0.6033215980407164, "grad_norm": 0.556194548459857, "learning_rate": 4.872291422777251e-06, "loss": 1.334, "step": 7883 }, { "epoch": 0.6033981325577835, "grad_norm": 0.5468403149019875, "learning_rate": 4.87134016927926e-06, "loss": 1.3747, "step": 7884 }, { "epoch": 0.6034746670748508, "grad_norm": 0.6590340039913551, "learning_rate": 4.870388985129503e-06, "loss": 1.6538, "step": 7885 }, { "epoch": 0.603551201591918, "grad_norm": 0.5662390132648233, "learning_rate": 4.869437870386423e-06, "loss": 1.2132, "step": 7886 }, { "epoch": 0.6036277361089851, "grad_norm": 0.6240930573209246, "learning_rate": 4.868486825108462e-06, "loss": 1.3853, "step": 7887 }, { "epoch": 0.6037042706260524, "grad_norm": 0.8740297505228733, "learning_rate": 4.867535849354052e-06, "loss": 1.3527, "step": 7888 }, { "epoch": 0.6037808051431195, "grad_norm": 0.49151671219246507, "learning_rate": 4.8665849431816244e-06, "loss": 1.0173, "step": 7889 }, { "epoch": 0.6038573396601867, "grad_norm": 0.9176832936881829, "learning_rate": 4.8656341066496034e-06, "loss": 1.7809, "step": 7890 }, { "epoch": 0.603933874177254, "grad_norm": 0.8268765750087923, "learning_rate": 4.864683339816419e-06, "loss": 1.5955, "step": 7891 }, { "epoch": 0.6040104086943211, "grad_norm": 0.6071850524915663, "learning_rate": 4.8637326427404814e-06, "loss": 1.5029, "step": 7892 }, { "epoch": 0.6040869432113883, "grad_norm": 0.57115056685685, "learning_rate": 4.8627820154802075e-06, "loss": 1.229, "step": 7893 }, { "epoch": 0.6041634777284556, "grad_norm": 0.6184117928258694, "learning_rate": 4.8618314580940064e-06, "loss": 1.4977, "step": 7894 }, { "epoch": 0.6042400122455227, "grad_norm": 0.7047155845668902, "learning_rate": 4.860880970640283e-06, "loss": 1.1804, "step": 7895 }, { "epoch": 0.60431654676259, "grad_norm": 0.6236399544735168, "learning_rate": 4.859930553177437e-06, "loss": 1.3504, "step": 7896 }, { "epoch": 0.6043930812796571, "grad_norm": 0.5441027890311444, "learning_rate": 4.858980205763868e-06, "loss": 1.143, "step": 7897 }, { "epoch": 0.6044696157967243, "grad_norm": 0.6319064059867263, "learning_rate": 4.858029928457965e-06, "loss": 1.4914, "step": 7898 }, { "epoch": 0.6045461503137916, "grad_norm": 0.49954678132649016, "learning_rate": 4.857079721318118e-06, "loss": 1.1648, "step": 7899 }, { "epoch": 0.6046226848308587, "grad_norm": 0.5408575297502234, "learning_rate": 4.856129584402709e-06, "loss": 1.1925, "step": 7900 }, { "epoch": 0.6046992193479259, "grad_norm": 0.8108543942710983, "learning_rate": 4.855179517770115e-06, "loss": 1.4282, "step": 7901 }, { "epoch": 0.6047757538649932, "grad_norm": 0.5899301233940923, "learning_rate": 4.8542295214787175e-06, "loss": 1.3825, "step": 7902 }, { "epoch": 0.6048522883820603, "grad_norm": 0.6084482609645816, "learning_rate": 4.8532795955868816e-06, "loss": 1.4643, "step": 7903 }, { "epoch": 0.6049288228991275, "grad_norm": 0.5131256362071818, "learning_rate": 4.852329740152977e-06, "loss": 1.2156, "step": 7904 }, { "epoch": 0.6050053574161947, "grad_norm": 0.619488548437101, "learning_rate": 4.851379955235363e-06, "loss": 1.6071, "step": 7905 }, { "epoch": 0.6050818919332619, "grad_norm": 0.6407076799767109, "learning_rate": 4.850430240892399e-06, "loss": 1.6225, "step": 7906 }, { "epoch": 0.6051584264503291, "grad_norm": 0.6573574344578148, "learning_rate": 4.849480597182436e-06, "loss": 1.5529, "step": 7907 }, { "epoch": 0.6052349609673963, "grad_norm": 0.5603974284200867, "learning_rate": 4.8485310241638265e-06, "loss": 1.3574, "step": 7908 }, { "epoch": 0.6053114954844635, "grad_norm": 0.6396373059196724, "learning_rate": 4.847581521894912e-06, "loss": 1.5554, "step": 7909 }, { "epoch": 0.6053880300015307, "grad_norm": 0.6152071783196237, "learning_rate": 4.846632090434034e-06, "loss": 1.5188, "step": 7910 }, { "epoch": 0.6054645645185979, "grad_norm": 0.6146740214904761, "learning_rate": 4.8456827298395295e-06, "loss": 1.3356, "step": 7911 }, { "epoch": 0.6055410990356651, "grad_norm": 0.5659174047525062, "learning_rate": 4.844733440169724e-06, "loss": 1.2315, "step": 7912 }, { "epoch": 0.6056176335527322, "grad_norm": 0.6900204502638377, "learning_rate": 4.843784221482955e-06, "loss": 1.5847, "step": 7913 }, { "epoch": 0.6056941680697995, "grad_norm": 0.7509375449492909, "learning_rate": 4.842835073837537e-06, "loss": 1.6544, "step": 7914 }, { "epoch": 0.6057707025868667, "grad_norm": 0.5224266723069385, "learning_rate": 4.841885997291793e-06, "loss": 1.1896, "step": 7915 }, { "epoch": 0.6058472371039338, "grad_norm": 0.7364866197480733, "learning_rate": 4.840936991904035e-06, "loss": 1.616, "step": 7916 }, { "epoch": 0.6059237716210011, "grad_norm": 0.6069056599050497, "learning_rate": 4.839988057732572e-06, "loss": 1.5462, "step": 7917 }, { "epoch": 0.6060003061380683, "grad_norm": 0.5732837793357347, "learning_rate": 4.839039194835712e-06, "loss": 1.302, "step": 7918 }, { "epoch": 0.6060768406551355, "grad_norm": 0.518829088833745, "learning_rate": 4.838090403271754e-06, "loss": 0.9491, "step": 7919 }, { "epoch": 0.6061533751722027, "grad_norm": 0.6775248211809274, "learning_rate": 4.837141683098995e-06, "loss": 1.2406, "step": 7920 }, { "epoch": 0.6062299096892698, "grad_norm": 0.6948020906999318, "learning_rate": 4.836193034375727e-06, "loss": 1.5167, "step": 7921 }, { "epoch": 0.6063064442063371, "grad_norm": 0.6148261373001269, "learning_rate": 4.83524445716024e-06, "loss": 1.623, "step": 7922 }, { "epoch": 0.6063829787234043, "grad_norm": 0.7175465958493934, "learning_rate": 4.8342959515108115e-06, "loss": 1.5084, "step": 7923 }, { "epoch": 0.6064595132404714, "grad_norm": 0.5432432594534977, "learning_rate": 4.833347517485727e-06, "loss": 1.2382, "step": 7924 }, { "epoch": 0.6065360477575387, "grad_norm": 0.7784266487151755, "learning_rate": 4.832399155143258e-06, "loss": 1.7615, "step": 7925 }, { "epoch": 0.6066125822746058, "grad_norm": 0.6582696432000363, "learning_rate": 4.831450864541677e-06, "loss": 1.5186, "step": 7926 }, { "epoch": 0.606689116791673, "grad_norm": 0.7565926136439026, "learning_rate": 4.830502645739247e-06, "loss": 1.4242, "step": 7927 }, { "epoch": 0.6067656513087403, "grad_norm": 0.6441855634179131, "learning_rate": 4.82955449879423e-06, "loss": 1.5841, "step": 7928 }, { "epoch": 0.6068421858258074, "grad_norm": 0.5841713102074814, "learning_rate": 4.828606423764884e-06, "loss": 1.4455, "step": 7929 }, { "epoch": 0.6069187203428746, "grad_norm": 0.7822476595161623, "learning_rate": 4.827658420709462e-06, "loss": 1.4573, "step": 7930 }, { "epoch": 0.6069952548599419, "grad_norm": 0.6496647815780492, "learning_rate": 4.82671048968621e-06, "loss": 1.6289, "step": 7931 }, { "epoch": 0.607071789377009, "grad_norm": 0.6705361098628482, "learning_rate": 4.825762630753375e-06, "loss": 1.5566, "step": 7932 }, { "epoch": 0.6071483238940762, "grad_norm": 0.7383645081025774, "learning_rate": 4.824814843969193e-06, "loss": 1.6569, "step": 7933 }, { "epoch": 0.6072248584111434, "grad_norm": 0.6122792474095551, "learning_rate": 4.823867129391897e-06, "loss": 1.1722, "step": 7934 }, { "epoch": 0.6073013929282106, "grad_norm": 0.5711938641579983, "learning_rate": 4.8229194870797245e-06, "loss": 1.3449, "step": 7935 }, { "epoch": 0.6073779274452779, "grad_norm": 0.5563875587586575, "learning_rate": 4.821971917090896e-06, "loss": 1.2243, "step": 7936 }, { "epoch": 0.607454461962345, "grad_norm": 0.573488246108279, "learning_rate": 4.821024419483635e-06, "loss": 1.3801, "step": 7937 }, { "epoch": 0.6075309964794122, "grad_norm": 0.537791789474332, "learning_rate": 4.8200769943161584e-06, "loss": 1.1437, "step": 7938 }, { "epoch": 0.6076075309964795, "grad_norm": 0.654608525191034, "learning_rate": 4.819129641646678e-06, "loss": 1.3631, "step": 7939 }, { "epoch": 0.6076840655135466, "grad_norm": 0.6917811218156182, "learning_rate": 4.8181823615334035e-06, "loss": 1.5724, "step": 7940 }, { "epoch": 0.6077606000306138, "grad_norm": 0.7195551979092039, "learning_rate": 4.817235154034537e-06, "loss": 1.7902, "step": 7941 }, { "epoch": 0.607837134547681, "grad_norm": 0.6864602609132852, "learning_rate": 4.816288019208279e-06, "loss": 1.7572, "step": 7942 }, { "epoch": 0.6079136690647482, "grad_norm": 0.5641534989158689, "learning_rate": 4.815340957112823e-06, "loss": 1.2267, "step": 7943 }, { "epoch": 0.6079902035818154, "grad_norm": 0.6047158479231651, "learning_rate": 4.814393967806361e-06, "loss": 1.5035, "step": 7944 }, { "epoch": 0.6080667380988826, "grad_norm": 0.6611129673755838, "learning_rate": 4.813447051347075e-06, "loss": 1.6196, "step": 7945 }, { "epoch": 0.6081432726159498, "grad_norm": 0.5565549407646859, "learning_rate": 4.812500207793151e-06, "loss": 1.4059, "step": 7946 }, { "epoch": 0.608219807133017, "grad_norm": 0.6570908346295803, "learning_rate": 4.811553437202766e-06, "loss": 1.5867, "step": 7947 }, { "epoch": 0.6082963416500842, "grad_norm": 0.7106457326858412, "learning_rate": 4.810606739634089e-06, "loss": 1.2232, "step": 7948 }, { "epoch": 0.6083728761671514, "grad_norm": 0.7696348362659562, "learning_rate": 4.80966011514529e-06, "loss": 1.5383, "step": 7949 }, { "epoch": 0.6084494106842185, "grad_norm": 0.847555664150494, "learning_rate": 4.808713563794532e-06, "loss": 1.6657, "step": 7950 }, { "epoch": 0.6085259452012858, "grad_norm": 0.6674402153339957, "learning_rate": 4.807767085639974e-06, "loss": 1.5822, "step": 7951 }, { "epoch": 0.608602479718353, "grad_norm": 0.5916036094527655, "learning_rate": 4.80682068073977e-06, "loss": 1.3014, "step": 7952 }, { "epoch": 0.6086790142354201, "grad_norm": 0.6534547174829148, "learning_rate": 4.8058743491520696e-06, "loss": 1.7401, "step": 7953 }, { "epoch": 0.6087555487524874, "grad_norm": 0.6043954186661213, "learning_rate": 4.804928090935021e-06, "loss": 1.482, "step": 7954 }, { "epoch": 0.6088320832695545, "grad_norm": 0.5734478144979832, "learning_rate": 4.803981906146761e-06, "loss": 1.4014, "step": 7955 }, { "epoch": 0.6089086177866218, "grad_norm": 0.6120842160760069, "learning_rate": 4.8030357948454274e-06, "loss": 1.5594, "step": 7956 }, { "epoch": 0.608985152303689, "grad_norm": 0.6000847723539103, "learning_rate": 4.802089757089155e-06, "loss": 1.5704, "step": 7957 }, { "epoch": 0.6090616868207561, "grad_norm": 0.6300744364226338, "learning_rate": 4.801143792936068e-06, "loss": 1.5444, "step": 7958 }, { "epoch": 0.6091382213378234, "grad_norm": 0.5606572053090543, "learning_rate": 4.80019790244429e-06, "loss": 1.2557, "step": 7959 }, { "epoch": 0.6092147558548906, "grad_norm": 0.5785174459467651, "learning_rate": 4.799252085671942e-06, "loss": 1.3852, "step": 7960 }, { "epoch": 0.6092912903719577, "grad_norm": 0.5172959072645985, "learning_rate": 4.7983063426771315e-06, "loss": 1.2103, "step": 7961 }, { "epoch": 0.609367824889025, "grad_norm": 0.5324005149934832, "learning_rate": 4.797360673517974e-06, "loss": 1.1952, "step": 7962 }, { "epoch": 0.6094443594060921, "grad_norm": 0.6399282641251851, "learning_rate": 4.796415078252572e-06, "loss": 1.406, "step": 7963 }, { "epoch": 0.6095208939231593, "grad_norm": 0.533746282779052, "learning_rate": 4.795469556939025e-06, "loss": 1.3063, "step": 7964 }, { "epoch": 0.6095974284402266, "grad_norm": 0.5710052351333684, "learning_rate": 4.79452410963543e-06, "loss": 1.3455, "step": 7965 }, { "epoch": 0.6096739629572937, "grad_norm": 0.575291288383115, "learning_rate": 4.7935787363998776e-06, "loss": 1.4666, "step": 7966 }, { "epoch": 0.6097504974743609, "grad_norm": 0.6180174222947399, "learning_rate": 4.7926334372904515e-06, "loss": 1.5141, "step": 7967 }, { "epoch": 0.6098270319914282, "grad_norm": 0.5682910913156489, "learning_rate": 4.791688212365238e-06, "loss": 1.3019, "step": 7968 }, { "epoch": 0.6099035665084953, "grad_norm": 0.5715311367970788, "learning_rate": 4.790743061682314e-06, "loss": 1.2571, "step": 7969 }, { "epoch": 0.6099801010255625, "grad_norm": 0.5556548534074397, "learning_rate": 4.7897979852997515e-06, "loss": 1.3987, "step": 7970 }, { "epoch": 0.6100566355426297, "grad_norm": 0.8894452902444577, "learning_rate": 4.788852983275617e-06, "loss": 1.8049, "step": 7971 }, { "epoch": 0.6101331700596969, "grad_norm": 1.2569017487090084, "learning_rate": 4.787908055667976e-06, "loss": 1.5463, "step": 7972 }, { "epoch": 0.6102097045767642, "grad_norm": 0.6744706414413233, "learning_rate": 4.786963202534888e-06, "loss": 1.6136, "step": 7973 }, { "epoch": 0.6102862390938313, "grad_norm": 0.5709611388616885, "learning_rate": 4.786018423934406e-06, "loss": 1.2516, "step": 7974 }, { "epoch": 0.6103627736108985, "grad_norm": 0.5806568399156271, "learning_rate": 4.785073719924583e-06, "loss": 1.3743, "step": 7975 }, { "epoch": 0.6104393081279658, "grad_norm": 0.621591833813637, "learning_rate": 4.784129090563463e-06, "loss": 1.5011, "step": 7976 }, { "epoch": 0.6105158426450329, "grad_norm": 0.9323344570459876, "learning_rate": 4.783184535909085e-06, "loss": 1.4924, "step": 7977 }, { "epoch": 0.6105923771621001, "grad_norm": 0.6123887986169675, "learning_rate": 4.7822400560194845e-06, "loss": 1.4459, "step": 7978 }, { "epoch": 0.6106689116791673, "grad_norm": 0.5164067536351505, "learning_rate": 4.781295650952699e-06, "loss": 1.1321, "step": 7979 }, { "epoch": 0.6107454461962345, "grad_norm": 0.5685598229886789, "learning_rate": 4.780351320766751e-06, "loss": 1.3288, "step": 7980 }, { "epoch": 0.6108219807133017, "grad_norm": 0.5809411970817389, "learning_rate": 4.779407065519663e-06, "loss": 1.3696, "step": 7981 }, { "epoch": 0.6108985152303689, "grad_norm": 0.6646261599573212, "learning_rate": 4.778462885269455e-06, "loss": 1.6215, "step": 7982 }, { "epoch": 0.6109750497474361, "grad_norm": 0.5848431899147075, "learning_rate": 4.777518780074138e-06, "loss": 1.3245, "step": 7983 }, { "epoch": 0.6110515842645033, "grad_norm": 0.5838739174122397, "learning_rate": 4.776574749991722e-06, "loss": 1.5438, "step": 7984 }, { "epoch": 0.6111281187815705, "grad_norm": 0.6348265444037927, "learning_rate": 4.775630795080212e-06, "loss": 1.5346, "step": 7985 }, { "epoch": 0.6112046532986377, "grad_norm": 0.706353262637962, "learning_rate": 4.7746869153976035e-06, "loss": 1.454, "step": 7986 }, { "epoch": 0.6112811878157048, "grad_norm": 0.5952775336315271, "learning_rate": 4.7737431110018964e-06, "loss": 1.397, "step": 7987 }, { "epoch": 0.6113577223327721, "grad_norm": 0.5934398582051175, "learning_rate": 4.772799381951078e-06, "loss": 1.3348, "step": 7988 }, { "epoch": 0.6114342568498393, "grad_norm": 0.5987853334524372, "learning_rate": 4.771855728303132e-06, "loss": 1.6015, "step": 7989 }, { "epoch": 0.6115107913669064, "grad_norm": 0.7578007483909871, "learning_rate": 4.770912150116042e-06, "loss": 1.5565, "step": 7990 }, { "epoch": 0.6115873258839737, "grad_norm": 0.543426245925448, "learning_rate": 4.769968647447786e-06, "loss": 1.3173, "step": 7991 }, { "epoch": 0.6116638604010408, "grad_norm": 0.5679384759489076, "learning_rate": 4.769025220356332e-06, "loss": 1.3456, "step": 7992 }, { "epoch": 0.611740394918108, "grad_norm": 0.5717933657787898, "learning_rate": 4.768081868899649e-06, "loss": 1.4288, "step": 7993 }, { "epoch": 0.6118169294351753, "grad_norm": 0.561022917620659, "learning_rate": 4.767138593135698e-06, "loss": 1.4551, "step": 7994 }, { "epoch": 0.6118934639522424, "grad_norm": 0.6024715676997918, "learning_rate": 4.766195393122438e-06, "loss": 1.4239, "step": 7995 }, { "epoch": 0.6119699984693097, "grad_norm": 0.7265889108308794, "learning_rate": 4.765252268917821e-06, "loss": 1.5406, "step": 7996 }, { "epoch": 0.6120465329863769, "grad_norm": 0.5267690803970233, "learning_rate": 4.764309220579797e-06, "loss": 1.0664, "step": 7997 }, { "epoch": 0.612123067503444, "grad_norm": 0.6019153184969362, "learning_rate": 4.763366248166307e-06, "loss": 1.3581, "step": 7998 }, { "epoch": 0.6121996020205113, "grad_norm": 0.6174212311276254, "learning_rate": 4.762423351735292e-06, "loss": 1.4139, "step": 7999 }, { "epoch": 0.6122761365375784, "grad_norm": 0.6216342005236505, "learning_rate": 4.761480531344685e-06, "loss": 1.3582, "step": 8000 }, { "epoch": 0.6123526710546456, "grad_norm": 1.0158530577441902, "learning_rate": 4.760537787052417e-06, "loss": 1.6051, "step": 8001 }, { "epoch": 0.6124292055717129, "grad_norm": 0.5810128172118069, "learning_rate": 4.759595118916414e-06, "loss": 1.4005, "step": 8002 }, { "epoch": 0.61250574008878, "grad_norm": 0.6285006224192243, "learning_rate": 4.7586525269945935e-06, "loss": 1.3061, "step": 8003 }, { "epoch": 0.6125822746058472, "grad_norm": 0.559988452357767, "learning_rate": 4.757710011344874e-06, "loss": 1.2329, "step": 8004 }, { "epoch": 0.6126588091229145, "grad_norm": 1.280493933279665, "learning_rate": 4.756767572025164e-06, "loss": 1.6942, "step": 8005 }, { "epoch": 0.6127353436399816, "grad_norm": 0.6538815930160391, "learning_rate": 4.755825209093371e-06, "loss": 1.6837, "step": 8006 }, { "epoch": 0.6128118781570489, "grad_norm": 0.7357390996770897, "learning_rate": 4.754882922607398e-06, "loss": 1.8445, "step": 8007 }, { "epoch": 0.612888412674116, "grad_norm": 0.6087721240056339, "learning_rate": 4.753940712625139e-06, "loss": 1.3779, "step": 8008 }, { "epoch": 0.6129649471911832, "grad_norm": 0.6145678245359923, "learning_rate": 4.752998579204487e-06, "loss": 1.5789, "step": 8009 }, { "epoch": 0.6130414817082505, "grad_norm": 0.5192934317705694, "learning_rate": 4.752056522403332e-06, "loss": 1.0498, "step": 8010 }, { "epoch": 0.6131180162253176, "grad_norm": 0.6250739977020572, "learning_rate": 4.751114542279553e-06, "loss": 1.5681, "step": 8011 }, { "epoch": 0.6131945507423848, "grad_norm": 0.6021128014686743, "learning_rate": 4.750172638891031e-06, "loss": 1.272, "step": 8012 }, { "epoch": 0.6132710852594521, "grad_norm": 1.2584236003675005, "learning_rate": 4.749230812295638e-06, "loss": 1.5797, "step": 8013 }, { "epoch": 0.6133476197765192, "grad_norm": 0.6679401547465581, "learning_rate": 4.748289062551243e-06, "loss": 1.4923, "step": 8014 }, { "epoch": 0.6134241542935864, "grad_norm": 0.7592469275962932, "learning_rate": 4.747347389715711e-06, "loss": 1.8472, "step": 8015 }, { "epoch": 0.6135006888106536, "grad_norm": 0.6792481190832063, "learning_rate": 4.746405793846898e-06, "loss": 1.505, "step": 8016 }, { "epoch": 0.6135772233277208, "grad_norm": 0.5769576998767286, "learning_rate": 4.745464275002663e-06, "loss": 1.3132, "step": 8017 }, { "epoch": 0.613653757844788, "grad_norm": 0.5233568767807031, "learning_rate": 4.744522833240852e-06, "loss": 1.2634, "step": 8018 }, { "epoch": 0.6137302923618552, "grad_norm": 0.6452711928032586, "learning_rate": 4.743581468619311e-06, "loss": 1.505, "step": 8019 }, { "epoch": 0.6138068268789224, "grad_norm": 0.6981938707032715, "learning_rate": 4.742640181195881e-06, "loss": 1.7234, "step": 8020 }, { "epoch": 0.6138833613959896, "grad_norm": 0.6368209702288528, "learning_rate": 4.741698971028396e-06, "loss": 1.5112, "step": 8021 }, { "epoch": 0.6139598959130568, "grad_norm": 0.654860271243878, "learning_rate": 4.74075783817469e-06, "loss": 1.7234, "step": 8022 }, { "epoch": 0.614036430430124, "grad_norm": 0.594253197216211, "learning_rate": 4.739816782692586e-06, "loss": 1.5005, "step": 8023 }, { "epoch": 0.6141129649471911, "grad_norm": 0.8217512712784788, "learning_rate": 4.738875804639905e-06, "loss": 1.8611, "step": 8024 }, { "epoch": 0.6141894994642584, "grad_norm": 0.6513600167195456, "learning_rate": 4.737934904074466e-06, "loss": 1.5234, "step": 8025 }, { "epoch": 0.6142660339813256, "grad_norm": 0.5926245630512241, "learning_rate": 4.7369940810540785e-06, "loss": 1.5281, "step": 8026 }, { "epoch": 0.6143425684983927, "grad_norm": 0.533243250057605, "learning_rate": 4.7360533356365504e-06, "loss": 1.2448, "step": 8027 }, { "epoch": 0.61441910301546, "grad_norm": 0.5643165095949878, "learning_rate": 4.735112667879684e-06, "loss": 1.3673, "step": 8028 }, { "epoch": 0.6144956375325271, "grad_norm": 0.6520276768679384, "learning_rate": 4.734172077841277e-06, "loss": 1.6727, "step": 8029 }, { "epoch": 0.6145721720495944, "grad_norm": 0.5652055747625881, "learning_rate": 4.7332315655791206e-06, "loss": 1.284, "step": 8030 }, { "epoch": 0.6146487065666616, "grad_norm": 0.6906990886712943, "learning_rate": 4.732291131151005e-06, "loss": 1.7845, "step": 8031 }, { "epoch": 0.6147252410837287, "grad_norm": 0.6353009627113435, "learning_rate": 4.731350774614709e-06, "loss": 1.5132, "step": 8032 }, { "epoch": 0.614801775600796, "grad_norm": 0.5435362044331323, "learning_rate": 4.730410496028017e-06, "loss": 1.2648, "step": 8033 }, { "epoch": 0.6148783101178632, "grad_norm": 0.5994472817146296, "learning_rate": 4.729470295448699e-06, "loss": 1.4701, "step": 8034 }, { "epoch": 0.6149548446349303, "grad_norm": 0.5769072154857533, "learning_rate": 4.728530172934524e-06, "loss": 1.3723, "step": 8035 }, { "epoch": 0.6150313791519976, "grad_norm": 0.6470700318585321, "learning_rate": 4.7275901285432565e-06, "loss": 1.6479, "step": 8036 }, { "epoch": 0.6151079136690647, "grad_norm": 0.634436564140064, "learning_rate": 4.7266501623326555e-06, "loss": 1.5351, "step": 8037 }, { "epoch": 0.6151844481861319, "grad_norm": 0.5689081427418249, "learning_rate": 4.725710274360474e-06, "loss": 1.3021, "step": 8038 }, { "epoch": 0.6152609827031992, "grad_norm": 0.7845856583484451, "learning_rate": 4.724770464684462e-06, "loss": 1.6533, "step": 8039 }, { "epoch": 0.6153375172202663, "grad_norm": 0.6470179834007204, "learning_rate": 4.723830733362367e-06, "loss": 1.299, "step": 8040 }, { "epoch": 0.6154140517373335, "grad_norm": 0.5858874008113776, "learning_rate": 4.722891080451927e-06, "loss": 1.2082, "step": 8041 }, { "epoch": 0.6154905862544008, "grad_norm": 0.5904338997789988, "learning_rate": 4.721951506010876e-06, "loss": 1.5453, "step": 8042 }, { "epoch": 0.6155671207714679, "grad_norm": 0.6136038701052404, "learning_rate": 4.7210120100969424e-06, "loss": 1.5789, "step": 8043 }, { "epoch": 0.6156436552885352, "grad_norm": 0.8371580595663765, "learning_rate": 4.720072592767857e-06, "loss": 1.6253, "step": 8044 }, { "epoch": 0.6157201898056023, "grad_norm": 0.5700810467602133, "learning_rate": 4.719133254081338e-06, "loss": 1.4273, "step": 8045 }, { "epoch": 0.6157967243226695, "grad_norm": 0.6634945297048451, "learning_rate": 4.718193994095101e-06, "loss": 1.3532, "step": 8046 }, { "epoch": 0.6158732588397368, "grad_norm": 0.699246326034229, "learning_rate": 4.717254812866856e-06, "loss": 1.5145, "step": 8047 }, { "epoch": 0.6159497933568039, "grad_norm": 0.5804522214974345, "learning_rate": 4.716315710454311e-06, "loss": 1.5029, "step": 8048 }, { "epoch": 0.6160263278738711, "grad_norm": 0.8255847448804798, "learning_rate": 4.715376686915164e-06, "loss": 1.6651, "step": 8049 }, { "epoch": 0.6161028623909384, "grad_norm": 0.6002682596564689, "learning_rate": 4.7144377423071155e-06, "loss": 1.4203, "step": 8050 }, { "epoch": 0.6161793969080055, "grad_norm": 0.5411748093050996, "learning_rate": 4.7134988766878556e-06, "loss": 1.2333, "step": 8051 }, { "epoch": 0.6162559314250727, "grad_norm": 0.650624363673381, "learning_rate": 4.712560090115069e-06, "loss": 1.4465, "step": 8052 }, { "epoch": 0.6163324659421399, "grad_norm": 0.6377332562372258, "learning_rate": 4.711621382646441e-06, "loss": 1.2568, "step": 8053 }, { "epoch": 0.6164090004592071, "grad_norm": 0.6661757064885389, "learning_rate": 4.710682754339644e-06, "loss": 1.6921, "step": 8054 }, { "epoch": 0.6164855349762743, "grad_norm": 0.581320194689655, "learning_rate": 4.709744205252356e-06, "loss": 1.3793, "step": 8055 }, { "epoch": 0.6165620694933415, "grad_norm": 0.580616296991811, "learning_rate": 4.708805735442239e-06, "loss": 1.5084, "step": 8056 }, { "epoch": 0.6166386040104087, "grad_norm": 0.5837381220940714, "learning_rate": 4.707867344966959e-06, "loss": 1.3029, "step": 8057 }, { "epoch": 0.6167151385274758, "grad_norm": 0.5958319673619958, "learning_rate": 4.7069290338841715e-06, "loss": 1.39, "step": 8058 }, { "epoch": 0.6167916730445431, "grad_norm": 0.5267092444455831, "learning_rate": 4.70599080225153e-06, "loss": 1.3505, "step": 8059 }, { "epoch": 0.6168682075616103, "grad_norm": 1.3056541867558282, "learning_rate": 4.7050526501266814e-06, "loss": 1.5684, "step": 8060 }, { "epoch": 0.6169447420786774, "grad_norm": 0.6444488228070647, "learning_rate": 4.704114577567269e-06, "loss": 1.6089, "step": 8061 }, { "epoch": 0.6170212765957447, "grad_norm": 0.6482810095755887, "learning_rate": 4.703176584630932e-06, "loss": 1.5503, "step": 8062 }, { "epoch": 0.6170978111128119, "grad_norm": 0.5926963517863509, "learning_rate": 4.702238671375301e-06, "loss": 1.4699, "step": 8063 }, { "epoch": 0.617174345629879, "grad_norm": 0.5184248281459307, "learning_rate": 4.701300837858007e-06, "loss": 1.046, "step": 8064 }, { "epoch": 0.6172508801469463, "grad_norm": 0.5891551183958682, "learning_rate": 4.70036308413667e-06, "loss": 1.3678, "step": 8065 }, { "epoch": 0.6173274146640134, "grad_norm": 0.7472214828625677, "learning_rate": 4.699425410268912e-06, "loss": 1.421, "step": 8066 }, { "epoch": 0.6174039491810807, "grad_norm": 0.6005967626277148, "learning_rate": 4.698487816312344e-06, "loss": 1.4571, "step": 8067 }, { "epoch": 0.6174804836981479, "grad_norm": 0.6193793877776854, "learning_rate": 4.697550302324576e-06, "loss": 1.4222, "step": 8068 }, { "epoch": 0.617557018215215, "grad_norm": 0.744147914993023, "learning_rate": 4.696612868363211e-06, "loss": 1.8029, "step": 8069 }, { "epoch": 0.6176335527322823, "grad_norm": 0.5946782648465466, "learning_rate": 4.695675514485847e-06, "loss": 1.5, "step": 8070 }, { "epoch": 0.6177100872493495, "grad_norm": 0.5091647436923444, "learning_rate": 4.694738240750079e-06, "loss": 1.036, "step": 8071 }, { "epoch": 0.6177866217664166, "grad_norm": 0.5195573932544922, "learning_rate": 4.6938010472134954e-06, "loss": 1.1602, "step": 8072 }, { "epoch": 0.6178631562834839, "grad_norm": 0.614525735154149, "learning_rate": 4.69286393393368e-06, "loss": 1.5499, "step": 8073 }, { "epoch": 0.617939690800551, "grad_norm": 0.552903258649516, "learning_rate": 4.691926900968213e-06, "loss": 1.2793, "step": 8074 }, { "epoch": 0.6180162253176182, "grad_norm": 0.6333473215899456, "learning_rate": 4.690989948374666e-06, "loss": 1.5868, "step": 8075 }, { "epoch": 0.6180927598346855, "grad_norm": 0.5904860611237637, "learning_rate": 4.69005307621061e-06, "loss": 1.2678, "step": 8076 }, { "epoch": 0.6181692943517526, "grad_norm": 0.6258907837451643, "learning_rate": 4.68911628453361e-06, "loss": 1.57, "step": 8077 }, { "epoch": 0.6182458288688198, "grad_norm": 0.6751900250575446, "learning_rate": 4.688179573401224e-06, "loss": 1.6782, "step": 8078 }, { "epoch": 0.6183223633858871, "grad_norm": 0.6647789360698937, "learning_rate": 4.6872429428710064e-06, "loss": 1.6732, "step": 8079 }, { "epoch": 0.6183988979029542, "grad_norm": 0.5779877840093822, "learning_rate": 4.686306393000508e-06, "loss": 1.3479, "step": 8080 }, { "epoch": 0.6184754324200215, "grad_norm": 0.7370579366543092, "learning_rate": 4.6853699238472705e-06, "loss": 1.1935, "step": 8081 }, { "epoch": 0.6185519669370886, "grad_norm": 0.9533795406172741, "learning_rate": 4.684433535468834e-06, "loss": 1.631, "step": 8082 }, { "epoch": 0.6186285014541558, "grad_norm": 0.8105382562788829, "learning_rate": 4.683497227922735e-06, "loss": 1.6415, "step": 8083 }, { "epoch": 0.6187050359712231, "grad_norm": 0.5704325745315186, "learning_rate": 4.682561001266504e-06, "loss": 1.363, "step": 8084 }, { "epoch": 0.6187815704882902, "grad_norm": 0.5780118437933649, "learning_rate": 4.681624855557661e-06, "loss": 1.4536, "step": 8085 }, { "epoch": 0.6188581050053574, "grad_norm": 0.5430583735035076, "learning_rate": 4.680688790853729e-06, "loss": 1.1015, "step": 8086 }, { "epoch": 0.6189346395224247, "grad_norm": 1.0075940518476674, "learning_rate": 4.679752807212221e-06, "loss": 1.6279, "step": 8087 }, { "epoch": 0.6190111740394918, "grad_norm": 0.6842463678805015, "learning_rate": 4.678816904690647e-06, "loss": 1.469, "step": 8088 }, { "epoch": 0.619087708556559, "grad_norm": 0.530011764908055, "learning_rate": 4.677881083346514e-06, "loss": 1.0517, "step": 8089 }, { "epoch": 0.6191642430736262, "grad_norm": 0.6046406238617547, "learning_rate": 4.676945343237318e-06, "loss": 1.3536, "step": 8090 }, { "epoch": 0.6192407775906934, "grad_norm": 0.8796716732822659, "learning_rate": 4.676009684420556e-06, "loss": 1.4746, "step": 8091 }, { "epoch": 0.6193173121077606, "grad_norm": 0.5564428922299244, "learning_rate": 4.675074106953716e-06, "loss": 1.3265, "step": 8092 }, { "epoch": 0.6193938466248278, "grad_norm": 0.5178825272373069, "learning_rate": 4.674138610894283e-06, "loss": 1.1319, "step": 8093 }, { "epoch": 0.619470381141895, "grad_norm": 0.8899015005727621, "learning_rate": 4.6732031962997384e-06, "loss": 1.5881, "step": 8094 }, { "epoch": 0.6195469156589621, "grad_norm": 0.5616394513856799, "learning_rate": 4.672267863227556e-06, "loss": 1.0399, "step": 8095 }, { "epoch": 0.6196234501760294, "grad_norm": 0.5164304074538624, "learning_rate": 4.6713326117352054e-06, "loss": 1.0613, "step": 8096 }, { "epoch": 0.6196999846930966, "grad_norm": 1.3366029599866003, "learning_rate": 4.6703974418801525e-06, "loss": 1.1952, "step": 8097 }, { "epoch": 0.6197765192101637, "grad_norm": 1.2813193991694756, "learning_rate": 4.669462353719853e-06, "loss": 1.644, "step": 8098 }, { "epoch": 0.619853053727231, "grad_norm": 0.6293090671843056, "learning_rate": 4.6685273473117645e-06, "loss": 1.4948, "step": 8099 }, { "epoch": 0.6199295882442982, "grad_norm": 0.5496432274575808, "learning_rate": 4.667592422713336e-06, "loss": 1.3097, "step": 8100 }, { "epoch": 0.6200061227613654, "grad_norm": 0.5157828738433996, "learning_rate": 4.666657579982014e-06, "loss": 1.1117, "step": 8101 }, { "epoch": 0.6200826572784326, "grad_norm": 0.6628553047963378, "learning_rate": 4.665722819175235e-06, "loss": 1.7391, "step": 8102 }, { "epoch": 0.6201591917954997, "grad_norm": 0.5879068399751146, "learning_rate": 4.664788140350435e-06, "loss": 1.3091, "step": 8103 }, { "epoch": 0.620235726312567, "grad_norm": 0.6414934960154319, "learning_rate": 4.663853543565041e-06, "loss": 1.726, "step": 8104 }, { "epoch": 0.6203122608296342, "grad_norm": 0.7020847724159075, "learning_rate": 4.662919028876482e-06, "loss": 1.6498, "step": 8105 }, { "epoch": 0.6203887953467013, "grad_norm": 0.6708933553542594, "learning_rate": 4.661984596342174e-06, "loss": 1.6218, "step": 8106 }, { "epoch": 0.6204653298637686, "grad_norm": 0.5124242304881089, "learning_rate": 4.661050246019533e-06, "loss": 1.1231, "step": 8107 }, { "epoch": 0.6205418643808358, "grad_norm": 0.5353963960363759, "learning_rate": 4.660115977965968e-06, "loss": 1.2456, "step": 8108 }, { "epoch": 0.6206183988979029, "grad_norm": 0.7313625638895208, "learning_rate": 4.6591817922388805e-06, "loss": 1.6232, "step": 8109 }, { "epoch": 0.6206949334149702, "grad_norm": 0.5958457018562229, "learning_rate": 4.658247688895674e-06, "loss": 1.4425, "step": 8110 }, { "epoch": 0.6207714679320373, "grad_norm": 0.58309105117669, "learning_rate": 4.65731366799374e-06, "loss": 1.2158, "step": 8111 }, { "epoch": 0.6208480024491045, "grad_norm": 0.5825648215384004, "learning_rate": 4.656379729590469e-06, "loss": 1.4782, "step": 8112 }, { "epoch": 0.6209245369661718, "grad_norm": 0.6204288291340423, "learning_rate": 4.655445873743243e-06, "loss": 1.3707, "step": 8113 }, { "epoch": 0.6210010714832389, "grad_norm": 0.5869424878388443, "learning_rate": 4.654512100509443e-06, "loss": 1.2166, "step": 8114 }, { "epoch": 0.6210776060003061, "grad_norm": 0.583049039017485, "learning_rate": 4.65357840994644e-06, "loss": 1.3411, "step": 8115 }, { "epoch": 0.6211541405173734, "grad_norm": 0.5647202149281895, "learning_rate": 4.652644802111606e-06, "loss": 1.2334, "step": 8116 }, { "epoch": 0.6212306750344405, "grad_norm": 0.7970553718924599, "learning_rate": 4.651711277062302e-06, "loss": 1.3744, "step": 8117 }, { "epoch": 0.6213072095515078, "grad_norm": 0.7264391155913642, "learning_rate": 4.650777834855888e-06, "loss": 1.5354, "step": 8118 }, { "epoch": 0.6213837440685749, "grad_norm": 0.5701616440115166, "learning_rate": 4.649844475549719e-06, "loss": 1.3732, "step": 8119 }, { "epoch": 0.6214602785856421, "grad_norm": 0.5597286244219856, "learning_rate": 4.648911199201139e-06, "loss": 1.2758, "step": 8120 }, { "epoch": 0.6215368131027094, "grad_norm": 0.707659416195644, "learning_rate": 4.647978005867496e-06, "loss": 1.327, "step": 8121 }, { "epoch": 0.6216133476197765, "grad_norm": 0.5845651093875538, "learning_rate": 4.647044895606125e-06, "loss": 1.5457, "step": 8122 }, { "epoch": 0.6216898821368437, "grad_norm": 0.7028102592169385, "learning_rate": 4.646111868474361e-06, "loss": 1.6422, "step": 8123 }, { "epoch": 0.6217664166539109, "grad_norm": 0.5619742559243052, "learning_rate": 4.6451789245295305e-06, "loss": 1.3241, "step": 8124 }, { "epoch": 0.6218429511709781, "grad_norm": 0.5030959345462214, "learning_rate": 4.644246063828959e-06, "loss": 1.1758, "step": 8125 }, { "epoch": 0.6219194856880453, "grad_norm": 0.5608725143971304, "learning_rate": 4.6433132864299575e-06, "loss": 1.1657, "step": 8126 }, { "epoch": 0.6219960202051125, "grad_norm": 2.394101270803498, "learning_rate": 4.642380592389848e-06, "loss": 1.6941, "step": 8127 }, { "epoch": 0.6220725547221797, "grad_norm": 0.5148090303073449, "learning_rate": 4.641447981765932e-06, "loss": 1.14, "step": 8128 }, { "epoch": 0.6221490892392469, "grad_norm": 0.5682049260488573, "learning_rate": 4.640515454615513e-06, "loss": 1.3063, "step": 8129 }, { "epoch": 0.6222256237563141, "grad_norm": 0.5591975142594632, "learning_rate": 4.639583010995889e-06, "loss": 1.5034, "step": 8130 }, { "epoch": 0.6223021582733813, "grad_norm": 0.6579449894645393, "learning_rate": 4.63865065096435e-06, "loss": 1.4426, "step": 8131 }, { "epoch": 0.6223786927904484, "grad_norm": 0.6184728187586412, "learning_rate": 4.637718374578186e-06, "loss": 1.3293, "step": 8132 }, { "epoch": 0.6224552273075157, "grad_norm": 0.5051177512466579, "learning_rate": 4.636786181894678e-06, "loss": 1.1243, "step": 8133 }, { "epoch": 0.6225317618245829, "grad_norm": 0.5524592118207595, "learning_rate": 4.635854072971102e-06, "loss": 1.1781, "step": 8134 }, { "epoch": 0.62260829634165, "grad_norm": 0.6785221870851447, "learning_rate": 4.6349220478647295e-06, "loss": 1.5724, "step": 8135 }, { "epoch": 0.6226848308587173, "grad_norm": 0.5400195232189211, "learning_rate": 4.633990106632829e-06, "loss": 1.2359, "step": 8136 }, { "epoch": 0.6227613653757845, "grad_norm": 0.5803232313538131, "learning_rate": 4.633058249332656e-06, "loss": 1.3251, "step": 8137 }, { "epoch": 0.6228378998928517, "grad_norm": 3.141629452810428, "learning_rate": 4.632126476021472e-06, "loss": 1.1309, "step": 8138 }, { "epoch": 0.6229144344099189, "grad_norm": 0.6049666045674095, "learning_rate": 4.631194786756528e-06, "loss": 1.3961, "step": 8139 }, { "epoch": 0.622990968926986, "grad_norm": 1.01675652558486, "learning_rate": 4.6302631815950674e-06, "loss": 1.464, "step": 8140 }, { "epoch": 0.6230675034440533, "grad_norm": 0.5708801566121764, "learning_rate": 4.629331660594332e-06, "loss": 1.266, "step": 8141 }, { "epoch": 0.6231440379611205, "grad_norm": 0.6025212510040249, "learning_rate": 4.628400223811556e-06, "loss": 1.4797, "step": 8142 }, { "epoch": 0.6232205724781876, "grad_norm": 0.7382631221562289, "learning_rate": 4.627468871303971e-06, "loss": 1.5701, "step": 8143 }, { "epoch": 0.6232971069952549, "grad_norm": 0.811757561619724, "learning_rate": 4.626537603128802e-06, "loss": 1.4619, "step": 8144 }, { "epoch": 0.6233736415123221, "grad_norm": 0.6339872887901798, "learning_rate": 4.625606419343269e-06, "loss": 1.5321, "step": 8145 }, { "epoch": 0.6234501760293892, "grad_norm": 0.5337189602673104, "learning_rate": 4.624675320004586e-06, "loss": 1.3504, "step": 8146 }, { "epoch": 0.6235267105464565, "grad_norm": 0.5992276248589816, "learning_rate": 4.623744305169962e-06, "loss": 1.4143, "step": 8147 }, { "epoch": 0.6236032450635236, "grad_norm": 0.6490330636893693, "learning_rate": 4.622813374896601e-06, "loss": 1.6451, "step": 8148 }, { "epoch": 0.6236797795805908, "grad_norm": 0.648642229845464, "learning_rate": 4.6218825292417045e-06, "loss": 1.548, "step": 8149 }, { "epoch": 0.6237563140976581, "grad_norm": 0.7096245462543347, "learning_rate": 4.620951768262466e-06, "loss": 1.6627, "step": 8150 }, { "epoch": 0.6238328486147252, "grad_norm": 0.5193172832506173, "learning_rate": 4.620021092016073e-06, "loss": 1.0146, "step": 8151 }, { "epoch": 0.6239093831317924, "grad_norm": 1.0032697301118316, "learning_rate": 4.6190905005597094e-06, "loss": 1.6536, "step": 8152 }, { "epoch": 0.6239859176488597, "grad_norm": 0.5671898591204574, "learning_rate": 4.618159993950553e-06, "loss": 1.3353, "step": 8153 }, { "epoch": 0.6240624521659268, "grad_norm": 0.5610058624342826, "learning_rate": 4.617229572245778e-06, "loss": 1.2748, "step": 8154 }, { "epoch": 0.6241389866829941, "grad_norm": 0.7143992930882838, "learning_rate": 4.6162992355025515e-06, "loss": 1.6183, "step": 8155 }, { "epoch": 0.6242155212000612, "grad_norm": 0.87828400538512, "learning_rate": 4.615368983778037e-06, "loss": 1.4108, "step": 8156 }, { "epoch": 0.6242920557171284, "grad_norm": 0.8957666196801296, "learning_rate": 4.614438817129392e-06, "loss": 1.2044, "step": 8157 }, { "epoch": 0.6243685902341957, "grad_norm": 0.575363785124535, "learning_rate": 4.613508735613768e-06, "loss": 1.3174, "step": 8158 }, { "epoch": 0.6244451247512628, "grad_norm": 0.4908827034552444, "learning_rate": 4.612578739288308e-06, "loss": 1.0762, "step": 8159 }, { "epoch": 0.62452165926833, "grad_norm": 0.5986822930377305, "learning_rate": 4.611648828210162e-06, "loss": 1.4675, "step": 8160 }, { "epoch": 0.6245981937853972, "grad_norm": 1.1970695698582263, "learning_rate": 4.6107190024364615e-06, "loss": 1.5527, "step": 8161 }, { "epoch": 0.6246747283024644, "grad_norm": 0.5459668104600315, "learning_rate": 4.609789262024339e-06, "loss": 1.336, "step": 8162 }, { "epoch": 0.6247512628195316, "grad_norm": 0.6584704555726055, "learning_rate": 4.60885960703092e-06, "loss": 1.4657, "step": 8163 }, { "epoch": 0.6248277973365988, "grad_norm": 0.5672544414093682, "learning_rate": 4.607930037513325e-06, "loss": 1.2929, "step": 8164 }, { "epoch": 0.624904331853666, "grad_norm": 0.5535266876880963, "learning_rate": 4.6070005535286696e-06, "loss": 1.2281, "step": 8165 }, { "epoch": 0.6249808663707332, "grad_norm": 0.7953656079163233, "learning_rate": 4.606071155134066e-06, "loss": 1.4479, "step": 8166 }, { "epoch": 0.6250574008878004, "grad_norm": 0.6387987772947135, "learning_rate": 4.6051418423866165e-06, "loss": 1.4444, "step": 8167 }, { "epoch": 0.6251339354048676, "grad_norm": 0.8392932930670489, "learning_rate": 4.604212615343423e-06, "loss": 1.4689, "step": 8168 }, { "epoch": 0.6252104699219347, "grad_norm": 0.7428943617490759, "learning_rate": 4.603283474061577e-06, "loss": 1.7899, "step": 8169 }, { "epoch": 0.625287004439002, "grad_norm": 0.5597106905006555, "learning_rate": 4.602354418598172e-06, "loss": 1.2578, "step": 8170 }, { "epoch": 0.6253635389560692, "grad_norm": 0.5056264850168571, "learning_rate": 4.601425449010288e-06, "loss": 1.1489, "step": 8171 }, { "epoch": 0.6254400734731363, "grad_norm": 0.5916009309513519, "learning_rate": 4.600496565355006e-06, "loss": 1.355, "step": 8172 }, { "epoch": 0.6255166079902036, "grad_norm": 0.579230122197354, "learning_rate": 4.599567767689401e-06, "loss": 1.498, "step": 8173 }, { "epoch": 0.6255931425072708, "grad_norm": 0.7729201158694753, "learning_rate": 4.5986390560705355e-06, "loss": 1.5575, "step": 8174 }, { "epoch": 0.625669677024338, "grad_norm": 0.60809107031559, "learning_rate": 4.5977104305554766e-06, "loss": 1.2919, "step": 8175 }, { "epoch": 0.6257462115414052, "grad_norm": 0.6871210712180339, "learning_rate": 4.596781891201281e-06, "loss": 1.7334, "step": 8176 }, { "epoch": 0.6258227460584723, "grad_norm": 0.5835155966351083, "learning_rate": 4.595853438065001e-06, "loss": 1.3164, "step": 8177 }, { "epoch": 0.6258992805755396, "grad_norm": 0.5310113193093655, "learning_rate": 4.594925071203684e-06, "loss": 1.0392, "step": 8178 }, { "epoch": 0.6259758150926068, "grad_norm": 0.6119093408573173, "learning_rate": 4.593996790674371e-06, "loss": 1.2802, "step": 8179 }, { "epoch": 0.6260523496096739, "grad_norm": 0.60016341835084, "learning_rate": 4.593068596534096e-06, "loss": 1.5148, "step": 8180 }, { "epoch": 0.6261288841267412, "grad_norm": 1.2910435660555413, "learning_rate": 4.592140488839895e-06, "loss": 1.325, "step": 8181 }, { "epoch": 0.6262054186438084, "grad_norm": 0.5617757147076546, "learning_rate": 4.59121246764879e-06, "loss": 1.29, "step": 8182 }, { "epoch": 0.6262819531608755, "grad_norm": 0.521630465697464, "learning_rate": 4.5902845330178025e-06, "loss": 1.1839, "step": 8183 }, { "epoch": 0.6263584876779428, "grad_norm": 0.7016951880908538, "learning_rate": 4.589356685003948e-06, "loss": 1.2959, "step": 8184 }, { "epoch": 0.6264350221950099, "grad_norm": 0.7892143690709861, "learning_rate": 4.588428923664236e-06, "loss": 1.5254, "step": 8185 }, { "epoch": 0.6265115567120771, "grad_norm": 0.8739576105459106, "learning_rate": 4.587501249055669e-06, "loss": 1.4694, "step": 8186 }, { "epoch": 0.6265880912291444, "grad_norm": 0.5317240277218968, "learning_rate": 4.586573661235249e-06, "loss": 1.2254, "step": 8187 }, { "epoch": 0.6266646257462115, "grad_norm": 1.1698497502588265, "learning_rate": 4.585646160259969e-06, "loss": 1.3033, "step": 8188 }, { "epoch": 0.6267411602632788, "grad_norm": 0.5417467417499953, "learning_rate": 4.584718746186817e-06, "loss": 1.2176, "step": 8189 }, { "epoch": 0.626817694780346, "grad_norm": 0.9376495299579249, "learning_rate": 4.583791419072776e-06, "loss": 1.1317, "step": 8190 }, { "epoch": 0.6268942292974131, "grad_norm": 0.6253603454697043, "learning_rate": 4.5828641789748215e-06, "loss": 1.3467, "step": 8191 }, { "epoch": 0.6269707638144804, "grad_norm": 0.9538801717061223, "learning_rate": 4.5819370259499314e-06, "loss": 1.2744, "step": 8192 }, { "epoch": 0.6270472983315475, "grad_norm": 0.5304163929669851, "learning_rate": 4.5810099600550686e-06, "loss": 1.2033, "step": 8193 }, { "epoch": 0.6271238328486147, "grad_norm": 0.6405839237410255, "learning_rate": 4.580082981347196e-06, "loss": 1.5181, "step": 8194 }, { "epoch": 0.627200367365682, "grad_norm": 0.5216710957817606, "learning_rate": 4.57915608988327e-06, "loss": 1.1276, "step": 8195 }, { "epoch": 0.6272769018827491, "grad_norm": 0.5888892668870624, "learning_rate": 4.578229285720241e-06, "loss": 1.3216, "step": 8196 }, { "epoch": 0.6273534363998163, "grad_norm": 0.5782252286907155, "learning_rate": 4.577302568915054e-06, "loss": 1.3901, "step": 8197 }, { "epoch": 0.6274299709168835, "grad_norm": 0.5334084870326871, "learning_rate": 4.576375939524652e-06, "loss": 1.2546, "step": 8198 }, { "epoch": 0.6275065054339507, "grad_norm": 0.6640783356319155, "learning_rate": 4.575449397605966e-06, "loss": 1.6683, "step": 8199 }, { "epoch": 0.6275830399510179, "grad_norm": 0.49346327493574405, "learning_rate": 4.5745229432159315e-06, "loss": 1.0001, "step": 8200 }, { "epoch": 0.6276595744680851, "grad_norm": 0.5996393885164084, "learning_rate": 4.573596576411466e-06, "loss": 1.1846, "step": 8201 }, { "epoch": 0.6277361089851523, "grad_norm": 0.5367000161282337, "learning_rate": 4.572670297249489e-06, "loss": 1.2609, "step": 8202 }, { "epoch": 0.6278126435022195, "grad_norm": 0.6997158413032287, "learning_rate": 4.571744105786918e-06, "loss": 1.6467, "step": 8203 }, { "epoch": 0.6278891780192867, "grad_norm": 0.5686373993288597, "learning_rate": 4.570818002080659e-06, "loss": 1.4844, "step": 8204 }, { "epoch": 0.6279657125363539, "grad_norm": 0.5781835336462101, "learning_rate": 4.569891986187615e-06, "loss": 1.3234, "step": 8205 }, { "epoch": 0.628042247053421, "grad_norm": 0.6747554094009661, "learning_rate": 4.568966058164681e-06, "loss": 1.747, "step": 8206 }, { "epoch": 0.6281187815704883, "grad_norm": 0.5582646671992629, "learning_rate": 4.568040218068751e-06, "loss": 1.2703, "step": 8207 }, { "epoch": 0.6281953160875555, "grad_norm": 0.8397160423077503, "learning_rate": 4.56711446595671e-06, "loss": 1.4149, "step": 8208 }, { "epoch": 0.6282718506046227, "grad_norm": 0.6161006565594562, "learning_rate": 4.566188801885442e-06, "loss": 1.4791, "step": 8209 }, { "epoch": 0.6283483851216899, "grad_norm": 0.5896474197404707, "learning_rate": 4.565263225911818e-06, "loss": 1.3109, "step": 8210 }, { "epoch": 0.6284249196387571, "grad_norm": 0.7629054622266616, "learning_rate": 4.564337738092712e-06, "loss": 1.4728, "step": 8211 }, { "epoch": 0.6285014541558243, "grad_norm": 0.5348208174075655, "learning_rate": 4.563412338484987e-06, "loss": 1.2517, "step": 8212 }, { "epoch": 0.6285779886728915, "grad_norm": 0.5828070187467332, "learning_rate": 4.5624870271455e-06, "loss": 1.4175, "step": 8213 }, { "epoch": 0.6286545231899586, "grad_norm": 0.615200693792759, "learning_rate": 4.561561804131109e-06, "loss": 1.4561, "step": 8214 }, { "epoch": 0.6287310577070259, "grad_norm": 0.7413523564772247, "learning_rate": 4.560636669498662e-06, "loss": 1.4386, "step": 8215 }, { "epoch": 0.6288075922240931, "grad_norm": 0.5256700867919161, "learning_rate": 4.559711623305e-06, "loss": 1.061, "step": 8216 }, { "epoch": 0.6288841267411602, "grad_norm": 0.7042396010246031, "learning_rate": 4.558786665606962e-06, "loss": 1.5099, "step": 8217 }, { "epoch": 0.6289606612582275, "grad_norm": 0.6181167008159544, "learning_rate": 4.5578617964613805e-06, "loss": 1.5016, "step": 8218 }, { "epoch": 0.6290371957752947, "grad_norm": 0.6564849692943674, "learning_rate": 4.556937015925079e-06, "loss": 1.2889, "step": 8219 }, { "epoch": 0.6291137302923618, "grad_norm": 0.622372721528235, "learning_rate": 4.556012324054883e-06, "loss": 1.5628, "step": 8220 }, { "epoch": 0.6291902648094291, "grad_norm": 0.5335792977205825, "learning_rate": 4.555087720907605e-06, "loss": 1.2665, "step": 8221 }, { "epoch": 0.6292667993264962, "grad_norm": 0.5238339758861585, "learning_rate": 4.554163206540059e-06, "loss": 1.0848, "step": 8222 }, { "epoch": 0.6293433338435634, "grad_norm": 0.5218974241345093, "learning_rate": 4.553238781009046e-06, "loss": 1.1263, "step": 8223 }, { "epoch": 0.6294198683606307, "grad_norm": 0.4946631188870799, "learning_rate": 4.552314444371366e-06, "loss": 1.0671, "step": 8224 }, { "epoch": 0.6294964028776978, "grad_norm": 0.5595186082120671, "learning_rate": 4.551390196683817e-06, "loss": 1.3726, "step": 8225 }, { "epoch": 0.629572937394765, "grad_norm": 0.5516255389685418, "learning_rate": 4.550466038003184e-06, "loss": 1.2151, "step": 8226 }, { "epoch": 0.6296494719118322, "grad_norm": 0.7050324203153349, "learning_rate": 4.549541968386252e-06, "loss": 1.6432, "step": 8227 }, { "epoch": 0.6297260064288994, "grad_norm": 0.5123304317053542, "learning_rate": 4.548617987889796e-06, "loss": 1.2136, "step": 8228 }, { "epoch": 0.6298025409459667, "grad_norm": 0.49327440637683834, "learning_rate": 4.5476940965705905e-06, "loss": 1.129, "step": 8229 }, { "epoch": 0.6298790754630338, "grad_norm": 0.4628659315965212, "learning_rate": 4.546770294485401e-06, "loss": 0.9746, "step": 8230 }, { "epoch": 0.629955609980101, "grad_norm": 0.5454137858498289, "learning_rate": 4.545846581690988e-06, "loss": 1.2362, "step": 8231 }, { "epoch": 0.6300321444971683, "grad_norm": 0.6577968342322197, "learning_rate": 4.544922958244109e-06, "loss": 1.6845, "step": 8232 }, { "epoch": 0.6301086790142354, "grad_norm": 0.5624621267712138, "learning_rate": 4.543999424201514e-06, "loss": 1.3052, "step": 8233 }, { "epoch": 0.6301852135313026, "grad_norm": 0.6579281507930912, "learning_rate": 4.543075979619947e-06, "loss": 1.4287, "step": 8234 }, { "epoch": 0.6302617480483698, "grad_norm": 0.5873713197707465, "learning_rate": 4.5421526245561445e-06, "loss": 1.3826, "step": 8235 }, { "epoch": 0.630338282565437, "grad_norm": 0.643428528019067, "learning_rate": 4.5412293590668446e-06, "loss": 1.4905, "step": 8236 }, { "epoch": 0.6304148170825042, "grad_norm": 0.6195987793563357, "learning_rate": 4.540306183208774e-06, "loss": 1.4323, "step": 8237 }, { "epoch": 0.6304913515995714, "grad_norm": 0.5973361724535663, "learning_rate": 4.539383097038655e-06, "loss": 1.4223, "step": 8238 }, { "epoch": 0.6305678861166386, "grad_norm": 0.5505739770195988, "learning_rate": 4.538460100613206e-06, "loss": 1.1136, "step": 8239 }, { "epoch": 0.6306444206337058, "grad_norm": 0.5847257111135891, "learning_rate": 4.537537193989137e-06, "loss": 1.2626, "step": 8240 }, { "epoch": 0.630720955150773, "grad_norm": 0.6088406144399153, "learning_rate": 4.536614377223153e-06, "loss": 1.3887, "step": 8241 }, { "epoch": 0.6307974896678402, "grad_norm": 0.5385618703453758, "learning_rate": 4.535691650371958e-06, "loss": 1.109, "step": 8242 }, { "epoch": 0.6308740241849073, "grad_norm": 0.5788751246136836, "learning_rate": 4.534769013492245e-06, "loss": 1.2523, "step": 8243 }, { "epoch": 0.6309505587019746, "grad_norm": 0.8433099573895529, "learning_rate": 4.5338464666407045e-06, "loss": 1.5057, "step": 8244 }, { "epoch": 0.6310270932190418, "grad_norm": 0.9355758755558895, "learning_rate": 4.53292400987402e-06, "loss": 1.3015, "step": 8245 }, { "epoch": 0.631103627736109, "grad_norm": 0.6195198710211732, "learning_rate": 4.532001643248868e-06, "loss": 1.4331, "step": 8246 }, { "epoch": 0.6311801622531762, "grad_norm": 0.5857338473237725, "learning_rate": 4.531079366821926e-06, "loss": 1.4677, "step": 8247 }, { "epoch": 0.6312566967702434, "grad_norm": 0.6140106850830602, "learning_rate": 4.530157180649858e-06, "loss": 1.4347, "step": 8248 }, { "epoch": 0.6313332312873106, "grad_norm": 0.8115801882657112, "learning_rate": 4.5292350847893275e-06, "loss": 1.4327, "step": 8249 }, { "epoch": 0.6314097658043778, "grad_norm": 0.6549038235368457, "learning_rate": 4.52831307929699e-06, "loss": 1.6139, "step": 8250 }, { "epoch": 0.6314863003214449, "grad_norm": 0.5226748986638715, "learning_rate": 4.527391164229496e-06, "loss": 1.0357, "step": 8251 }, { "epoch": 0.6315628348385122, "grad_norm": 0.741058804518059, "learning_rate": 4.526469339643491e-06, "loss": 1.6879, "step": 8252 }, { "epoch": 0.6316393693555794, "grad_norm": 0.6322087449429736, "learning_rate": 4.525547605595615e-06, "loss": 1.6201, "step": 8253 }, { "epoch": 0.6317159038726465, "grad_norm": 0.7609326680565931, "learning_rate": 4.524625962142501e-06, "loss": 1.6215, "step": 8254 }, { "epoch": 0.6317924383897138, "grad_norm": 0.6297548520539672, "learning_rate": 4.5237044093407815e-06, "loss": 1.5018, "step": 8255 }, { "epoch": 0.631868972906781, "grad_norm": 0.5480332968184299, "learning_rate": 4.522782947247073e-06, "loss": 1.2679, "step": 8256 }, { "epoch": 0.6319455074238481, "grad_norm": 0.6655676383887711, "learning_rate": 4.5218615759179964e-06, "loss": 1.3113, "step": 8257 }, { "epoch": 0.6320220419409154, "grad_norm": 0.9488889113804106, "learning_rate": 4.520940295410164e-06, "loss": 1.8081, "step": 8258 }, { "epoch": 0.6320985764579825, "grad_norm": 0.6616022917640678, "learning_rate": 4.520019105780183e-06, "loss": 1.5318, "step": 8259 }, { "epoch": 0.6321751109750497, "grad_norm": 0.7083430985271866, "learning_rate": 4.519098007084651e-06, "loss": 1.7186, "step": 8260 }, { "epoch": 0.632251645492117, "grad_norm": 0.6001378652292626, "learning_rate": 4.518176999380165e-06, "loss": 1.4128, "step": 8261 }, { "epoch": 0.6323281800091841, "grad_norm": 0.5997314641370582, "learning_rate": 4.517256082723313e-06, "loss": 1.4251, "step": 8262 }, { "epoch": 0.6324047145262514, "grad_norm": 0.6019336908439411, "learning_rate": 4.516335257170681e-06, "loss": 1.4116, "step": 8263 }, { "epoch": 0.6324812490433185, "grad_norm": 0.5400744245326868, "learning_rate": 4.515414522778846e-06, "loss": 1.2939, "step": 8264 }, { "epoch": 0.6325577835603857, "grad_norm": 0.5318518694106268, "learning_rate": 4.51449387960438e-06, "loss": 1.0991, "step": 8265 }, { "epoch": 0.632634318077453, "grad_norm": 0.5888985080184505, "learning_rate": 4.513573327703852e-06, "loss": 1.3488, "step": 8266 }, { "epoch": 0.6327108525945201, "grad_norm": 0.5762453923873537, "learning_rate": 4.512652867133822e-06, "loss": 1.5192, "step": 8267 }, { "epoch": 0.6327873871115873, "grad_norm": 0.7293007764343379, "learning_rate": 4.511732497950844e-06, "loss": 1.1285, "step": 8268 }, { "epoch": 0.6328639216286546, "grad_norm": 0.6283466361456297, "learning_rate": 4.510812220211473e-06, "loss": 1.6556, "step": 8269 }, { "epoch": 0.6329404561457217, "grad_norm": 0.6765278893434256, "learning_rate": 4.50989203397225e-06, "loss": 1.2443, "step": 8270 }, { "epoch": 0.6330169906627889, "grad_norm": 0.5600032001983858, "learning_rate": 4.508971939289716e-06, "loss": 1.3118, "step": 8271 }, { "epoch": 0.6330935251798561, "grad_norm": 0.5356981101283116, "learning_rate": 4.5080519362204024e-06, "loss": 1.1976, "step": 8272 }, { "epoch": 0.6331700596969233, "grad_norm": 0.5117470781226622, "learning_rate": 4.507132024820838e-06, "loss": 1.0557, "step": 8273 }, { "epoch": 0.6332465942139905, "grad_norm": 0.5751733709361111, "learning_rate": 4.506212205147544e-06, "loss": 1.4254, "step": 8274 }, { "epoch": 0.6333231287310577, "grad_norm": 0.5701097508576142, "learning_rate": 4.505292477257039e-06, "loss": 1.3317, "step": 8275 }, { "epoch": 0.6333996632481249, "grad_norm": 1.5058974159332037, "learning_rate": 4.504372841205833e-06, "loss": 1.8101, "step": 8276 }, { "epoch": 0.6334761977651922, "grad_norm": 0.536383639464881, "learning_rate": 4.5034532970504294e-06, "loss": 1.1452, "step": 8277 }, { "epoch": 0.6335527322822593, "grad_norm": 0.5857275722231745, "learning_rate": 4.50253384484733e-06, "loss": 1.3224, "step": 8278 }, { "epoch": 0.6336292667993265, "grad_norm": 0.5637666657374703, "learning_rate": 4.501614484653025e-06, "loss": 1.2357, "step": 8279 }, { "epoch": 0.6337058013163936, "grad_norm": 0.7880519933990473, "learning_rate": 4.500695216524008e-06, "loss": 1.6357, "step": 8280 }, { "epoch": 0.6337823358334609, "grad_norm": 0.6320811903534044, "learning_rate": 4.499776040516759e-06, "loss": 1.5126, "step": 8281 }, { "epoch": 0.6338588703505281, "grad_norm": 0.7443416396503089, "learning_rate": 4.498856956687755e-06, "loss": 1.6109, "step": 8282 }, { "epoch": 0.6339354048675953, "grad_norm": 0.5756694102339551, "learning_rate": 4.497937965093468e-06, "loss": 1.2283, "step": 8283 }, { "epoch": 0.6340119393846625, "grad_norm": 0.6419498385245406, "learning_rate": 4.49701906579036e-06, "loss": 1.292, "step": 8284 }, { "epoch": 0.6340884739017297, "grad_norm": 0.5849457776961697, "learning_rate": 4.496100258834895e-06, "loss": 1.2808, "step": 8285 }, { "epoch": 0.6341650084187969, "grad_norm": 0.5695084148860157, "learning_rate": 4.495181544283525e-06, "loss": 1.2794, "step": 8286 }, { "epoch": 0.6342415429358641, "grad_norm": 0.7045367371529874, "learning_rate": 4.494262922192701e-06, "loss": 1.8311, "step": 8287 }, { "epoch": 0.6343180774529312, "grad_norm": 0.7077275326845306, "learning_rate": 4.493344392618865e-06, "loss": 1.2684, "step": 8288 }, { "epoch": 0.6343946119699985, "grad_norm": 0.5232502586432825, "learning_rate": 4.492425955618452e-06, "loss": 1.1808, "step": 8289 }, { "epoch": 0.6344711464870657, "grad_norm": 0.6077467257991489, "learning_rate": 4.491507611247894e-06, "loss": 1.419, "step": 8290 }, { "epoch": 0.6345476810041328, "grad_norm": 0.6389518121409228, "learning_rate": 4.490589359563619e-06, "loss": 1.4594, "step": 8291 }, { "epoch": 0.6346242155212001, "grad_norm": 0.6751776547024786, "learning_rate": 4.489671200622047e-06, "loss": 1.3792, "step": 8292 }, { "epoch": 0.6347007500382673, "grad_norm": 0.5649325488480238, "learning_rate": 4.488753134479591e-06, "loss": 1.3054, "step": 8293 }, { "epoch": 0.6347772845553344, "grad_norm": 0.5516603506280963, "learning_rate": 4.48783516119266e-06, "loss": 1.346, "step": 8294 }, { "epoch": 0.6348538190724017, "grad_norm": 0.770335079987103, "learning_rate": 4.486917280817657e-06, "loss": 1.5502, "step": 8295 }, { "epoch": 0.6349303535894688, "grad_norm": 0.6834829698807304, "learning_rate": 4.48599949341098e-06, "loss": 1.5957, "step": 8296 }, { "epoch": 0.635006888106536, "grad_norm": 0.5502150235796129, "learning_rate": 4.4850817990290196e-06, "loss": 1.3238, "step": 8297 }, { "epoch": 0.6350834226236033, "grad_norm": 0.6413436095185793, "learning_rate": 4.484164197728163e-06, "loss": 1.676, "step": 8298 }, { "epoch": 0.6351599571406704, "grad_norm": 0.6609500960454341, "learning_rate": 4.483246689564789e-06, "loss": 1.5357, "step": 8299 }, { "epoch": 0.6352364916577377, "grad_norm": 0.6471384094604467, "learning_rate": 4.482329274595271e-06, "loss": 1.3142, "step": 8300 }, { "epoch": 0.6353130261748048, "grad_norm": 0.7680203364721564, "learning_rate": 4.4814119528759804e-06, "loss": 1.4755, "step": 8301 }, { "epoch": 0.635389560691872, "grad_norm": 0.5647584161988722, "learning_rate": 4.48049472446328e-06, "loss": 1.3543, "step": 8302 }, { "epoch": 0.6354660952089393, "grad_norm": 0.5239199847274705, "learning_rate": 4.479577589413526e-06, "loss": 1.1036, "step": 8303 }, { "epoch": 0.6355426297260064, "grad_norm": 0.8877052915557093, "learning_rate": 4.47866054778307e-06, "loss": 1.2472, "step": 8304 }, { "epoch": 0.6356191642430736, "grad_norm": 0.5739013556882784, "learning_rate": 4.4777435996282585e-06, "loss": 1.2461, "step": 8305 }, { "epoch": 0.6356956987601409, "grad_norm": 0.5653590471349136, "learning_rate": 4.476826745005431e-06, "loss": 1.1835, "step": 8306 }, { "epoch": 0.635772233277208, "grad_norm": 0.6436027357215995, "learning_rate": 4.475909983970918e-06, "loss": 1.2874, "step": 8307 }, { "epoch": 0.6358487677942752, "grad_norm": 0.5176414635793806, "learning_rate": 4.474993316581057e-06, "loss": 1.161, "step": 8308 }, { "epoch": 0.6359253023113424, "grad_norm": 0.6419848001216221, "learning_rate": 4.4740767428921635e-06, "loss": 1.429, "step": 8309 }, { "epoch": 0.6360018368284096, "grad_norm": 0.648559047182301, "learning_rate": 4.473160262960556e-06, "loss": 1.5663, "step": 8310 }, { "epoch": 0.6360783713454768, "grad_norm": 0.5810972355378055, "learning_rate": 4.472243876842547e-06, "loss": 1.2842, "step": 8311 }, { "epoch": 0.636154905862544, "grad_norm": 1.744536320232509, "learning_rate": 4.47132758459444e-06, "loss": 1.5282, "step": 8312 }, { "epoch": 0.6362314403796112, "grad_norm": 0.80622127334505, "learning_rate": 4.470411386272537e-06, "loss": 1.3093, "step": 8313 }, { "epoch": 0.6363079748966785, "grad_norm": 0.6095149043586333, "learning_rate": 4.469495281933132e-06, "loss": 1.5225, "step": 8314 }, { "epoch": 0.6363845094137456, "grad_norm": 0.6114544550344647, "learning_rate": 4.468579271632511e-06, "loss": 1.2792, "step": 8315 }, { "epoch": 0.6364610439308128, "grad_norm": 0.9083308605577478, "learning_rate": 4.467663355426958e-06, "loss": 1.447, "step": 8316 }, { "epoch": 0.63653757844788, "grad_norm": 0.5151337055856152, "learning_rate": 4.466747533372748e-06, "loss": 1.1794, "step": 8317 }, { "epoch": 0.6366141129649472, "grad_norm": 0.6104909619628138, "learning_rate": 4.465831805526154e-06, "loss": 1.3829, "step": 8318 }, { "epoch": 0.6366906474820144, "grad_norm": 0.6290903479802676, "learning_rate": 4.464916171943441e-06, "loss": 1.5019, "step": 8319 }, { "epoch": 0.6367671819990816, "grad_norm": 0.6154923390221511, "learning_rate": 4.464000632680867e-06, "loss": 1.3796, "step": 8320 }, { "epoch": 0.6368437165161488, "grad_norm": 0.744566355048515, "learning_rate": 4.4630851877946866e-06, "loss": 1.6827, "step": 8321 }, { "epoch": 0.636920251033216, "grad_norm": 0.8175589690361882, "learning_rate": 4.462169837341146e-06, "loss": 1.3668, "step": 8322 }, { "epoch": 0.6369967855502832, "grad_norm": 0.6109164585284518, "learning_rate": 4.461254581376486e-06, "loss": 1.4914, "step": 8323 }, { "epoch": 0.6370733200673504, "grad_norm": 0.838167739214879, "learning_rate": 4.460339419956948e-06, "loss": 1.7347, "step": 8324 }, { "epoch": 0.6371498545844175, "grad_norm": 0.6333587176735923, "learning_rate": 4.459424353138756e-06, "loss": 1.4299, "step": 8325 }, { "epoch": 0.6372263891014848, "grad_norm": 0.6147156689061865, "learning_rate": 4.458509380978139e-06, "loss": 1.7417, "step": 8326 }, { "epoch": 0.637302923618552, "grad_norm": 0.5555803519703035, "learning_rate": 4.457594503531312e-06, "loss": 1.3012, "step": 8327 }, { "epoch": 0.6373794581356191, "grad_norm": 0.6942603818737462, "learning_rate": 4.4566797208544895e-06, "loss": 1.539, "step": 8328 }, { "epoch": 0.6374559926526864, "grad_norm": 0.9624697843250475, "learning_rate": 4.4557650330038805e-06, "loss": 1.9905, "step": 8329 }, { "epoch": 0.6375325271697535, "grad_norm": 0.6080929996226027, "learning_rate": 4.454850440035683e-06, "loss": 1.2481, "step": 8330 }, { "epoch": 0.6376090616868207, "grad_norm": 0.5952468117977922, "learning_rate": 4.453935942006094e-06, "loss": 1.3375, "step": 8331 }, { "epoch": 0.637685596203888, "grad_norm": 0.6015643441982347, "learning_rate": 4.453021538971302e-06, "loss": 1.4725, "step": 8332 }, { "epoch": 0.6377621307209551, "grad_norm": 0.6505739661396979, "learning_rate": 4.452107230987493e-06, "loss": 1.3957, "step": 8333 }, { "epoch": 0.6378386652380224, "grad_norm": 0.642559742925966, "learning_rate": 4.451193018110842e-06, "loss": 1.5608, "step": 8334 }, { "epoch": 0.6379151997550896, "grad_norm": 0.6516712459173776, "learning_rate": 4.4502789003975216e-06, "loss": 1.6008, "step": 8335 }, { "epoch": 0.6379917342721567, "grad_norm": 0.6712614248267933, "learning_rate": 4.449364877903699e-06, "loss": 1.5484, "step": 8336 }, { "epoch": 0.638068268789224, "grad_norm": 0.8219462621285204, "learning_rate": 4.448450950685533e-06, "loss": 1.7677, "step": 8337 }, { "epoch": 0.6381448033062911, "grad_norm": 0.5824095836293071, "learning_rate": 4.44753711879918e-06, "loss": 1.2247, "step": 8338 }, { "epoch": 0.6382213378233583, "grad_norm": 0.562280176670326, "learning_rate": 4.446623382300785e-06, "loss": 1.229, "step": 8339 }, { "epoch": 0.6382978723404256, "grad_norm": 0.7117113991946054, "learning_rate": 4.445709741246495e-06, "loss": 1.1176, "step": 8340 }, { "epoch": 0.6383744068574927, "grad_norm": 0.5829075015262488, "learning_rate": 4.444796195692446e-06, "loss": 1.438, "step": 8341 }, { "epoch": 0.6384509413745599, "grad_norm": 1.2641198363189214, "learning_rate": 4.443882745694766e-06, "loss": 2.0402, "step": 8342 }, { "epoch": 0.6385274758916272, "grad_norm": 0.7790236332503285, "learning_rate": 4.442969391309583e-06, "loss": 1.172, "step": 8343 }, { "epoch": 0.6386040104086943, "grad_norm": 0.6348193549789358, "learning_rate": 4.442056132593015e-06, "loss": 1.3846, "step": 8344 }, { "epoch": 0.6386805449257615, "grad_norm": 0.7767976313469406, "learning_rate": 4.441142969601175e-06, "loss": 1.6788, "step": 8345 }, { "epoch": 0.6387570794428287, "grad_norm": 0.5372111037411199, "learning_rate": 4.440229902390172e-06, "loss": 1.2536, "step": 8346 }, { "epoch": 0.6388336139598959, "grad_norm": 0.6810829349580306, "learning_rate": 4.439316931016106e-06, "loss": 1.6404, "step": 8347 }, { "epoch": 0.6389101484769631, "grad_norm": 0.535064358245084, "learning_rate": 4.438404055535074e-06, "loss": 1.2401, "step": 8348 }, { "epoch": 0.6389866829940303, "grad_norm": 0.7657383967584337, "learning_rate": 4.437491276003165e-06, "loss": 1.691, "step": 8349 }, { "epoch": 0.6390632175110975, "grad_norm": 0.6424859738777773, "learning_rate": 4.4365785924764605e-06, "loss": 1.5083, "step": 8350 }, { "epoch": 0.6391397520281648, "grad_norm": 1.0582676661352943, "learning_rate": 4.435666005011044e-06, "loss": 1.583, "step": 8351 }, { "epoch": 0.6392162865452319, "grad_norm": 0.6352881626597767, "learning_rate": 4.434753513662983e-06, "loss": 1.2749, "step": 8352 }, { "epoch": 0.6392928210622991, "grad_norm": 0.6678422293569882, "learning_rate": 4.433841118488347e-06, "loss": 1.5113, "step": 8353 }, { "epoch": 0.6393693555793662, "grad_norm": 0.6439262346962547, "learning_rate": 4.432928819543194e-06, "loss": 1.5748, "step": 8354 }, { "epoch": 0.6394458900964335, "grad_norm": 0.5560914301071239, "learning_rate": 4.432016616883578e-06, "loss": 1.1002, "step": 8355 }, { "epoch": 0.6395224246135007, "grad_norm": 0.7479294961961569, "learning_rate": 4.431104510565549e-06, "loss": 1.631, "step": 8356 }, { "epoch": 0.6395989591305679, "grad_norm": 0.4895320423792559, "learning_rate": 4.430192500645149e-06, "loss": 1.0461, "step": 8357 }, { "epoch": 0.6396754936476351, "grad_norm": 0.6064905814882686, "learning_rate": 4.429280587178415e-06, "loss": 1.5349, "step": 8358 }, { "epoch": 0.6397520281647023, "grad_norm": 0.573829936674675, "learning_rate": 4.428368770221377e-06, "loss": 1.2224, "step": 8359 }, { "epoch": 0.6398285626817695, "grad_norm": 0.4926462774059117, "learning_rate": 4.427457049830059e-06, "loss": 1.0368, "step": 8360 }, { "epoch": 0.6399050971988367, "grad_norm": 0.5441653967234706, "learning_rate": 4.426545426060481e-06, "loss": 1.2525, "step": 8361 }, { "epoch": 0.6399816317159038, "grad_norm": 0.6200642574797873, "learning_rate": 4.425633898968658e-06, "loss": 1.473, "step": 8362 }, { "epoch": 0.6400581662329711, "grad_norm": 0.5838610156934027, "learning_rate": 4.424722468610593e-06, "loss": 1.3032, "step": 8363 }, { "epoch": 0.6401347007500383, "grad_norm": 0.7536209423160147, "learning_rate": 4.423811135042288e-06, "loss": 1.5708, "step": 8364 }, { "epoch": 0.6402112352671054, "grad_norm": 0.6354182731897365, "learning_rate": 4.42289989831974e-06, "loss": 1.4346, "step": 8365 }, { "epoch": 0.6402877697841727, "grad_norm": 0.5946796777941057, "learning_rate": 4.421988758498937e-06, "loss": 1.3816, "step": 8366 }, { "epoch": 0.6403643043012398, "grad_norm": 0.6179928875256963, "learning_rate": 4.421077715635862e-06, "loss": 1.5166, "step": 8367 }, { "epoch": 0.640440838818307, "grad_norm": 0.5688095978075471, "learning_rate": 4.420166769786492e-06, "loss": 1.1851, "step": 8368 }, { "epoch": 0.6405173733353743, "grad_norm": 0.5623464561114888, "learning_rate": 4.419255921006799e-06, "loss": 1.184, "step": 8369 }, { "epoch": 0.6405939078524414, "grad_norm": 0.5553693760364761, "learning_rate": 4.418345169352746e-06, "loss": 1.1897, "step": 8370 }, { "epoch": 0.6406704423695087, "grad_norm": 0.5450447307208849, "learning_rate": 4.417434514880296e-06, "loss": 1.2542, "step": 8371 }, { "epoch": 0.6407469768865759, "grad_norm": 0.6376997721915365, "learning_rate": 4.416523957645399e-06, "loss": 1.4449, "step": 8372 }, { "epoch": 0.640823511403643, "grad_norm": 0.7292217409747952, "learning_rate": 4.415613497704004e-06, "loss": 1.1462, "step": 8373 }, { "epoch": 0.6409000459207103, "grad_norm": 0.6536131460997281, "learning_rate": 4.414703135112053e-06, "loss": 1.3027, "step": 8374 }, { "epoch": 0.6409765804377774, "grad_norm": 0.49431264554569015, "learning_rate": 4.4137928699254795e-06, "loss": 1.0275, "step": 8375 }, { "epoch": 0.6410531149548446, "grad_norm": 0.5902178100999005, "learning_rate": 4.412882702200214e-06, "loss": 1.3815, "step": 8376 }, { "epoch": 0.6411296494719119, "grad_norm": 0.5789551766286789, "learning_rate": 4.411972631992181e-06, "loss": 1.2769, "step": 8377 }, { "epoch": 0.641206183988979, "grad_norm": 0.7915449170036231, "learning_rate": 4.411062659357295e-06, "loss": 1.5701, "step": 8378 }, { "epoch": 0.6412827185060462, "grad_norm": 1.1752532303531957, "learning_rate": 4.41015278435147e-06, "loss": 1.2959, "step": 8379 }, { "epoch": 0.6413592530231135, "grad_norm": 0.6135254258894486, "learning_rate": 4.409243007030612e-06, "loss": 1.4286, "step": 8380 }, { "epoch": 0.6414357875401806, "grad_norm": 0.6165089967253732, "learning_rate": 4.408333327450618e-06, "loss": 1.4628, "step": 8381 }, { "epoch": 0.6415123220572478, "grad_norm": 0.5456902651512794, "learning_rate": 4.407423745667383e-06, "loss": 1.3519, "step": 8382 }, { "epoch": 0.641588856574315, "grad_norm": 0.6375033865909673, "learning_rate": 4.406514261736791e-06, "loss": 1.628, "step": 8383 }, { "epoch": 0.6416653910913822, "grad_norm": 0.5300844874348256, "learning_rate": 4.40560487571473e-06, "loss": 1.106, "step": 8384 }, { "epoch": 0.6417419256084494, "grad_norm": 0.5936109919384237, "learning_rate": 4.40469558765707e-06, "loss": 1.3999, "step": 8385 }, { "epoch": 0.6418184601255166, "grad_norm": 0.9723258091878604, "learning_rate": 4.403786397619684e-06, "loss": 1.2143, "step": 8386 }, { "epoch": 0.6418949946425838, "grad_norm": 0.5412658558938833, "learning_rate": 4.402877305658432e-06, "loss": 1.2474, "step": 8387 }, { "epoch": 0.641971529159651, "grad_norm": 0.5547674872757091, "learning_rate": 4.4019683118291736e-06, "loss": 1.3949, "step": 8388 }, { "epoch": 0.6420480636767182, "grad_norm": 1.0652592595348487, "learning_rate": 4.401059416187757e-06, "loss": 1.5261, "step": 8389 }, { "epoch": 0.6421245981937854, "grad_norm": 0.6462051488393437, "learning_rate": 4.400150618790032e-06, "loss": 1.619, "step": 8390 }, { "epoch": 0.6422011327108526, "grad_norm": 0.7496359390924744, "learning_rate": 4.399241919691836e-06, "loss": 1.6388, "step": 8391 }, { "epoch": 0.6422776672279198, "grad_norm": 0.6086770214177659, "learning_rate": 4.398333318949001e-06, "loss": 1.5529, "step": 8392 }, { "epoch": 0.642354201744987, "grad_norm": 0.759432711670508, "learning_rate": 4.397424816617355e-06, "loss": 1.3894, "step": 8393 }, { "epoch": 0.6424307362620542, "grad_norm": 0.6022480320389227, "learning_rate": 4.3965164127527175e-06, "loss": 1.4429, "step": 8394 }, { "epoch": 0.6425072707791214, "grad_norm": 0.5811434225895932, "learning_rate": 4.395608107410907e-06, "loss": 1.2281, "step": 8395 }, { "epoch": 0.6425838052961885, "grad_norm": 0.5821393147569152, "learning_rate": 4.394699900647731e-06, "loss": 1.1444, "step": 8396 }, { "epoch": 0.6426603398132558, "grad_norm": 0.6413999697153785, "learning_rate": 4.393791792518991e-06, "loss": 1.6749, "step": 8397 }, { "epoch": 0.642736874330323, "grad_norm": 0.6247147009648786, "learning_rate": 4.392883783080485e-06, "loss": 1.646, "step": 8398 }, { "epoch": 0.6428134088473901, "grad_norm": 0.505348577116971, "learning_rate": 4.391975872388005e-06, "loss": 1.2627, "step": 8399 }, { "epoch": 0.6428899433644574, "grad_norm": 0.6290620443066873, "learning_rate": 4.391068060497332e-06, "loss": 1.4467, "step": 8400 }, { "epoch": 0.6429664778815246, "grad_norm": 0.6731699879889896, "learning_rate": 4.39016034746425e-06, "loss": 1.6101, "step": 8401 }, { "epoch": 0.6430430123985917, "grad_norm": 0.6071190221515673, "learning_rate": 4.3892527333445264e-06, "loss": 1.5301, "step": 8402 }, { "epoch": 0.643119546915659, "grad_norm": 0.5651683751315717, "learning_rate": 4.388345218193932e-06, "loss": 1.3055, "step": 8403 }, { "epoch": 0.6431960814327261, "grad_norm": 0.6503748243413738, "learning_rate": 4.387437802068224e-06, "loss": 1.6042, "step": 8404 }, { "epoch": 0.6432726159497933, "grad_norm": 0.6195796963249995, "learning_rate": 4.386530485023157e-06, "loss": 1.519, "step": 8405 }, { "epoch": 0.6433491504668606, "grad_norm": 0.5706571037010809, "learning_rate": 4.385623267114482e-06, "loss": 1.4005, "step": 8406 }, { "epoch": 0.6434256849839277, "grad_norm": 0.5832196817039417, "learning_rate": 4.38471614839794e-06, "loss": 1.3241, "step": 8407 }, { "epoch": 0.643502219500995, "grad_norm": 0.6151091172363637, "learning_rate": 4.383809128929266e-06, "loss": 1.3365, "step": 8408 }, { "epoch": 0.6435787540180622, "grad_norm": 0.6477189749000831, "learning_rate": 4.38290220876419e-06, "loss": 1.5595, "step": 8409 }, { "epoch": 0.6436552885351293, "grad_norm": 0.6088223827219692, "learning_rate": 4.381995387958437e-06, "loss": 1.3561, "step": 8410 }, { "epoch": 0.6437318230521966, "grad_norm": 0.6142019835451266, "learning_rate": 4.381088666567724e-06, "loss": 1.5228, "step": 8411 }, { "epoch": 0.6438083575692637, "grad_norm": 0.5971394600349738, "learning_rate": 4.380182044647764e-06, "loss": 1.4479, "step": 8412 }, { "epoch": 0.6438848920863309, "grad_norm": 0.6967562975853208, "learning_rate": 4.37927552225426e-06, "loss": 1.2719, "step": 8413 }, { "epoch": 0.6439614266033982, "grad_norm": 0.6985126544728635, "learning_rate": 4.378369099442915e-06, "loss": 1.2667, "step": 8414 }, { "epoch": 0.6440379611204653, "grad_norm": 0.6236883183310207, "learning_rate": 4.377462776269419e-06, "loss": 1.5948, "step": 8415 }, { "epoch": 0.6441144956375325, "grad_norm": 0.5483787225826499, "learning_rate": 4.376556552789459e-06, "loss": 1.345, "step": 8416 }, { "epoch": 0.6441910301545998, "grad_norm": 0.6083741645784612, "learning_rate": 4.37565042905872e-06, "loss": 1.253, "step": 8417 }, { "epoch": 0.6442675646716669, "grad_norm": 0.5189025062734276, "learning_rate": 4.374744405132873e-06, "loss": 0.9127, "step": 8418 }, { "epoch": 0.6443440991887341, "grad_norm": 0.8315555188096024, "learning_rate": 4.3738384810675895e-06, "loss": 1.5585, "step": 8419 }, { "epoch": 0.6444206337058013, "grad_norm": 0.6294508854676413, "learning_rate": 4.37293265691853e-06, "loss": 1.605, "step": 8420 }, { "epoch": 0.6444971682228685, "grad_norm": 1.4379294285576163, "learning_rate": 4.3720269327413526e-06, "loss": 1.368, "step": 8421 }, { "epoch": 0.6445737027399357, "grad_norm": 0.7393248284402887, "learning_rate": 4.371121308591707e-06, "loss": 1.6928, "step": 8422 }, { "epoch": 0.6446502372570029, "grad_norm": 0.5754858095687889, "learning_rate": 4.370215784525238e-06, "loss": 1.3918, "step": 8423 }, { "epoch": 0.6447267717740701, "grad_norm": 0.49635203643909714, "learning_rate": 4.369310360597583e-06, "loss": 1.1073, "step": 8424 }, { "epoch": 0.6448033062911374, "grad_norm": 0.617555286936728, "learning_rate": 4.368405036864375e-06, "loss": 1.5421, "step": 8425 }, { "epoch": 0.6448798408082045, "grad_norm": 0.8457102107662666, "learning_rate": 4.367499813381239e-06, "loss": 1.7768, "step": 8426 }, { "epoch": 0.6449563753252717, "grad_norm": 0.5251568541555289, "learning_rate": 4.366594690203792e-06, "loss": 1.2208, "step": 8427 }, { "epoch": 0.6450329098423389, "grad_norm": 0.5845357695385047, "learning_rate": 4.3656896673876525e-06, "loss": 1.283, "step": 8428 }, { "epoch": 0.6451094443594061, "grad_norm": 0.9036800269576831, "learning_rate": 4.3647847449884255e-06, "loss": 1.399, "step": 8429 }, { "epoch": 0.6451859788764733, "grad_norm": 0.6080790724250612, "learning_rate": 4.363879923061711e-06, "loss": 1.5423, "step": 8430 }, { "epoch": 0.6452625133935405, "grad_norm": 0.5708727949532861, "learning_rate": 4.362975201663106e-06, "loss": 1.3003, "step": 8431 }, { "epoch": 0.6453390479106077, "grad_norm": 2.239714138615098, "learning_rate": 4.3620705808482e-06, "loss": 1.9899, "step": 8432 }, { "epoch": 0.6454155824276748, "grad_norm": 0.6270203517708803, "learning_rate": 4.361166060672571e-06, "loss": 1.4072, "step": 8433 }, { "epoch": 0.6454921169447421, "grad_norm": 0.48580997318045727, "learning_rate": 4.3602616411918e-06, "loss": 1.1051, "step": 8434 }, { "epoch": 0.6455686514618093, "grad_norm": 0.5178586761671142, "learning_rate": 4.359357322461456e-06, "loss": 1.2224, "step": 8435 }, { "epoch": 0.6456451859788764, "grad_norm": 0.5882146543653844, "learning_rate": 4.358453104537103e-06, "loss": 1.5441, "step": 8436 }, { "epoch": 0.6457217204959437, "grad_norm": 0.5891446947920608, "learning_rate": 4.357548987474297e-06, "loss": 1.5179, "step": 8437 }, { "epoch": 0.6457982550130109, "grad_norm": 2.008105837026623, "learning_rate": 4.356644971328591e-06, "loss": 1.4597, "step": 8438 }, { "epoch": 0.645874789530078, "grad_norm": 0.7319028003266086, "learning_rate": 4.355741056155532e-06, "loss": 1.3475, "step": 8439 }, { "epoch": 0.6459513240471453, "grad_norm": 0.7703236824729034, "learning_rate": 4.354837242010657e-06, "loss": 1.6653, "step": 8440 }, { "epoch": 0.6460278585642124, "grad_norm": 0.5614984115867973, "learning_rate": 4.353933528949501e-06, "loss": 1.2984, "step": 8441 }, { "epoch": 0.6461043930812796, "grad_norm": 1.4405352315686377, "learning_rate": 4.353029917027589e-06, "loss": 1.1285, "step": 8442 }, { "epoch": 0.6461809275983469, "grad_norm": 0.5291299884377818, "learning_rate": 4.352126406300442e-06, "loss": 1.1736, "step": 8443 }, { "epoch": 0.646257462115414, "grad_norm": 0.7388502679009034, "learning_rate": 4.351222996823575e-06, "loss": 1.6599, "step": 8444 }, { "epoch": 0.6463339966324813, "grad_norm": 0.7464935827215078, "learning_rate": 4.3503196886524965e-06, "loss": 1.4593, "step": 8445 }, { "epoch": 0.6464105311495485, "grad_norm": 0.5946735261934843, "learning_rate": 4.349416481842709e-06, "loss": 1.4465, "step": 8446 }, { "epoch": 0.6464870656666156, "grad_norm": 0.6732575935331164, "learning_rate": 4.348513376449705e-06, "loss": 1.3634, "step": 8447 }, { "epoch": 0.6465636001836829, "grad_norm": 0.715392084547733, "learning_rate": 4.3476103725289785e-06, "loss": 1.1634, "step": 8448 }, { "epoch": 0.64664013470075, "grad_norm": 0.5821072085552903, "learning_rate": 4.346707470136007e-06, "loss": 1.2264, "step": 8449 }, { "epoch": 0.6467166692178172, "grad_norm": 0.6261392118675699, "learning_rate": 4.3458046693262736e-06, "loss": 1.5351, "step": 8450 }, { "epoch": 0.6467932037348845, "grad_norm": 0.621131201888078, "learning_rate": 4.344901970155245e-06, "loss": 1.5743, "step": 8451 }, { "epoch": 0.6468697382519516, "grad_norm": 0.5079052913342097, "learning_rate": 4.343999372678389e-06, "loss": 1.1138, "step": 8452 }, { "epoch": 0.6469462727690188, "grad_norm": 0.4925036792610572, "learning_rate": 4.3430968769511605e-06, "loss": 0.9326, "step": 8453 }, { "epoch": 0.6470228072860861, "grad_norm": 0.6079822226360688, "learning_rate": 4.342194483029014e-06, "loss": 1.3913, "step": 8454 }, { "epoch": 0.6470993418031532, "grad_norm": 0.5414276803985782, "learning_rate": 4.341292190967393e-06, "loss": 1.3215, "step": 8455 }, { "epoch": 0.6471758763202204, "grad_norm": 1.4053916529190609, "learning_rate": 4.340390000821741e-06, "loss": 1.2533, "step": 8456 }, { "epoch": 0.6472524108372876, "grad_norm": 1.3922212710365647, "learning_rate": 4.339487912647486e-06, "loss": 1.738, "step": 8457 }, { "epoch": 0.6473289453543548, "grad_norm": 0.5392795286315041, "learning_rate": 4.33858592650006e-06, "loss": 1.1836, "step": 8458 }, { "epoch": 0.647405479871422, "grad_norm": 0.6369669034069917, "learning_rate": 4.337684042434879e-06, "loss": 1.6192, "step": 8459 }, { "epoch": 0.6474820143884892, "grad_norm": 0.5373111428629886, "learning_rate": 4.336782260507359e-06, "loss": 1.2121, "step": 8460 }, { "epoch": 0.6475585489055564, "grad_norm": 0.6177664022673981, "learning_rate": 4.335880580772911e-06, "loss": 1.3984, "step": 8461 }, { "epoch": 0.6476350834226237, "grad_norm": 0.8001581522497304, "learning_rate": 4.334979003286935e-06, "loss": 1.6384, "step": 8462 }, { "epoch": 0.6477116179396908, "grad_norm": 0.559088308268215, "learning_rate": 4.334077528104827e-06, "loss": 1.2644, "step": 8463 }, { "epoch": 0.647788152456758, "grad_norm": 0.7663307087027407, "learning_rate": 4.333176155281974e-06, "loss": 1.2498, "step": 8464 }, { "epoch": 0.6478646869738252, "grad_norm": 0.586440840114652, "learning_rate": 4.332274884873763e-06, "loss": 1.4062, "step": 8465 }, { "epoch": 0.6479412214908924, "grad_norm": 0.6146820985999816, "learning_rate": 4.331373716935565e-06, "loss": 1.527, "step": 8466 }, { "epoch": 0.6480177560079596, "grad_norm": 0.4792893270983208, "learning_rate": 4.330472651522757e-06, "loss": 0.9841, "step": 8467 }, { "epoch": 0.6480942905250268, "grad_norm": 0.6661659030757643, "learning_rate": 4.3295716886907e-06, "loss": 1.3991, "step": 8468 }, { "epoch": 0.648170825042094, "grad_norm": 0.6061396316365802, "learning_rate": 4.328670828494752e-06, "loss": 1.4434, "step": 8469 }, { "epoch": 0.6482473595591611, "grad_norm": 0.5919516549248178, "learning_rate": 4.327770070990265e-06, "loss": 1.3642, "step": 8470 }, { "epoch": 0.6483238940762284, "grad_norm": 0.731975011455349, "learning_rate": 4.3268694162325834e-06, "loss": 1.3712, "step": 8471 }, { "epoch": 0.6484004285932956, "grad_norm": 0.6140950056250378, "learning_rate": 4.325968864277047e-06, "loss": 1.3418, "step": 8472 }, { "epoch": 0.6484769631103627, "grad_norm": 0.5816384443572052, "learning_rate": 4.325068415178987e-06, "loss": 1.3387, "step": 8473 }, { "epoch": 0.64855349762743, "grad_norm": 0.9856585790084699, "learning_rate": 4.324168068993731e-06, "loss": 1.5154, "step": 8474 }, { "epoch": 0.6486300321444972, "grad_norm": 3.1258827326032805, "learning_rate": 4.3232678257766e-06, "loss": 1.5094, "step": 8475 }, { "epoch": 0.6487065666615643, "grad_norm": 0.6495743800860028, "learning_rate": 4.322367685582903e-06, "loss": 1.6637, "step": 8476 }, { "epoch": 0.6487831011786316, "grad_norm": 0.6007512100184211, "learning_rate": 4.321467648467953e-06, "loss": 1.4708, "step": 8477 }, { "epoch": 0.6488596356956987, "grad_norm": 0.5232025078320517, "learning_rate": 4.320567714487048e-06, "loss": 1.0991, "step": 8478 }, { "epoch": 0.648936170212766, "grad_norm": 0.59547009934694, "learning_rate": 4.319667883695484e-06, "loss": 1.5861, "step": 8479 }, { "epoch": 0.6490127047298332, "grad_norm": 0.651949157851718, "learning_rate": 4.3187681561485486e-06, "loss": 1.3079, "step": 8480 }, { "epoch": 0.6490892392469003, "grad_norm": 0.5831065221597755, "learning_rate": 4.317868531901523e-06, "loss": 1.128, "step": 8481 }, { "epoch": 0.6491657737639676, "grad_norm": 0.5568200716820431, "learning_rate": 4.316969011009681e-06, "loss": 1.2763, "step": 8482 }, { "epoch": 0.6492423082810348, "grad_norm": 0.5324064335255182, "learning_rate": 4.316069593528297e-06, "loss": 1.1491, "step": 8483 }, { "epoch": 0.6493188427981019, "grad_norm": 0.568289337418916, "learning_rate": 4.31517027951263e-06, "loss": 1.3215, "step": 8484 }, { "epoch": 0.6493953773151692, "grad_norm": 0.5657175322175554, "learning_rate": 4.314271069017939e-06, "loss": 1.317, "step": 8485 }, { "epoch": 0.6494719118322363, "grad_norm": 0.5560086361841906, "learning_rate": 4.313371962099473e-06, "loss": 1.3401, "step": 8486 }, { "epoch": 0.6495484463493035, "grad_norm": 0.6731347393598828, "learning_rate": 4.3124729588124725e-06, "loss": 1.8497, "step": 8487 }, { "epoch": 0.6496249808663708, "grad_norm": 0.5774941466954474, "learning_rate": 4.311574059212181e-06, "loss": 1.4936, "step": 8488 }, { "epoch": 0.6497015153834379, "grad_norm": 0.45756657182267296, "learning_rate": 4.310675263353827e-06, "loss": 0.999, "step": 8489 }, { "epoch": 0.6497780499005051, "grad_norm": 0.6594748954502806, "learning_rate": 4.309776571292634e-06, "loss": 1.2031, "step": 8490 }, { "epoch": 0.6498545844175724, "grad_norm": 0.7126940526731917, "learning_rate": 4.308877983083821e-06, "loss": 1.7311, "step": 8491 }, { "epoch": 0.6499311189346395, "grad_norm": 0.4896786236386408, "learning_rate": 4.307979498782602e-06, "loss": 1.014, "step": 8492 }, { "epoch": 0.6500076534517067, "grad_norm": 0.9381966055518863, "learning_rate": 4.30708111844418e-06, "loss": 1.5038, "step": 8493 }, { "epoch": 0.6500841879687739, "grad_norm": 0.624948044449014, "learning_rate": 4.306182842123754e-06, "loss": 1.5881, "step": 8494 }, { "epoch": 0.6501607224858411, "grad_norm": 0.6248509915161261, "learning_rate": 4.3052846698765184e-06, "loss": 1.4867, "step": 8495 }, { "epoch": 0.6502372570029084, "grad_norm": 0.5659166682616863, "learning_rate": 4.304386601757659e-06, "loss": 1.3781, "step": 8496 }, { "epoch": 0.6503137915199755, "grad_norm": 0.5930240593753647, "learning_rate": 4.303488637822355e-06, "loss": 1.4031, "step": 8497 }, { "epoch": 0.6503903260370427, "grad_norm": 0.6254629646731531, "learning_rate": 4.302590778125779e-06, "loss": 1.504, "step": 8498 }, { "epoch": 0.6504668605541098, "grad_norm": 0.5680138083624422, "learning_rate": 4.301693022723103e-06, "loss": 1.2304, "step": 8499 }, { "epoch": 0.6505433950711771, "grad_norm": 0.5482320151142475, "learning_rate": 4.300795371669484e-06, "loss": 1.281, "step": 8500 }, { "epoch": 0.6506199295882443, "grad_norm": 0.7601561959265105, "learning_rate": 4.2998978250200765e-06, "loss": 1.6855, "step": 8501 }, { "epoch": 0.6506964641053115, "grad_norm": 0.6014309338904404, "learning_rate": 4.299000382830029e-06, "loss": 1.4242, "step": 8502 }, { "epoch": 0.6507729986223787, "grad_norm": 0.5293303087655102, "learning_rate": 4.298103045154484e-06, "loss": 1.2069, "step": 8503 }, { "epoch": 0.6508495331394459, "grad_norm": 0.6105012754557119, "learning_rate": 4.297205812048573e-06, "loss": 1.41, "step": 8504 }, { "epoch": 0.6509260676565131, "grad_norm": 0.5217076050414761, "learning_rate": 4.296308683567429e-06, "loss": 1.1789, "step": 8505 }, { "epoch": 0.6510026021735803, "grad_norm": 0.5953985785813384, "learning_rate": 4.295411659766173e-06, "loss": 1.551, "step": 8506 }, { "epoch": 0.6510791366906474, "grad_norm": 0.6601811929194424, "learning_rate": 4.29451474069992e-06, "loss": 1.6384, "step": 8507 }, { "epoch": 0.6511556712077147, "grad_norm": 0.7004607815891439, "learning_rate": 4.29361792642378e-06, "loss": 1.4698, "step": 8508 }, { "epoch": 0.6512322057247819, "grad_norm": 0.5994674240472088, "learning_rate": 4.292721216992854e-06, "loss": 1.3997, "step": 8509 }, { "epoch": 0.651308740241849, "grad_norm": 0.6077959172768277, "learning_rate": 4.291824612462242e-06, "loss": 1.2131, "step": 8510 }, { "epoch": 0.6513852747589163, "grad_norm": 0.6596473301549869, "learning_rate": 4.290928112887032e-06, "loss": 1.3937, "step": 8511 }, { "epoch": 0.6514618092759835, "grad_norm": 0.5749460430999892, "learning_rate": 4.290031718322308e-06, "loss": 1.3951, "step": 8512 }, { "epoch": 0.6515383437930506, "grad_norm": 0.598279930094556, "learning_rate": 4.289135428823147e-06, "loss": 1.4076, "step": 8513 }, { "epoch": 0.6516148783101179, "grad_norm": 0.6960627990306997, "learning_rate": 4.28823924444462e-06, "loss": 1.4456, "step": 8514 }, { "epoch": 0.651691412827185, "grad_norm": 0.9484096673573248, "learning_rate": 4.287343165241791e-06, "loss": 1.6513, "step": 8515 }, { "epoch": 0.6517679473442523, "grad_norm": 0.6280918760282725, "learning_rate": 4.286447191269717e-06, "loss": 1.5657, "step": 8516 }, { "epoch": 0.6518444818613195, "grad_norm": 0.5779714398230426, "learning_rate": 4.285551322583451e-06, "loss": 1.2783, "step": 8517 }, { "epoch": 0.6519210163783866, "grad_norm": 0.6415014731160412, "learning_rate": 4.284655559238037e-06, "loss": 1.8335, "step": 8518 }, { "epoch": 0.6519975508954539, "grad_norm": 0.6668011490324889, "learning_rate": 4.283759901288513e-06, "loss": 1.3733, "step": 8519 }, { "epoch": 0.6520740854125211, "grad_norm": 0.9068444442482315, "learning_rate": 4.282864348789909e-06, "loss": 1.7518, "step": 8520 }, { "epoch": 0.6521506199295882, "grad_norm": 0.6499705563051732, "learning_rate": 4.281968901797256e-06, "loss": 1.4906, "step": 8521 }, { "epoch": 0.6522271544466555, "grad_norm": 0.6576970288481911, "learning_rate": 4.281073560365569e-06, "loss": 1.2898, "step": 8522 }, { "epoch": 0.6523036889637226, "grad_norm": 0.5252212357814952, "learning_rate": 4.28017832454986e-06, "loss": 1.1697, "step": 8523 }, { "epoch": 0.6523802234807898, "grad_norm": 0.6783123854480542, "learning_rate": 4.279283194405136e-06, "loss": 1.4199, "step": 8524 }, { "epoch": 0.6524567579978571, "grad_norm": 0.6613792834670786, "learning_rate": 4.278388169986397e-06, "loss": 1.4771, "step": 8525 }, { "epoch": 0.6525332925149242, "grad_norm": 0.5019225390349161, "learning_rate": 4.277493251348635e-06, "loss": 1.1925, "step": 8526 }, { "epoch": 0.6526098270319914, "grad_norm": 0.5657239388312554, "learning_rate": 4.276598438546837e-06, "loss": 1.3478, "step": 8527 }, { "epoch": 0.6526863615490587, "grad_norm": 0.6809215649393363, "learning_rate": 4.275703731635984e-06, "loss": 1.4554, "step": 8528 }, { "epoch": 0.6527628960661258, "grad_norm": 0.6367828129333207, "learning_rate": 4.274809130671048e-06, "loss": 1.3671, "step": 8529 }, { "epoch": 0.652839430583193, "grad_norm": 0.771916366285792, "learning_rate": 4.273914635706996e-06, "loss": 1.6932, "step": 8530 }, { "epoch": 0.6529159651002602, "grad_norm": 0.6390698853795661, "learning_rate": 4.273020246798787e-06, "loss": 1.5978, "step": 8531 }, { "epoch": 0.6529924996173274, "grad_norm": 0.5622208415228923, "learning_rate": 4.272125964001379e-06, "loss": 1.2365, "step": 8532 }, { "epoch": 0.6530690341343947, "grad_norm": 0.5639324187617216, "learning_rate": 4.271231787369717e-06, "loss": 1.464, "step": 8533 }, { "epoch": 0.6531455686514618, "grad_norm": 0.5828706593511254, "learning_rate": 4.270337716958743e-06, "loss": 1.3458, "step": 8534 }, { "epoch": 0.653222103168529, "grad_norm": 0.5671299407695138, "learning_rate": 4.269443752823388e-06, "loss": 1.2843, "step": 8535 }, { "epoch": 0.6532986376855962, "grad_norm": 0.6257074763173961, "learning_rate": 4.268549895018584e-06, "loss": 1.6934, "step": 8536 }, { "epoch": 0.6533751722026634, "grad_norm": 0.6048271019289607, "learning_rate": 4.267656143599248e-06, "loss": 1.2514, "step": 8537 }, { "epoch": 0.6534517067197306, "grad_norm": 0.5376198669361945, "learning_rate": 4.2667624986203e-06, "loss": 1.222, "step": 8538 }, { "epoch": 0.6535282412367978, "grad_norm": 0.5639367701047954, "learning_rate": 4.265868960136645e-06, "loss": 1.3653, "step": 8539 }, { "epoch": 0.653604775753865, "grad_norm": 0.5590927070590715, "learning_rate": 4.264975528203185e-06, "loss": 1.1377, "step": 8540 }, { "epoch": 0.6536813102709322, "grad_norm": 0.6464545334088717, "learning_rate": 4.2640822028748165e-06, "loss": 1.7116, "step": 8541 }, { "epoch": 0.6537578447879994, "grad_norm": 0.6919331578625502, "learning_rate": 4.263188984206424e-06, "loss": 1.6419, "step": 8542 }, { "epoch": 0.6538343793050666, "grad_norm": 0.8133555968293076, "learning_rate": 4.262295872252893e-06, "loss": 1.29, "step": 8543 }, { "epoch": 0.6539109138221337, "grad_norm": 0.5657387534085668, "learning_rate": 4.2614028670691e-06, "loss": 1.3916, "step": 8544 }, { "epoch": 0.653987448339201, "grad_norm": 0.5269365411296246, "learning_rate": 4.260509968709911e-06, "loss": 1.0818, "step": 8545 }, { "epoch": 0.6540639828562682, "grad_norm": 0.5392108667617309, "learning_rate": 4.25961717723019e-06, "loss": 1.1981, "step": 8546 }, { "epoch": 0.6541405173733353, "grad_norm": 0.5427672130165341, "learning_rate": 4.258724492684793e-06, "loss": 1.3247, "step": 8547 }, { "epoch": 0.6542170518904026, "grad_norm": 0.5800867786368383, "learning_rate": 4.257831915128566e-06, "loss": 1.3054, "step": 8548 }, { "epoch": 0.6542935864074698, "grad_norm": 0.5940498139988583, "learning_rate": 4.256939444616357e-06, "loss": 1.2647, "step": 8549 }, { "epoch": 0.654370120924537, "grad_norm": 0.6161831181551262, "learning_rate": 4.256047081202999e-06, "loss": 1.6117, "step": 8550 }, { "epoch": 0.6544466554416042, "grad_norm": 0.5301590685859413, "learning_rate": 4.255154824943321e-06, "loss": 1.1838, "step": 8551 }, { "epoch": 0.6545231899586713, "grad_norm": 0.6206189426980819, "learning_rate": 4.254262675892147e-06, "loss": 1.3957, "step": 8552 }, { "epoch": 0.6545997244757386, "grad_norm": 0.6454129237374547, "learning_rate": 4.2533706341042916e-06, "loss": 1.6142, "step": 8553 }, { "epoch": 0.6546762589928058, "grad_norm": 0.5648399566748471, "learning_rate": 4.252478699634567e-06, "loss": 1.3468, "step": 8554 }, { "epoch": 0.6547527935098729, "grad_norm": 0.6647857037575448, "learning_rate": 4.251586872537775e-06, "loss": 1.3605, "step": 8555 }, { "epoch": 0.6548293280269402, "grad_norm": 0.5161411959746969, "learning_rate": 4.250695152868713e-06, "loss": 1.1458, "step": 8556 }, { "epoch": 0.6549058625440074, "grad_norm": 0.5769097156852268, "learning_rate": 4.24980354068217e-06, "loss": 1.406, "step": 8557 }, { "epoch": 0.6549823970610745, "grad_norm": 0.8229001052270288, "learning_rate": 4.248912036032929e-06, "loss": 1.5225, "step": 8558 }, { "epoch": 0.6550589315781418, "grad_norm": 0.5608503837192842, "learning_rate": 4.248020638975767e-06, "loss": 1.3378, "step": 8559 }, { "epoch": 0.6551354660952089, "grad_norm": 0.6203937686397194, "learning_rate": 4.2471293495654555e-06, "loss": 1.3531, "step": 8560 }, { "epoch": 0.6552120006122761, "grad_norm": 0.6115356821988938, "learning_rate": 4.246238167856756e-06, "loss": 1.3946, "step": 8561 }, { "epoch": 0.6552885351293434, "grad_norm": 0.5915258157446086, "learning_rate": 4.245347093904427e-06, "loss": 1.2693, "step": 8562 }, { "epoch": 0.6553650696464105, "grad_norm": 0.8071983673648282, "learning_rate": 4.244456127763218e-06, "loss": 1.5877, "step": 8563 }, { "epoch": 0.6554416041634777, "grad_norm": 0.8531794284690453, "learning_rate": 4.243565269487872e-06, "loss": 1.4393, "step": 8564 }, { "epoch": 0.6555181386805449, "grad_norm": 0.5568948602720286, "learning_rate": 4.242674519133127e-06, "loss": 1.3989, "step": 8565 }, { "epoch": 0.6555946731976121, "grad_norm": 0.5768437452411019, "learning_rate": 4.241783876753713e-06, "loss": 1.3688, "step": 8566 }, { "epoch": 0.6556712077146793, "grad_norm": 0.566198485668061, "learning_rate": 4.240893342404353e-06, "loss": 1.3782, "step": 8567 }, { "epoch": 0.6557477422317465, "grad_norm": 0.7491580984610495, "learning_rate": 4.240002916139766e-06, "loss": 1.7222, "step": 8568 }, { "epoch": 0.6558242767488137, "grad_norm": 0.594117625806602, "learning_rate": 4.2391125980146596e-06, "loss": 1.2262, "step": 8569 }, { "epoch": 0.655900811265881, "grad_norm": 0.5588641231330382, "learning_rate": 4.238222388083739e-06, "loss": 1.1494, "step": 8570 }, { "epoch": 0.6559773457829481, "grad_norm": 0.49394716303302433, "learning_rate": 4.237332286401703e-06, "loss": 1.02, "step": 8571 }, { "epoch": 0.6560538803000153, "grad_norm": 0.6103731951536827, "learning_rate": 4.23644229302324e-06, "loss": 1.481, "step": 8572 }, { "epoch": 0.6561304148170825, "grad_norm": 0.5775044298710129, "learning_rate": 4.235552408003036e-06, "loss": 1.3884, "step": 8573 }, { "epoch": 0.6562069493341497, "grad_norm": 0.7409327346303587, "learning_rate": 4.234662631395765e-06, "loss": 1.5782, "step": 8574 }, { "epoch": 0.6562834838512169, "grad_norm": 0.6922124585240444, "learning_rate": 4.233772963256099e-06, "loss": 1.5799, "step": 8575 }, { "epoch": 0.6563600183682841, "grad_norm": 0.5174685972584386, "learning_rate": 4.2328834036387024e-06, "loss": 1.2623, "step": 8576 }, { "epoch": 0.6564365528853513, "grad_norm": 0.6809371779507146, "learning_rate": 4.231993952598233e-06, "loss": 1.6525, "step": 8577 }, { "epoch": 0.6565130874024185, "grad_norm": 0.622089812851541, "learning_rate": 4.23110461018934e-06, "loss": 1.4718, "step": 8578 }, { "epoch": 0.6565896219194857, "grad_norm": 0.5963759438825911, "learning_rate": 4.230215376466669e-06, "loss": 1.5064, "step": 8579 }, { "epoch": 0.6566661564365529, "grad_norm": 0.6068047947160768, "learning_rate": 4.229326251484855e-06, "loss": 1.3444, "step": 8580 }, { "epoch": 0.65674269095362, "grad_norm": 0.5636296574880768, "learning_rate": 4.228437235298527e-06, "loss": 1.2677, "step": 8581 }, { "epoch": 0.6568192254706873, "grad_norm": 1.1615825880140032, "learning_rate": 4.227548327962314e-06, "loss": 1.5984, "step": 8582 }, { "epoch": 0.6568957599877545, "grad_norm": 0.7134685967705278, "learning_rate": 4.22665952953083e-06, "loss": 1.4845, "step": 8583 }, { "epoch": 0.6569722945048216, "grad_norm": 0.6143466445794429, "learning_rate": 4.225770840058686e-06, "loss": 1.3236, "step": 8584 }, { "epoch": 0.6570488290218889, "grad_norm": 0.759958888534844, "learning_rate": 4.2248822596004855e-06, "loss": 1.3124, "step": 8585 }, { "epoch": 0.6571253635389561, "grad_norm": 0.5115237956768929, "learning_rate": 4.223993788210825e-06, "loss": 1.0587, "step": 8586 }, { "epoch": 0.6572018980560232, "grad_norm": 0.5839489796231186, "learning_rate": 4.223105425944296e-06, "loss": 1.1816, "step": 8587 }, { "epoch": 0.6572784325730905, "grad_norm": 0.5962456614326216, "learning_rate": 4.222217172855482e-06, "loss": 1.4263, "step": 8588 }, { "epoch": 0.6573549670901576, "grad_norm": 0.8725905846421016, "learning_rate": 4.221329028998958e-06, "loss": 1.1705, "step": 8589 }, { "epoch": 0.6574315016072249, "grad_norm": 0.7997857736531299, "learning_rate": 4.220440994429297e-06, "loss": 1.5301, "step": 8590 }, { "epoch": 0.6575080361242921, "grad_norm": 0.552890509576049, "learning_rate": 4.21955306920106e-06, "loss": 1.3582, "step": 8591 }, { "epoch": 0.6575845706413592, "grad_norm": 1.4881065363018935, "learning_rate": 4.218665253368804e-06, "loss": 1.2536, "step": 8592 }, { "epoch": 0.6576611051584265, "grad_norm": 0.5972217901782649, "learning_rate": 4.2177775469870815e-06, "loss": 1.4837, "step": 8593 }, { "epoch": 0.6577376396754937, "grad_norm": 0.7182388119410001, "learning_rate": 4.216889950110436e-06, "loss": 1.7416, "step": 8594 }, { "epoch": 0.6578141741925608, "grad_norm": 0.9107333778114679, "learning_rate": 4.2160024627933994e-06, "loss": 1.1879, "step": 8595 }, { "epoch": 0.6578907087096281, "grad_norm": 0.5835870367539138, "learning_rate": 4.215115085090506e-06, "loss": 1.0696, "step": 8596 }, { "epoch": 0.6579672432266952, "grad_norm": 0.6644905398480596, "learning_rate": 4.214227817056277e-06, "loss": 1.5565, "step": 8597 }, { "epoch": 0.6580437777437624, "grad_norm": 0.6343633042523643, "learning_rate": 4.21334065874523e-06, "loss": 1.2355, "step": 8598 }, { "epoch": 0.6581203122608297, "grad_norm": 0.5727825634996737, "learning_rate": 4.212453610211874e-06, "loss": 1.2032, "step": 8599 }, { "epoch": 0.6581968467778968, "grad_norm": 0.541653281040907, "learning_rate": 4.211566671510712e-06, "loss": 1.0803, "step": 8600 }, { "epoch": 0.658273381294964, "grad_norm": 0.5823965818972016, "learning_rate": 4.2106798426962405e-06, "loss": 1.2266, "step": 8601 }, { "epoch": 0.6583499158120312, "grad_norm": 0.6179051577548936, "learning_rate": 4.209793123822947e-06, "loss": 1.3554, "step": 8602 }, { "epoch": 0.6584264503290984, "grad_norm": 0.7642689076742015, "learning_rate": 4.208906514945316e-06, "loss": 1.9289, "step": 8603 }, { "epoch": 0.6585029848461657, "grad_norm": 0.5307885033079125, "learning_rate": 4.208020016117824e-06, "loss": 1.1258, "step": 8604 }, { "epoch": 0.6585795193632328, "grad_norm": 0.6443209080731318, "learning_rate": 4.2071336273949385e-06, "loss": 1.5349, "step": 8605 }, { "epoch": 0.6586560538803, "grad_norm": 0.6346738943540414, "learning_rate": 4.206247348831124e-06, "loss": 1.6243, "step": 8606 }, { "epoch": 0.6587325883973673, "grad_norm": 0.791695847625567, "learning_rate": 4.205361180480833e-06, "loss": 1.3969, "step": 8607 }, { "epoch": 0.6588091229144344, "grad_norm": 0.5723212767444006, "learning_rate": 4.204475122398517e-06, "loss": 1.3438, "step": 8608 }, { "epoch": 0.6588856574315016, "grad_norm": 0.8117366885307533, "learning_rate": 4.203589174638617e-06, "loss": 1.2239, "step": 8609 }, { "epoch": 0.6589621919485688, "grad_norm": 0.6278486845455709, "learning_rate": 4.202703337255568e-06, "loss": 1.6433, "step": 8610 }, { "epoch": 0.659038726465636, "grad_norm": 0.5165832120386638, "learning_rate": 4.2018176103038e-06, "loss": 1.2437, "step": 8611 }, { "epoch": 0.6591152609827032, "grad_norm": 0.6163351928738594, "learning_rate": 4.200931993837733e-06, "loss": 1.4595, "step": 8612 }, { "epoch": 0.6591917954997704, "grad_norm": 1.2386608882956145, "learning_rate": 4.200046487911781e-06, "loss": 1.6908, "step": 8613 }, { "epoch": 0.6592683300168376, "grad_norm": 0.5926267754960453, "learning_rate": 4.199161092580354e-06, "loss": 1.4014, "step": 8614 }, { "epoch": 0.6593448645339048, "grad_norm": 0.7932727405515663, "learning_rate": 4.198275807897853e-06, "loss": 1.5416, "step": 8615 }, { "epoch": 0.659421399050972, "grad_norm": 0.6208929255011502, "learning_rate": 4.197390633918673e-06, "loss": 1.0755, "step": 8616 }, { "epoch": 0.6594979335680392, "grad_norm": 0.5870515860204003, "learning_rate": 4.196505570697202e-06, "loss": 1.3919, "step": 8617 }, { "epoch": 0.6595744680851063, "grad_norm": 0.5861008253684955, "learning_rate": 4.19562061828782e-06, "loss": 1.5349, "step": 8618 }, { "epoch": 0.6596510026021736, "grad_norm": 0.5802783881386433, "learning_rate": 4.1947357767448994e-06, "loss": 1.4608, "step": 8619 }, { "epoch": 0.6597275371192408, "grad_norm": 0.4963609189939164, "learning_rate": 4.193851046122811e-06, "loss": 0.9525, "step": 8620 }, { "epoch": 0.6598040716363079, "grad_norm": 0.5593141602353497, "learning_rate": 4.1929664264759126e-06, "loss": 1.2553, "step": 8621 }, { "epoch": 0.6598806061533752, "grad_norm": 0.5951385890653187, "learning_rate": 4.192081917858561e-06, "loss": 1.4124, "step": 8622 }, { "epoch": 0.6599571406704424, "grad_norm": 0.6304982583742829, "learning_rate": 4.1911975203251e-06, "loss": 1.3642, "step": 8623 }, { "epoch": 0.6600336751875095, "grad_norm": 0.6852403221630453, "learning_rate": 4.190313233929869e-06, "loss": 1.6588, "step": 8624 }, { "epoch": 0.6601102097045768, "grad_norm": 0.6530595658129911, "learning_rate": 4.189429058727204e-06, "loss": 1.5384, "step": 8625 }, { "epoch": 0.6601867442216439, "grad_norm": 0.6742197994100961, "learning_rate": 4.18854499477143e-06, "loss": 1.4851, "step": 8626 }, { "epoch": 0.6602632787387112, "grad_norm": 0.5238250520991479, "learning_rate": 4.187661042116868e-06, "loss": 1.21, "step": 8627 }, { "epoch": 0.6603398132557784, "grad_norm": 0.5994826642976655, "learning_rate": 4.1867772008178295e-06, "loss": 1.4634, "step": 8628 }, { "epoch": 0.6604163477728455, "grad_norm": 0.48743565809561334, "learning_rate": 4.185893470928619e-06, "loss": 0.9891, "step": 8629 }, { "epoch": 0.6604928822899128, "grad_norm": 0.527551305052199, "learning_rate": 4.185009852503538e-06, "loss": 1.0269, "step": 8630 }, { "epoch": 0.66056941680698, "grad_norm": 0.5420811942524515, "learning_rate": 4.184126345596877e-06, "loss": 1.1235, "step": 8631 }, { "epoch": 0.6606459513240471, "grad_norm": 0.5778601613554297, "learning_rate": 4.183242950262921e-06, "loss": 1.3552, "step": 8632 }, { "epoch": 0.6607224858411144, "grad_norm": 0.6264126827323051, "learning_rate": 4.18235966655595e-06, "loss": 1.482, "step": 8633 }, { "epoch": 0.6607990203581815, "grad_norm": 0.6292573566830579, "learning_rate": 4.181476494530234e-06, "loss": 1.595, "step": 8634 }, { "epoch": 0.6608755548752487, "grad_norm": 0.5591825953450443, "learning_rate": 4.180593434240038e-06, "loss": 1.1774, "step": 8635 }, { "epoch": 0.660952089392316, "grad_norm": 0.5374812996449062, "learning_rate": 4.179710485739622e-06, "loss": 1.3249, "step": 8636 }, { "epoch": 0.6610286239093831, "grad_norm": 0.5592419154125989, "learning_rate": 4.178827649083235e-06, "loss": 1.1825, "step": 8637 }, { "epoch": 0.6611051584264503, "grad_norm": 0.5263483369568247, "learning_rate": 4.177944924325121e-06, "loss": 1.0285, "step": 8638 }, { "epoch": 0.6611816929435175, "grad_norm": 0.5895191497203359, "learning_rate": 4.177062311519519e-06, "loss": 1.4658, "step": 8639 }, { "epoch": 0.6612582274605847, "grad_norm": 0.7092985950340233, "learning_rate": 4.176179810720657e-06, "loss": 1.5375, "step": 8640 }, { "epoch": 0.661334761977652, "grad_norm": 0.5324642858270339, "learning_rate": 4.175297421982758e-06, "loss": 1.1242, "step": 8641 }, { "epoch": 0.6614112964947191, "grad_norm": 0.5768196485847003, "learning_rate": 4.174415145360042e-06, "loss": 1.3042, "step": 8642 }, { "epoch": 0.6614878310117863, "grad_norm": 0.5566324515712054, "learning_rate": 4.173532980906718e-06, "loss": 1.3752, "step": 8643 }, { "epoch": 0.6615643655288536, "grad_norm": 0.4619469671019941, "learning_rate": 4.172650928676987e-06, "loss": 1.0152, "step": 8644 }, { "epoch": 0.6616409000459207, "grad_norm": 0.5543629647423399, "learning_rate": 4.171768988725044e-06, "loss": 1.411, "step": 8645 }, { "epoch": 0.6617174345629879, "grad_norm": 0.5222646293699557, "learning_rate": 4.170887161105081e-06, "loss": 1.177, "step": 8646 }, { "epoch": 0.6617939690800551, "grad_norm": 0.5257853648561619, "learning_rate": 4.1700054458712785e-06, "loss": 1.2432, "step": 8647 }, { "epoch": 0.6618705035971223, "grad_norm": 0.5193736578089936, "learning_rate": 4.1691238430778125e-06, "loss": 1.2416, "step": 8648 }, { "epoch": 0.6619470381141895, "grad_norm": 0.5653600572352311, "learning_rate": 4.168242352778851e-06, "loss": 1.5187, "step": 8649 }, { "epoch": 0.6620235726312567, "grad_norm": 0.49109663761060196, "learning_rate": 4.167360975028556e-06, "loss": 1.1006, "step": 8650 }, { "epoch": 0.6621001071483239, "grad_norm": 0.756911240713731, "learning_rate": 4.166479709881079e-06, "loss": 1.3492, "step": 8651 }, { "epoch": 0.6621766416653911, "grad_norm": 0.5308252026791764, "learning_rate": 4.165598557390572e-06, "loss": 1.0021, "step": 8652 }, { "epoch": 0.6622531761824583, "grad_norm": 0.710900771483513, "learning_rate": 4.164717517611172e-06, "loss": 1.7256, "step": 8653 }, { "epoch": 0.6623297106995255, "grad_norm": 1.061038007061394, "learning_rate": 4.163836590597016e-06, "loss": 1.338, "step": 8654 }, { "epoch": 0.6624062452165926, "grad_norm": 1.0111051183747268, "learning_rate": 4.1629557764022274e-06, "loss": 1.6179, "step": 8655 }, { "epoch": 0.6624827797336599, "grad_norm": 0.9427249999337094, "learning_rate": 4.162075075080928e-06, "loss": 1.2981, "step": 8656 }, { "epoch": 0.6625593142507271, "grad_norm": 0.521864938665019, "learning_rate": 4.161194486687229e-06, "loss": 1.2734, "step": 8657 }, { "epoch": 0.6626358487677942, "grad_norm": 0.5277175957125563, "learning_rate": 4.160314011275239e-06, "loss": 1.231, "step": 8658 }, { "epoch": 0.6627123832848615, "grad_norm": 3.3254271279184713, "learning_rate": 4.159433648899056e-06, "loss": 1.5636, "step": 8659 }, { "epoch": 0.6627889178019287, "grad_norm": 0.7692953488715204, "learning_rate": 4.158553399612771e-06, "loss": 1.4698, "step": 8660 }, { "epoch": 0.6628654523189959, "grad_norm": 0.5146876417176358, "learning_rate": 4.157673263470471e-06, "loss": 1.2537, "step": 8661 }, { "epoch": 0.6629419868360631, "grad_norm": 0.544806695006846, "learning_rate": 4.156793240526233e-06, "loss": 1.1365, "step": 8662 }, { "epoch": 0.6630185213531302, "grad_norm": 0.5767866360179484, "learning_rate": 4.155913330834127e-06, "loss": 1.3381, "step": 8663 }, { "epoch": 0.6630950558701975, "grad_norm": 0.6548115022802279, "learning_rate": 4.15503353444822e-06, "loss": 1.6698, "step": 8664 }, { "epoch": 0.6631715903872647, "grad_norm": 0.5921571495348469, "learning_rate": 4.154153851422568e-06, "loss": 1.4248, "step": 8665 }, { "epoch": 0.6632481249043318, "grad_norm": 0.5405150776302247, "learning_rate": 4.153274281811221e-06, "loss": 1.273, "step": 8666 }, { "epoch": 0.6633246594213991, "grad_norm": 0.4787813338823711, "learning_rate": 4.152394825668224e-06, "loss": 1.0521, "step": 8667 }, { "epoch": 0.6634011939384662, "grad_norm": 0.48393936259767056, "learning_rate": 4.15151548304761e-06, "loss": 0.9528, "step": 8668 }, { "epoch": 0.6634777284555334, "grad_norm": 2.2073526251857807, "learning_rate": 4.150636254003412e-06, "loss": 1.2844, "step": 8669 }, { "epoch": 0.6635542629726007, "grad_norm": 0.8649961106150361, "learning_rate": 4.149757138589652e-06, "loss": 1.8085, "step": 8670 }, { "epoch": 0.6636307974896678, "grad_norm": 1.7277710932334145, "learning_rate": 4.148878136860344e-06, "loss": 0.9903, "step": 8671 }, { "epoch": 0.663707332006735, "grad_norm": 0.6124309056664045, "learning_rate": 4.147999248869497e-06, "loss": 1.5431, "step": 8672 }, { "epoch": 0.6637838665238023, "grad_norm": 0.5372326022529471, "learning_rate": 4.147120474671111e-06, "loss": 1.2402, "step": 8673 }, { "epoch": 0.6638604010408694, "grad_norm": 0.6103146595791461, "learning_rate": 4.1462418143191845e-06, "loss": 1.4181, "step": 8674 }, { "epoch": 0.6639369355579366, "grad_norm": 0.6761793325888709, "learning_rate": 4.145363267867701e-06, "loss": 1.2967, "step": 8675 }, { "epoch": 0.6640134700750038, "grad_norm": 0.5287153350710191, "learning_rate": 4.144484835370644e-06, "loss": 1.3256, "step": 8676 }, { "epoch": 0.664090004592071, "grad_norm": 0.5763395799811972, "learning_rate": 4.143606516881985e-06, "loss": 1.288, "step": 8677 }, { "epoch": 0.6641665391091383, "grad_norm": 0.7142831722946866, "learning_rate": 4.1427283124556905e-06, "loss": 1.4173, "step": 8678 }, { "epoch": 0.6642430736262054, "grad_norm": 0.6093744300099972, "learning_rate": 4.14185022214572e-06, "loss": 1.5411, "step": 8679 }, { "epoch": 0.6643196081432726, "grad_norm": 0.5966820979281158, "learning_rate": 4.1409722460060285e-06, "loss": 1.43, "step": 8680 }, { "epoch": 0.6643961426603399, "grad_norm": 0.6135254691193207, "learning_rate": 4.140094384090559e-06, "loss": 1.64, "step": 8681 }, { "epoch": 0.664472677177407, "grad_norm": 0.5057793133853501, "learning_rate": 4.139216636453251e-06, "loss": 1.1422, "step": 8682 }, { "epoch": 0.6645492116944742, "grad_norm": 0.9493865038356963, "learning_rate": 4.138339003148035e-06, "loss": 1.2533, "step": 8683 }, { "epoch": 0.6646257462115414, "grad_norm": 0.5608551281766802, "learning_rate": 4.137461484228836e-06, "loss": 1.261, "step": 8684 }, { "epoch": 0.6647022807286086, "grad_norm": 0.6106016920231938, "learning_rate": 4.13658407974957e-06, "loss": 1.4654, "step": 8685 }, { "epoch": 0.6647788152456758, "grad_norm": 0.5362916831051665, "learning_rate": 4.13570678976415e-06, "loss": 1.3317, "step": 8686 }, { "epoch": 0.664855349762743, "grad_norm": 0.593298328881746, "learning_rate": 4.1348296143264785e-06, "loss": 1.3096, "step": 8687 }, { "epoch": 0.6649318842798102, "grad_norm": 0.6088891924353935, "learning_rate": 4.1339525534904514e-06, "loss": 1.3966, "step": 8688 }, { "epoch": 0.6650084187968774, "grad_norm": 0.666157675143666, "learning_rate": 4.133075607309957e-06, "loss": 1.7956, "step": 8689 }, { "epoch": 0.6650849533139446, "grad_norm": 0.5094013767083025, "learning_rate": 4.132198775838877e-06, "loss": 1.0739, "step": 8690 }, { "epoch": 0.6651614878310118, "grad_norm": 0.5375055964523839, "learning_rate": 4.13132205913109e-06, "loss": 1.1644, "step": 8691 }, { "epoch": 0.6652380223480789, "grad_norm": 0.5236998334276711, "learning_rate": 4.130445457240462e-06, "loss": 1.0425, "step": 8692 }, { "epoch": 0.6653145568651462, "grad_norm": 0.6692655958497896, "learning_rate": 4.129568970220853e-06, "loss": 1.5309, "step": 8693 }, { "epoch": 0.6653910913822134, "grad_norm": 0.7297553210710865, "learning_rate": 4.1286925981261185e-06, "loss": 1.481, "step": 8694 }, { "epoch": 0.6654676258992805, "grad_norm": 0.5522328028139128, "learning_rate": 4.1278163410101055e-06, "loss": 1.3275, "step": 8695 }, { "epoch": 0.6655441604163478, "grad_norm": 0.5447155855603819, "learning_rate": 4.126940198926652e-06, "loss": 1.1185, "step": 8696 }, { "epoch": 0.665620694933415, "grad_norm": 0.5139564097651831, "learning_rate": 4.126064171929593e-06, "loss": 1.0592, "step": 8697 }, { "epoch": 0.6656972294504822, "grad_norm": 0.5960105325678888, "learning_rate": 4.1251882600727535e-06, "loss": 1.4392, "step": 8698 }, { "epoch": 0.6657737639675494, "grad_norm": 0.597604052195862, "learning_rate": 4.124312463409952e-06, "loss": 1.2712, "step": 8699 }, { "epoch": 0.6658502984846165, "grad_norm": 0.5641803444201414, "learning_rate": 4.123436781994999e-06, "loss": 1.2267, "step": 8700 }, { "epoch": 0.6659268330016838, "grad_norm": 0.6249212264131202, "learning_rate": 4.1225612158817e-06, "loss": 1.4898, "step": 8701 }, { "epoch": 0.666003367518751, "grad_norm": 0.5927912261242215, "learning_rate": 4.121685765123853e-06, "loss": 1.3516, "step": 8702 }, { "epoch": 0.6660799020358181, "grad_norm": 0.6398767780813831, "learning_rate": 4.120810429775247e-06, "loss": 1.6938, "step": 8703 }, { "epoch": 0.6661564365528854, "grad_norm": 0.5886773132364393, "learning_rate": 4.119935209889667e-06, "loss": 1.3468, "step": 8704 }, { "epoch": 0.6662329710699525, "grad_norm": 0.5121526259025457, "learning_rate": 4.119060105520887e-06, "loss": 1.1623, "step": 8705 }, { "epoch": 0.6663095055870197, "grad_norm": 0.5675765850528537, "learning_rate": 4.118185116722677e-06, "loss": 1.4433, "step": 8706 }, { "epoch": 0.666386040104087, "grad_norm": 0.6544863722482358, "learning_rate": 4.117310243548799e-06, "loss": 1.4913, "step": 8707 }, { "epoch": 0.6664625746211541, "grad_norm": 0.5668520087854708, "learning_rate": 4.116435486053008e-06, "loss": 1.2721, "step": 8708 }, { "epoch": 0.6665391091382213, "grad_norm": 0.5477157258831234, "learning_rate": 4.115560844289052e-06, "loss": 1.2664, "step": 8709 }, { "epoch": 0.6666156436552886, "grad_norm": 0.5098793552972759, "learning_rate": 4.114686318310671e-06, "loss": 1.0145, "step": 8710 }, { "epoch": 0.6666921781723557, "grad_norm": 0.5574400959869567, "learning_rate": 4.113811908171599e-06, "loss": 1.2601, "step": 8711 }, { "epoch": 0.666768712689423, "grad_norm": 0.4971722804020738, "learning_rate": 4.11293761392556e-06, "loss": 1.0766, "step": 8712 }, { "epoch": 0.6668452472064901, "grad_norm": 0.5927816104136746, "learning_rate": 4.112063435626277e-06, "loss": 1.2895, "step": 8713 }, { "epoch": 0.6669217817235573, "grad_norm": 0.6668511184179999, "learning_rate": 4.1111893733274596e-06, "loss": 1.5476, "step": 8714 }, { "epoch": 0.6669983162406246, "grad_norm": 0.7364865007804366, "learning_rate": 4.110315427082815e-06, "loss": 1.4154, "step": 8715 }, { "epoch": 0.6670748507576917, "grad_norm": 0.742838540876565, "learning_rate": 4.109441596946039e-06, "loss": 1.5982, "step": 8716 }, { "epoch": 0.6671513852747589, "grad_norm": 0.6678748758244354, "learning_rate": 4.108567882970822e-06, "loss": 1.209, "step": 8717 }, { "epoch": 0.6672279197918262, "grad_norm": 0.618241839494098, "learning_rate": 4.10769428521085e-06, "loss": 1.4993, "step": 8718 }, { "epoch": 0.6673044543088933, "grad_norm": 0.6519342163633229, "learning_rate": 4.106820803719798e-06, "loss": 1.6335, "step": 8719 }, { "epoch": 0.6673809888259605, "grad_norm": 0.5880990419270297, "learning_rate": 4.105947438551334e-06, "loss": 1.4429, "step": 8720 }, { "epoch": 0.6674575233430277, "grad_norm": 0.5385263410067177, "learning_rate": 4.105074189759123e-06, "loss": 1.1992, "step": 8721 }, { "epoch": 0.6675340578600949, "grad_norm": 0.9080559652675168, "learning_rate": 4.1042010573968175e-06, "loss": 1.7017, "step": 8722 }, { "epoch": 0.6676105923771621, "grad_norm": 0.8284603171355686, "learning_rate": 4.103328041518065e-06, "loss": 1.9042, "step": 8723 }, { "epoch": 0.6676871268942293, "grad_norm": 0.6264798403035443, "learning_rate": 4.102455142176509e-06, "loss": 1.5365, "step": 8724 }, { "epoch": 0.6677636614112965, "grad_norm": 0.5238015478471698, "learning_rate": 4.101582359425782e-06, "loss": 1.2105, "step": 8725 }, { "epoch": 0.6678401959283637, "grad_norm": 0.6789210035950494, "learning_rate": 4.10070969331951e-06, "loss": 1.4514, "step": 8726 }, { "epoch": 0.6679167304454309, "grad_norm": 0.6641954811020394, "learning_rate": 4.0998371439113124e-06, "loss": 1.6102, "step": 8727 }, { "epoch": 0.6679932649624981, "grad_norm": 0.553515645022328, "learning_rate": 4.0989647112548015e-06, "loss": 1.1694, "step": 8728 }, { "epoch": 0.6680697994795652, "grad_norm": 0.5440548753638038, "learning_rate": 4.098092395403579e-06, "loss": 1.1698, "step": 8729 }, { "epoch": 0.6681463339966325, "grad_norm": 1.3937593893234583, "learning_rate": 4.0972201964112475e-06, "loss": 1.2444, "step": 8730 }, { "epoch": 0.6682228685136997, "grad_norm": 0.5962486421243881, "learning_rate": 4.096348114331396e-06, "loss": 1.5211, "step": 8731 }, { "epoch": 0.6682994030307668, "grad_norm": 0.5246166655033517, "learning_rate": 4.095476149217607e-06, "loss": 1.359, "step": 8732 }, { "epoch": 0.6683759375478341, "grad_norm": 0.6238483435813458, "learning_rate": 4.094604301123456e-06, "loss": 1.3602, "step": 8733 }, { "epoch": 0.6684524720649012, "grad_norm": 0.6000476798159473, "learning_rate": 4.093732570102511e-06, "loss": 1.4551, "step": 8734 }, { "epoch": 0.6685290065819685, "grad_norm": 1.8576906752777123, "learning_rate": 4.0928609562083376e-06, "loss": 1.4058, "step": 8735 }, { "epoch": 0.6686055410990357, "grad_norm": 0.5870675905172138, "learning_rate": 4.0919894594944885e-06, "loss": 1.3211, "step": 8736 }, { "epoch": 0.6686820756161028, "grad_norm": 0.5953942567846021, "learning_rate": 4.091118080014511e-06, "loss": 1.3753, "step": 8737 }, { "epoch": 0.6687586101331701, "grad_norm": 0.5729835783177807, "learning_rate": 4.090246817821945e-06, "loss": 1.1342, "step": 8738 }, { "epoch": 0.6688351446502373, "grad_norm": 1.4012275203823892, "learning_rate": 4.089375672970324e-06, "loss": 1.4931, "step": 8739 }, { "epoch": 0.6689116791673044, "grad_norm": 0.5207714283038711, "learning_rate": 4.088504645513173e-06, "loss": 1.2421, "step": 8740 }, { "epoch": 0.6689882136843717, "grad_norm": 0.5808996969577551, "learning_rate": 4.0876337355040106e-06, "loss": 1.5178, "step": 8741 }, { "epoch": 0.6690647482014388, "grad_norm": 0.5708556097658158, "learning_rate": 4.086762942996349e-06, "loss": 1.2857, "step": 8742 }, { "epoch": 0.669141282718506, "grad_norm": 0.533982782228568, "learning_rate": 4.08589226804369e-06, "loss": 1.1712, "step": 8743 }, { "epoch": 0.6692178172355733, "grad_norm": 0.6001434095206426, "learning_rate": 4.085021710699534e-06, "loss": 1.501, "step": 8744 }, { "epoch": 0.6692943517526404, "grad_norm": 0.5960714024494311, "learning_rate": 4.0841512710173665e-06, "loss": 1.3992, "step": 8745 }, { "epoch": 0.6693708862697076, "grad_norm": 0.6394673313813013, "learning_rate": 4.083280949050674e-06, "loss": 1.5112, "step": 8746 }, { "epoch": 0.6694474207867749, "grad_norm": 0.621582286450149, "learning_rate": 4.082410744852929e-06, "loss": 1.2517, "step": 8747 }, { "epoch": 0.669523955303842, "grad_norm": 0.573749893810405, "learning_rate": 4.081540658477602e-06, "loss": 1.323, "step": 8748 }, { "epoch": 0.6696004898209093, "grad_norm": 0.7023731961038774, "learning_rate": 4.0806706899781516e-06, "loss": 1.4955, "step": 8749 }, { "epoch": 0.6696770243379764, "grad_norm": 0.5417969401964695, "learning_rate": 4.07980083940803e-06, "loss": 1.3841, "step": 8750 }, { "epoch": 0.6697535588550436, "grad_norm": 1.302340272373209, "learning_rate": 4.078931106820685e-06, "loss": 1.371, "step": 8751 }, { "epoch": 0.6698300933721109, "grad_norm": 0.7175091421745691, "learning_rate": 4.078061492269557e-06, "loss": 1.3939, "step": 8752 }, { "epoch": 0.669906627889178, "grad_norm": 0.5170893662640659, "learning_rate": 4.0771919958080775e-06, "loss": 1.2556, "step": 8753 }, { "epoch": 0.6699831624062452, "grad_norm": 0.6486321221401421, "learning_rate": 4.0763226174896685e-06, "loss": 1.714, "step": 8754 }, { "epoch": 0.6700596969233125, "grad_norm": 0.5227010043823485, "learning_rate": 4.075453357367748e-06, "loss": 1.2019, "step": 8755 }, { "epoch": 0.6701362314403796, "grad_norm": 0.5380042828059233, "learning_rate": 4.074584215495727e-06, "loss": 1.2412, "step": 8756 }, { "epoch": 0.6702127659574468, "grad_norm": 0.6690948321760923, "learning_rate": 4.073715191927008e-06, "loss": 1.3332, "step": 8757 }, { "epoch": 0.670289300474514, "grad_norm": 0.5998455582152777, "learning_rate": 4.072846286714986e-06, "loss": 1.5348, "step": 8758 }, { "epoch": 0.6703658349915812, "grad_norm": 0.48554512055625404, "learning_rate": 4.071977499913049e-06, "loss": 1.1243, "step": 8759 }, { "epoch": 0.6704423695086484, "grad_norm": 0.6042202830253243, "learning_rate": 4.071108831574579e-06, "loss": 1.5537, "step": 8760 }, { "epoch": 0.6705189040257156, "grad_norm": 0.5526478785797805, "learning_rate": 4.070240281752948e-06, "loss": 1.2557, "step": 8761 }, { "epoch": 0.6705954385427828, "grad_norm": 0.7356669147216979, "learning_rate": 4.069371850501522e-06, "loss": 1.4945, "step": 8762 }, { "epoch": 0.67067197305985, "grad_norm": 0.5778606079481634, "learning_rate": 4.068503537873662e-06, "loss": 1.3181, "step": 8763 }, { "epoch": 0.6707485075769172, "grad_norm": 0.6205640822910594, "learning_rate": 4.067635343922721e-06, "loss": 1.5815, "step": 8764 }, { "epoch": 0.6708250420939844, "grad_norm": 0.5958704768945238, "learning_rate": 4.066767268702039e-06, "loss": 1.3147, "step": 8765 }, { "epoch": 0.6709015766110515, "grad_norm": 0.6113386073181117, "learning_rate": 4.065899312264956e-06, "loss": 1.5464, "step": 8766 }, { "epoch": 0.6709781111281188, "grad_norm": 0.5433692981489522, "learning_rate": 4.065031474664798e-06, "loss": 1.074, "step": 8767 }, { "epoch": 0.671054645645186, "grad_norm": 0.6193763277558897, "learning_rate": 4.064163755954893e-06, "loss": 1.4941, "step": 8768 }, { "epoch": 0.6711311801622531, "grad_norm": 0.5955925018402276, "learning_rate": 4.063296156188554e-06, "loss": 1.338, "step": 8769 }, { "epoch": 0.6712077146793204, "grad_norm": 0.6646674622676598, "learning_rate": 4.062428675419089e-06, "loss": 1.6131, "step": 8770 }, { "epoch": 0.6712842491963875, "grad_norm": 0.6399198581696123, "learning_rate": 4.061561313699797e-06, "loss": 1.6818, "step": 8771 }, { "epoch": 0.6713607837134548, "grad_norm": 0.6530441691814057, "learning_rate": 4.060694071083974e-06, "loss": 1.3883, "step": 8772 }, { "epoch": 0.671437318230522, "grad_norm": 0.567194544295325, "learning_rate": 4.059826947624904e-06, "loss": 1.2729, "step": 8773 }, { "epoch": 0.6715138527475891, "grad_norm": 0.6014003221883233, "learning_rate": 4.058959943375867e-06, "loss": 1.1913, "step": 8774 }, { "epoch": 0.6715903872646564, "grad_norm": 0.6038885339883638, "learning_rate": 4.058093058390134e-06, "loss": 1.4132, "step": 8775 }, { "epoch": 0.6716669217817236, "grad_norm": 0.5028253703747512, "learning_rate": 4.057226292720968e-06, "loss": 1.1501, "step": 8776 }, { "epoch": 0.6717434562987907, "grad_norm": 0.5600610546799346, "learning_rate": 4.056359646421627e-06, "loss": 1.2832, "step": 8777 }, { "epoch": 0.671819990815858, "grad_norm": 0.7330746507261285, "learning_rate": 4.0554931195453584e-06, "loss": 1.4664, "step": 8778 }, { "epoch": 0.6718965253329251, "grad_norm": 0.5121783475423253, "learning_rate": 4.0546267121454055e-06, "loss": 0.921, "step": 8779 }, { "epoch": 0.6719730598499923, "grad_norm": 0.6233992224535515, "learning_rate": 4.053760424275005e-06, "loss": 1.3223, "step": 8780 }, { "epoch": 0.6720495943670596, "grad_norm": 0.7383916727843681, "learning_rate": 4.052894255987381e-06, "loss": 1.4998, "step": 8781 }, { "epoch": 0.6721261288841267, "grad_norm": 1.0939262510422054, "learning_rate": 4.052028207335754e-06, "loss": 1.6907, "step": 8782 }, { "epoch": 0.672202663401194, "grad_norm": 0.882032653045167, "learning_rate": 4.051162278373337e-06, "loss": 1.6234, "step": 8783 }, { "epoch": 0.6722791979182612, "grad_norm": 0.586038652803305, "learning_rate": 4.0502964691533365e-06, "loss": 1.2137, "step": 8784 }, { "epoch": 0.6723557324353283, "grad_norm": 0.5623648606652899, "learning_rate": 4.04943077972895e-06, "loss": 1.2844, "step": 8785 }, { "epoch": 0.6724322669523956, "grad_norm": 0.5243450067188663, "learning_rate": 4.048565210153367e-06, "loss": 1.1636, "step": 8786 }, { "epoch": 0.6725088014694627, "grad_norm": 1.1188288099043835, "learning_rate": 4.047699760479771e-06, "loss": 1.7479, "step": 8787 }, { "epoch": 0.6725853359865299, "grad_norm": 0.653140739896565, "learning_rate": 4.046834430761338e-06, "loss": 1.6683, "step": 8788 }, { "epoch": 0.6726618705035972, "grad_norm": 0.5503463918468777, "learning_rate": 4.045969221051237e-06, "loss": 1.1307, "step": 8789 }, { "epoch": 0.6727384050206643, "grad_norm": 0.5444894835918427, "learning_rate": 4.045104131402628e-06, "loss": 1.2976, "step": 8790 }, { "epoch": 0.6728149395377315, "grad_norm": 0.5558226395060462, "learning_rate": 4.0442391618686656e-06, "loss": 1.156, "step": 8791 }, { "epoch": 0.6728914740547988, "grad_norm": 0.5829219180098599, "learning_rate": 4.043374312502495e-06, "loss": 1.3826, "step": 8792 }, { "epoch": 0.6729680085718659, "grad_norm": 0.6361896249540715, "learning_rate": 4.042509583357257e-06, "loss": 1.1756, "step": 8793 }, { "epoch": 0.6730445430889331, "grad_norm": 0.6495430190180932, "learning_rate": 4.0416449744860795e-06, "loss": 1.6076, "step": 8794 }, { "epoch": 0.6731210776060003, "grad_norm": 0.6056177959751342, "learning_rate": 4.040780485942091e-06, "loss": 1.2731, "step": 8795 }, { "epoch": 0.6731976121230675, "grad_norm": 0.5300748810701528, "learning_rate": 4.0399161177784055e-06, "loss": 1.3481, "step": 8796 }, { "epoch": 0.6732741466401347, "grad_norm": 0.5759405875148217, "learning_rate": 4.0390518700481345e-06, "loss": 1.4041, "step": 8797 }, { "epoch": 0.6733506811572019, "grad_norm": 0.9418965263281455, "learning_rate": 4.0381877428043784e-06, "loss": 1.3427, "step": 8798 }, { "epoch": 0.6734272156742691, "grad_norm": 0.5372190316591388, "learning_rate": 4.037323736100232e-06, "loss": 1.1649, "step": 8799 }, { "epoch": 0.6735037501913363, "grad_norm": 0.616371309399484, "learning_rate": 4.036459849988781e-06, "loss": 1.3785, "step": 8800 }, { "epoch": 0.6735802847084035, "grad_norm": 0.6098355893282956, "learning_rate": 4.035596084523108e-06, "loss": 1.483, "step": 8801 }, { "epoch": 0.6736568192254707, "grad_norm": 0.6239903239239719, "learning_rate": 4.034732439756284e-06, "loss": 1.4549, "step": 8802 }, { "epoch": 0.6737333537425378, "grad_norm": 0.6191824317209847, "learning_rate": 4.033868915741374e-06, "loss": 1.4495, "step": 8803 }, { "epoch": 0.6738098882596051, "grad_norm": 0.666343736084391, "learning_rate": 4.033005512531435e-06, "loss": 1.7549, "step": 8804 }, { "epoch": 0.6738864227766723, "grad_norm": 0.5946620511491789, "learning_rate": 4.032142230179517e-06, "loss": 1.459, "step": 8805 }, { "epoch": 0.6739629572937395, "grad_norm": 0.5156266752735443, "learning_rate": 4.031279068738664e-06, "loss": 1.041, "step": 8806 }, { "epoch": 0.6740394918108067, "grad_norm": 0.6084691835858508, "learning_rate": 4.030416028261911e-06, "loss": 1.2723, "step": 8807 }, { "epoch": 0.6741160263278738, "grad_norm": 0.560258261489574, "learning_rate": 4.029553108802284e-06, "loss": 1.4176, "step": 8808 }, { "epoch": 0.6741925608449411, "grad_norm": 0.6464187935551146, "learning_rate": 4.0286903104128064e-06, "loss": 1.3924, "step": 8809 }, { "epoch": 0.6742690953620083, "grad_norm": 0.6015603226410559, "learning_rate": 4.027827633146488e-06, "loss": 1.3844, "step": 8810 }, { "epoch": 0.6743456298790754, "grad_norm": 0.6928877515018659, "learning_rate": 4.026965077056335e-06, "loss": 1.242, "step": 8811 }, { "epoch": 0.6744221643961427, "grad_norm": 0.6146192339688039, "learning_rate": 4.0261026421953465e-06, "loss": 1.4792, "step": 8812 }, { "epoch": 0.6744986989132099, "grad_norm": 0.632771403120142, "learning_rate": 4.025240328616513e-06, "loss": 1.3079, "step": 8813 }, { "epoch": 0.674575233430277, "grad_norm": 1.9863506915831048, "learning_rate": 4.024378136372816e-06, "loss": 1.6204, "step": 8814 }, { "epoch": 0.6746517679473443, "grad_norm": 0.8126012044083438, "learning_rate": 4.023516065517234e-06, "loss": 1.543, "step": 8815 }, { "epoch": 0.6747283024644114, "grad_norm": 0.5896115467411039, "learning_rate": 4.02265411610273e-06, "loss": 1.3297, "step": 8816 }, { "epoch": 0.6748048369814786, "grad_norm": 0.5602396201229641, "learning_rate": 4.021792288182271e-06, "loss": 1.3663, "step": 8817 }, { "epoch": 0.6748813714985459, "grad_norm": 0.6316804253578483, "learning_rate": 4.020930581808806e-06, "loss": 1.5483, "step": 8818 }, { "epoch": 0.674957906015613, "grad_norm": 0.6729603630535999, "learning_rate": 4.020068997035284e-06, "loss": 1.4877, "step": 8819 }, { "epoch": 0.6750344405326802, "grad_norm": 0.6742810835463324, "learning_rate": 4.019207533914641e-06, "loss": 1.293, "step": 8820 }, { "epoch": 0.6751109750497475, "grad_norm": 0.5671231251813285, "learning_rate": 4.018346192499809e-06, "loss": 1.1774, "step": 8821 }, { "epoch": 0.6751875095668146, "grad_norm": 0.5626294560075943, "learning_rate": 4.017484972843709e-06, "loss": 1.2727, "step": 8822 }, { "epoch": 0.6752640440838819, "grad_norm": 0.7185770113649975, "learning_rate": 4.016623874999261e-06, "loss": 1.4661, "step": 8823 }, { "epoch": 0.675340578600949, "grad_norm": 0.6234146092154419, "learning_rate": 4.015762899019371e-06, "loss": 1.5586, "step": 8824 }, { "epoch": 0.6754171131180162, "grad_norm": 3.802211480104787, "learning_rate": 4.01490204495694e-06, "loss": 1.6288, "step": 8825 }, { "epoch": 0.6754936476350835, "grad_norm": 0.5239809979735186, "learning_rate": 4.014041312864861e-06, "loss": 1.1893, "step": 8826 }, { "epoch": 0.6755701821521506, "grad_norm": 0.6229176033029048, "learning_rate": 4.01318070279602e-06, "loss": 1.4638, "step": 8827 }, { "epoch": 0.6756467166692178, "grad_norm": 0.8654453265641806, "learning_rate": 4.012320214803298e-06, "loss": 1.6633, "step": 8828 }, { "epoch": 0.6757232511862851, "grad_norm": 0.6937739398728635, "learning_rate": 4.0114598489395625e-06, "loss": 1.6475, "step": 8829 }, { "epoch": 0.6757997857033522, "grad_norm": 0.6629899638459853, "learning_rate": 4.0105996052576805e-06, "loss": 1.5292, "step": 8830 }, { "epoch": 0.6758763202204194, "grad_norm": 0.5798167988202451, "learning_rate": 4.009739483810504e-06, "loss": 1.4866, "step": 8831 }, { "epoch": 0.6759528547374866, "grad_norm": 0.5516381789667608, "learning_rate": 4.008879484650884e-06, "loss": 1.197, "step": 8832 }, { "epoch": 0.6760293892545538, "grad_norm": 0.5897025931705002, "learning_rate": 4.008019607831661e-06, "loss": 1.3914, "step": 8833 }, { "epoch": 0.676105923771621, "grad_norm": 0.5858187890899342, "learning_rate": 4.007159853405669e-06, "loss": 1.502, "step": 8834 }, { "epoch": 0.6761824582886882, "grad_norm": 0.6226920017672776, "learning_rate": 4.006300221425732e-06, "loss": 1.5891, "step": 8835 }, { "epoch": 0.6762589928057554, "grad_norm": 0.5773252965345116, "learning_rate": 4.005440711944669e-06, "loss": 1.3453, "step": 8836 }, { "epoch": 0.6763355273228225, "grad_norm": 0.5194683006549824, "learning_rate": 4.004581325015293e-06, "loss": 1.2685, "step": 8837 }, { "epoch": 0.6764120618398898, "grad_norm": 0.6270379044496233, "learning_rate": 4.0037220606904024e-06, "loss": 1.7299, "step": 8838 }, { "epoch": 0.676488596356957, "grad_norm": 0.6546068069249128, "learning_rate": 4.002862919022799e-06, "loss": 1.4922, "step": 8839 }, { "epoch": 0.6765651308740241, "grad_norm": 0.6247864640942044, "learning_rate": 4.002003900065267e-06, "loss": 1.6554, "step": 8840 }, { "epoch": 0.6766416653910914, "grad_norm": 0.7260392150338285, "learning_rate": 4.001145003870589e-06, "loss": 1.6423, "step": 8841 }, { "epoch": 0.6767181999081586, "grad_norm": 0.5699287571670978, "learning_rate": 4.0002862304915365e-06, "loss": 1.3903, "step": 8842 }, { "epoch": 0.6767947344252258, "grad_norm": 0.5910008115603685, "learning_rate": 3.999427579980876e-06, "loss": 1.3382, "step": 8843 }, { "epoch": 0.676871268942293, "grad_norm": 0.591951728782211, "learning_rate": 3.998569052391366e-06, "loss": 1.4424, "step": 8844 }, { "epoch": 0.6769478034593601, "grad_norm": 0.5196551873000209, "learning_rate": 3.997710647775756e-06, "loss": 1.2368, "step": 8845 }, { "epoch": 0.6770243379764274, "grad_norm": 0.5874578828970245, "learning_rate": 3.996852366186789e-06, "loss": 1.3296, "step": 8846 }, { "epoch": 0.6771008724934946, "grad_norm": 0.854772465531892, "learning_rate": 3.995994207677203e-06, "loss": 1.6527, "step": 8847 }, { "epoch": 0.6771774070105617, "grad_norm": 0.5783054638495931, "learning_rate": 3.995136172299723e-06, "loss": 0.9958, "step": 8848 }, { "epoch": 0.677253941527629, "grad_norm": 0.6465178563393555, "learning_rate": 3.994278260107069e-06, "loss": 1.4101, "step": 8849 }, { "epoch": 0.6773304760446962, "grad_norm": 0.6011442259518011, "learning_rate": 3.993420471151954e-06, "loss": 1.4941, "step": 8850 }, { "epoch": 0.6774070105617633, "grad_norm": 0.8549594320309772, "learning_rate": 3.992562805487086e-06, "loss": 1.5935, "step": 8851 }, { "epoch": 0.6774835450788306, "grad_norm": 0.646423918377188, "learning_rate": 3.991705263165159e-06, "loss": 1.4219, "step": 8852 }, { "epoch": 0.6775600795958977, "grad_norm": 0.6405917382616597, "learning_rate": 3.990847844238865e-06, "loss": 1.5818, "step": 8853 }, { "epoch": 0.6776366141129649, "grad_norm": 0.5836593713306442, "learning_rate": 3.989990548760886e-06, "loss": 1.2676, "step": 8854 }, { "epoch": 0.6777131486300322, "grad_norm": 0.6254550468019076, "learning_rate": 3.9891333767838955e-06, "loss": 1.4182, "step": 8855 }, { "epoch": 0.6777896831470993, "grad_norm": 0.5806151664710318, "learning_rate": 3.9882763283605625e-06, "loss": 1.4566, "step": 8856 }, { "epoch": 0.6778662176641665, "grad_norm": 0.7419611473529911, "learning_rate": 3.987419403543546e-06, "loss": 1.5956, "step": 8857 }, { "epoch": 0.6779427521812338, "grad_norm": 0.6311609474167967, "learning_rate": 3.986562602385496e-06, "loss": 1.6477, "step": 8858 }, { "epoch": 0.6780192866983009, "grad_norm": 0.8049710605425681, "learning_rate": 3.98570592493906e-06, "loss": 1.7672, "step": 8859 }, { "epoch": 0.6780958212153682, "grad_norm": 0.5656607574093069, "learning_rate": 3.984849371256873e-06, "loss": 1.2917, "step": 8860 }, { "epoch": 0.6781723557324353, "grad_norm": 0.6508181812950571, "learning_rate": 3.983992941391565e-06, "loss": 1.5144, "step": 8861 }, { "epoch": 0.6782488902495025, "grad_norm": 0.554077743334971, "learning_rate": 3.9831366353957576e-06, "loss": 1.1827, "step": 8862 }, { "epoch": 0.6783254247665698, "grad_norm": 0.6159087430326424, "learning_rate": 3.982280453322065e-06, "loss": 1.4098, "step": 8863 }, { "epoch": 0.6784019592836369, "grad_norm": 0.5497049370664514, "learning_rate": 3.981424395223093e-06, "loss": 1.4556, "step": 8864 }, { "epoch": 0.6784784938007041, "grad_norm": 0.5497810527582175, "learning_rate": 3.9805684611514405e-06, "loss": 1.3045, "step": 8865 }, { "epoch": 0.6785550283177714, "grad_norm": 0.5450444339634685, "learning_rate": 3.979712651159697e-06, "loss": 1.3262, "step": 8866 }, { "epoch": 0.6786315628348385, "grad_norm": 0.6754515871930654, "learning_rate": 3.978856965300449e-06, "loss": 1.5238, "step": 8867 }, { "epoch": 0.6787080973519057, "grad_norm": 0.6665845465173078, "learning_rate": 3.9780014036262716e-06, "loss": 1.51, "step": 8868 }, { "epoch": 0.6787846318689729, "grad_norm": 0.7659463727098602, "learning_rate": 3.9771459661897325e-06, "loss": 1.4577, "step": 8869 }, { "epoch": 0.6788611663860401, "grad_norm": 0.5697962644227672, "learning_rate": 3.976290653043393e-06, "loss": 1.304, "step": 8870 }, { "epoch": 0.6789377009031073, "grad_norm": 0.5678592152947585, "learning_rate": 3.975435464239803e-06, "loss": 1.2923, "step": 8871 }, { "epoch": 0.6790142354201745, "grad_norm": 0.563119769290449, "learning_rate": 3.9745803998315125e-06, "loss": 1.2032, "step": 8872 }, { "epoch": 0.6790907699372417, "grad_norm": 0.5223109776409938, "learning_rate": 3.973725459871056e-06, "loss": 1.198, "step": 8873 }, { "epoch": 0.6791673044543088, "grad_norm": 0.679797409789621, "learning_rate": 3.972870644410966e-06, "loss": 1.6311, "step": 8874 }, { "epoch": 0.6792438389713761, "grad_norm": 0.6569566325014335, "learning_rate": 3.972015953503763e-06, "loss": 1.6893, "step": 8875 }, { "epoch": 0.6793203734884433, "grad_norm": 0.9252973839654056, "learning_rate": 3.971161387201963e-06, "loss": 1.5672, "step": 8876 }, { "epoch": 0.6793969080055104, "grad_norm": 0.6123441226485896, "learning_rate": 3.970306945558071e-06, "loss": 1.4925, "step": 8877 }, { "epoch": 0.6794734425225777, "grad_norm": 0.5104883278876564, "learning_rate": 3.96945262862459e-06, "loss": 1.0907, "step": 8878 }, { "epoch": 0.6795499770396449, "grad_norm": 0.5162353225353473, "learning_rate": 3.968598436454007e-06, "loss": 1.1615, "step": 8879 }, { "epoch": 0.679626511556712, "grad_norm": 0.888591582172232, "learning_rate": 3.967744369098811e-06, "loss": 1.5273, "step": 8880 }, { "epoch": 0.6797030460737793, "grad_norm": 0.5629623579870551, "learning_rate": 3.9668904266114755e-06, "loss": 1.3349, "step": 8881 }, { "epoch": 0.6797795805908464, "grad_norm": 0.5483196797561414, "learning_rate": 3.966036609044468e-06, "loss": 1.1159, "step": 8882 }, { "epoch": 0.6798561151079137, "grad_norm": 0.6071128552309054, "learning_rate": 3.965182916450254e-06, "loss": 1.4225, "step": 8883 }, { "epoch": 0.6799326496249809, "grad_norm": 0.5334909364957158, "learning_rate": 3.964329348881284e-06, "loss": 1.3575, "step": 8884 }, { "epoch": 0.680009184142048, "grad_norm": 0.6795350760779614, "learning_rate": 3.963475906390005e-06, "loss": 1.5271, "step": 8885 }, { "epoch": 0.6800857186591153, "grad_norm": 0.5604697055106392, "learning_rate": 3.962622589028854e-06, "loss": 1.2899, "step": 8886 }, { "epoch": 0.6801622531761825, "grad_norm": 0.741234636463558, "learning_rate": 3.961769396850261e-06, "loss": 1.4056, "step": 8887 }, { "epoch": 0.6802387876932496, "grad_norm": 0.5416818615019354, "learning_rate": 3.960916329906649e-06, "loss": 1.1774, "step": 8888 }, { "epoch": 0.6803153222103169, "grad_norm": 0.626951986757848, "learning_rate": 3.960063388250434e-06, "loss": 1.313, "step": 8889 }, { "epoch": 0.680391856727384, "grad_norm": 0.8695622392725079, "learning_rate": 3.959210571934024e-06, "loss": 1.8286, "step": 8890 }, { "epoch": 0.6804683912444512, "grad_norm": 0.5822926506709208, "learning_rate": 3.9583578810098155e-06, "loss": 1.4509, "step": 8891 }, { "epoch": 0.6805449257615185, "grad_norm": 0.5387697302825035, "learning_rate": 3.957505315530203e-06, "loss": 1.0601, "step": 8892 }, { "epoch": 0.6806214602785856, "grad_norm": 0.66458729599447, "learning_rate": 3.956652875547569e-06, "loss": 1.4123, "step": 8893 }, { "epoch": 0.6806979947956528, "grad_norm": 0.7084378380195495, "learning_rate": 3.955800561114291e-06, "loss": 1.4987, "step": 8894 }, { "epoch": 0.6807745293127201, "grad_norm": 0.49567777396742263, "learning_rate": 3.9549483722827385e-06, "loss": 1.0557, "step": 8895 }, { "epoch": 0.6808510638297872, "grad_norm": 0.5186037108276176, "learning_rate": 3.954096309105271e-06, "loss": 1.117, "step": 8896 }, { "epoch": 0.6809275983468545, "grad_norm": 0.6476531951831404, "learning_rate": 3.953244371634242e-06, "loss": 1.3782, "step": 8897 }, { "epoch": 0.6810041328639216, "grad_norm": 0.6198493496896195, "learning_rate": 3.952392559921998e-06, "loss": 1.4659, "step": 8898 }, { "epoch": 0.6810806673809888, "grad_norm": 0.5567375562786561, "learning_rate": 3.951540874020876e-06, "loss": 1.2841, "step": 8899 }, { "epoch": 0.6811572018980561, "grad_norm": 0.5295996193891125, "learning_rate": 3.950689313983207e-06, "loss": 1.0822, "step": 8900 }, { "epoch": 0.6812337364151232, "grad_norm": 0.6085223931490393, "learning_rate": 3.949837879861313e-06, "loss": 1.4922, "step": 8901 }, { "epoch": 0.6813102709321904, "grad_norm": 0.4898970258699682, "learning_rate": 3.9489865717075075e-06, "loss": 1.106, "step": 8902 }, { "epoch": 0.6813868054492576, "grad_norm": 0.49578991560895397, "learning_rate": 3.9481353895740985e-06, "loss": 1.0605, "step": 8903 }, { "epoch": 0.6814633399663248, "grad_norm": 0.9928383526469122, "learning_rate": 3.947284333513383e-06, "loss": 1.3908, "step": 8904 }, { "epoch": 0.681539874483392, "grad_norm": 0.5795014637276492, "learning_rate": 3.946433403577658e-06, "loss": 1.3665, "step": 8905 }, { "epoch": 0.6816164090004592, "grad_norm": 0.5855944140725678, "learning_rate": 3.945582599819203e-06, "loss": 1.3121, "step": 8906 }, { "epoch": 0.6816929435175264, "grad_norm": 0.5503604733248655, "learning_rate": 3.944731922290294e-06, "loss": 1.2478, "step": 8907 }, { "epoch": 0.6817694780345936, "grad_norm": 0.6018200318461927, "learning_rate": 3.9438813710432e-06, "loss": 1.4737, "step": 8908 }, { "epoch": 0.6818460125516608, "grad_norm": 0.7077796490058049, "learning_rate": 3.943030946130181e-06, "loss": 1.4616, "step": 8909 }, { "epoch": 0.681922547068728, "grad_norm": 0.5041818337135021, "learning_rate": 3.942180647603489e-06, "loss": 1.1149, "step": 8910 }, { "epoch": 0.6819990815857951, "grad_norm": 0.6070068825169077, "learning_rate": 3.94133047551537e-06, "loss": 1.5127, "step": 8911 }, { "epoch": 0.6820756161028624, "grad_norm": 0.555919899194365, "learning_rate": 3.9404804299180624e-06, "loss": 1.395, "step": 8912 }, { "epoch": 0.6821521506199296, "grad_norm": 0.5269217422378502, "learning_rate": 3.939630510863793e-06, "loss": 1.221, "step": 8913 }, { "epoch": 0.6822286851369967, "grad_norm": 0.5980137850037283, "learning_rate": 3.938780718404784e-06, "loss": 1.3947, "step": 8914 }, { "epoch": 0.682305219654064, "grad_norm": 0.5982195240245384, "learning_rate": 3.937931052593249e-06, "loss": 1.4706, "step": 8915 }, { "epoch": 0.6823817541711312, "grad_norm": 0.6629616039519872, "learning_rate": 3.937081513481395e-06, "loss": 1.5449, "step": 8916 }, { "epoch": 0.6824582886881984, "grad_norm": 0.5496641963017285, "learning_rate": 3.936232101121421e-06, "loss": 1.28, "step": 8917 }, { "epoch": 0.6825348232052656, "grad_norm": 0.5491464824362912, "learning_rate": 3.935382815565515e-06, "loss": 1.2587, "step": 8918 }, { "epoch": 0.6826113577223327, "grad_norm": 0.6010881674647184, "learning_rate": 3.934533656865862e-06, "loss": 1.3851, "step": 8919 }, { "epoch": 0.6826878922394, "grad_norm": 0.5109065121021008, "learning_rate": 3.933684625074634e-06, "loss": 0.9896, "step": 8920 }, { "epoch": 0.6827644267564672, "grad_norm": 1.3315356589275298, "learning_rate": 3.932835720244001e-06, "loss": 1.3998, "step": 8921 }, { "epoch": 0.6828409612735343, "grad_norm": 0.697730376145948, "learning_rate": 3.931986942426121e-06, "loss": 1.5866, "step": 8922 }, { "epoch": 0.6829174957906016, "grad_norm": 0.649659122459477, "learning_rate": 3.931138291673145e-06, "loss": 1.566, "step": 8923 }, { "epoch": 0.6829940303076688, "grad_norm": 0.5937591754525515, "learning_rate": 3.930289768037217e-06, "loss": 1.4776, "step": 8924 }, { "epoch": 0.6830705648247359, "grad_norm": 0.6257397180453582, "learning_rate": 3.929441371570474e-06, "loss": 1.4068, "step": 8925 }, { "epoch": 0.6831470993418032, "grad_norm": 0.5429551111871553, "learning_rate": 3.92859310232504e-06, "loss": 1.165, "step": 8926 }, { "epoch": 0.6832236338588703, "grad_norm": 1.370093182433168, "learning_rate": 3.927744960353041e-06, "loss": 1.7898, "step": 8927 }, { "epoch": 0.6833001683759375, "grad_norm": 0.567903066276596, "learning_rate": 3.9268969457065854e-06, "loss": 1.263, "step": 8928 }, { "epoch": 0.6833767028930048, "grad_norm": 0.5268972821061984, "learning_rate": 3.926049058437779e-06, "loss": 1.1448, "step": 8929 }, { "epoch": 0.6834532374100719, "grad_norm": 0.5766871922392289, "learning_rate": 3.925201298598719e-06, "loss": 1.2152, "step": 8930 }, { "epoch": 0.6835297719271392, "grad_norm": 0.5459320450129276, "learning_rate": 3.924353666241492e-06, "loss": 1.2039, "step": 8931 }, { "epoch": 0.6836063064442064, "grad_norm": 0.6052399415579064, "learning_rate": 3.9235061614181814e-06, "loss": 1.3607, "step": 8932 }, { "epoch": 0.6836828409612735, "grad_norm": 0.5487012811007616, "learning_rate": 3.92265878418086e-06, "loss": 1.232, "step": 8933 }, { "epoch": 0.6837593754783408, "grad_norm": 0.5690202917301891, "learning_rate": 3.921811534581594e-06, "loss": 1.4224, "step": 8934 }, { "epoch": 0.6838359099954079, "grad_norm": 0.6570679241636411, "learning_rate": 3.92096441267244e-06, "loss": 1.6285, "step": 8935 }, { "epoch": 0.6839124445124751, "grad_norm": 1.6423746153864003, "learning_rate": 3.920117418505446e-06, "loss": 1.238, "step": 8936 }, { "epoch": 0.6839889790295424, "grad_norm": 0.6148082099210213, "learning_rate": 3.919270552132655e-06, "loss": 1.4114, "step": 8937 }, { "epoch": 0.6840655135466095, "grad_norm": 0.6573530437932561, "learning_rate": 3.918423813606102e-06, "loss": 1.8017, "step": 8938 }, { "epoch": 0.6841420480636767, "grad_norm": 0.6852870559042894, "learning_rate": 3.9175772029778145e-06, "loss": 1.6712, "step": 8939 }, { "epoch": 0.6842185825807439, "grad_norm": 0.6021718738115769, "learning_rate": 3.916730720299808e-06, "loss": 1.3121, "step": 8940 }, { "epoch": 0.6842951170978111, "grad_norm": 0.607301174951472, "learning_rate": 3.915884365624094e-06, "loss": 1.4647, "step": 8941 }, { "epoch": 0.6843716516148783, "grad_norm": 0.7756829008706785, "learning_rate": 3.915038139002675e-06, "loss": 1.7646, "step": 8942 }, { "epoch": 0.6844481861319455, "grad_norm": 0.6477211886187735, "learning_rate": 3.914192040487544e-06, "loss": 1.4537, "step": 8943 }, { "epoch": 0.6845247206490127, "grad_norm": 2.0460436396453376, "learning_rate": 3.913346070130691e-06, "loss": 1.495, "step": 8944 }, { "epoch": 0.68460125516608, "grad_norm": 0.5401394995884055, "learning_rate": 3.9125002279840925e-06, "loss": 1.1961, "step": 8945 }, { "epoch": 0.6846777896831471, "grad_norm": 0.5308735192603632, "learning_rate": 3.911654514099722e-06, "loss": 1.0864, "step": 8946 }, { "epoch": 0.6847543242002143, "grad_norm": 0.5375262939052552, "learning_rate": 3.910808928529539e-06, "loss": 1.3713, "step": 8947 }, { "epoch": 0.6848308587172814, "grad_norm": 0.5262843192986167, "learning_rate": 3.909963471325501e-06, "loss": 1.2188, "step": 8948 }, { "epoch": 0.6849073932343487, "grad_norm": 0.5404281843567743, "learning_rate": 3.909118142539556e-06, "loss": 1.1913, "step": 8949 }, { "epoch": 0.6849839277514159, "grad_norm": 0.7282399771835563, "learning_rate": 3.908272942223643e-06, "loss": 1.4659, "step": 8950 }, { "epoch": 0.685060462268483, "grad_norm": 0.6094507620637067, "learning_rate": 3.9074278704296945e-06, "loss": 1.3586, "step": 8951 }, { "epoch": 0.6851369967855503, "grad_norm": 1.0776643486971904, "learning_rate": 3.906582927209632e-06, "loss": 1.576, "step": 8952 }, { "epoch": 0.6852135313026175, "grad_norm": 0.6645168201457333, "learning_rate": 3.905738112615372e-06, "loss": 1.7027, "step": 8953 }, { "epoch": 0.6852900658196847, "grad_norm": 0.5959390141695748, "learning_rate": 3.9048934266988235e-06, "loss": 1.3262, "step": 8954 }, { "epoch": 0.6853666003367519, "grad_norm": 0.6127255825679331, "learning_rate": 3.904048869511886e-06, "loss": 1.3954, "step": 8955 }, { "epoch": 0.685443134853819, "grad_norm": 0.6045607410521706, "learning_rate": 3.903204441106453e-06, "loss": 1.3002, "step": 8956 }, { "epoch": 0.6855196693708863, "grad_norm": 0.5554600432189497, "learning_rate": 3.902360141534406e-06, "loss": 1.3601, "step": 8957 }, { "epoch": 0.6855962038879535, "grad_norm": 0.579285187734379, "learning_rate": 3.901515970847623e-06, "loss": 1.4132, "step": 8958 }, { "epoch": 0.6856727384050206, "grad_norm": 0.5326105706153612, "learning_rate": 3.900671929097971e-06, "loss": 1.2327, "step": 8959 }, { "epoch": 0.6857492729220879, "grad_norm": 0.5462916412342851, "learning_rate": 3.8998280163373126e-06, "loss": 1.3938, "step": 8960 }, { "epoch": 0.6858258074391551, "grad_norm": 0.5672153368660607, "learning_rate": 3.8989842326174996e-06, "loss": 1.325, "step": 8961 }, { "epoch": 0.6859023419562222, "grad_norm": 0.5374192486101383, "learning_rate": 3.898140577990376e-06, "loss": 1.0961, "step": 8962 }, { "epoch": 0.6859788764732895, "grad_norm": 0.5600992815634511, "learning_rate": 3.897297052507778e-06, "loss": 1.2725, "step": 8963 }, { "epoch": 0.6860554109903566, "grad_norm": 0.6148902486594436, "learning_rate": 3.896453656221536e-06, "loss": 1.4072, "step": 8964 }, { "epoch": 0.6861319455074238, "grad_norm": 0.6662930551956695, "learning_rate": 3.89561038918347e-06, "loss": 1.6573, "step": 8965 }, { "epoch": 0.6862084800244911, "grad_norm": 0.5887035709947415, "learning_rate": 3.894767251445391e-06, "loss": 1.4619, "step": 8966 }, { "epoch": 0.6862850145415582, "grad_norm": 0.6562938348352986, "learning_rate": 3.893924243059106e-06, "loss": 1.524, "step": 8967 }, { "epoch": 0.6863615490586255, "grad_norm": 0.6928294649763653, "learning_rate": 3.893081364076412e-06, "loss": 1.6058, "step": 8968 }, { "epoch": 0.6864380835756927, "grad_norm": 0.5641616477193978, "learning_rate": 3.892238614549098e-06, "loss": 1.0093, "step": 8969 }, { "epoch": 0.6865146180927598, "grad_norm": 0.633205681631284, "learning_rate": 3.891395994528943e-06, "loss": 1.3642, "step": 8970 }, { "epoch": 0.6865911526098271, "grad_norm": 0.5267696192072233, "learning_rate": 3.8905535040677215e-06, "loss": 1.1611, "step": 8971 }, { "epoch": 0.6866676871268942, "grad_norm": 0.6080153636364064, "learning_rate": 3.8897111432172e-06, "loss": 1.5234, "step": 8972 }, { "epoch": 0.6867442216439614, "grad_norm": 0.5242845736979332, "learning_rate": 3.888868912029135e-06, "loss": 1.2157, "step": 8973 }, { "epoch": 0.6868207561610287, "grad_norm": 0.6604234703127733, "learning_rate": 3.888026810555273e-06, "loss": 1.4264, "step": 8974 }, { "epoch": 0.6868972906780958, "grad_norm": 0.7048875091074867, "learning_rate": 3.8871848388473574e-06, "loss": 1.7241, "step": 8975 }, { "epoch": 0.686973825195163, "grad_norm": 0.6164984952370662, "learning_rate": 3.886342996957123e-06, "loss": 1.3448, "step": 8976 }, { "epoch": 0.6870503597122302, "grad_norm": 0.6891395374952796, "learning_rate": 3.885501284936293e-06, "loss": 1.6228, "step": 8977 }, { "epoch": 0.6871268942292974, "grad_norm": 0.608121008643236, "learning_rate": 3.884659702836586e-06, "loss": 1.2713, "step": 8978 }, { "epoch": 0.6872034287463646, "grad_norm": 0.6296414250838068, "learning_rate": 3.883818250709709e-06, "loss": 1.5759, "step": 8979 }, { "epoch": 0.6872799632634318, "grad_norm": 1.1654986038968742, "learning_rate": 3.882976928607366e-06, "loss": 1.6351, "step": 8980 }, { "epoch": 0.687356497780499, "grad_norm": 0.647081591723051, "learning_rate": 3.882135736581247e-06, "loss": 1.5377, "step": 8981 }, { "epoch": 0.6874330322975662, "grad_norm": 0.6456838059169854, "learning_rate": 3.881294674683042e-06, "loss": 1.5054, "step": 8982 }, { "epoch": 0.6875095668146334, "grad_norm": 0.5449512598026157, "learning_rate": 3.880453742964426e-06, "loss": 1.2136, "step": 8983 }, { "epoch": 0.6875861013317006, "grad_norm": 0.5368286941180215, "learning_rate": 3.879612941477068e-06, "loss": 1.3371, "step": 8984 }, { "epoch": 0.6876626358487677, "grad_norm": 0.6842308827449052, "learning_rate": 3.878772270272628e-06, "loss": 1.2375, "step": 8985 }, { "epoch": 0.687739170365835, "grad_norm": 1.051957956971879, "learning_rate": 3.877931729402762e-06, "loss": 1.9881, "step": 8986 }, { "epoch": 0.6878157048829022, "grad_norm": 2.418069514401122, "learning_rate": 3.877091318919114e-06, "loss": 1.1887, "step": 8987 }, { "epoch": 0.6878922393999694, "grad_norm": 0.556868739787226, "learning_rate": 3.876251038873324e-06, "loss": 1.1802, "step": 8988 }, { "epoch": 0.6879687739170366, "grad_norm": 0.5507109171480009, "learning_rate": 3.8754108893170175e-06, "loss": 1.2091, "step": 8989 }, { "epoch": 0.6880453084341038, "grad_norm": 0.5779915854283552, "learning_rate": 3.874570870301817e-06, "loss": 1.385, "step": 8990 }, { "epoch": 0.688121842951171, "grad_norm": 0.5316610983421918, "learning_rate": 3.873730981879336e-06, "loss": 1.2648, "step": 8991 }, { "epoch": 0.6881983774682382, "grad_norm": 0.652823128413962, "learning_rate": 3.872891224101178e-06, "loss": 1.2987, "step": 8992 }, { "epoch": 0.6882749119853053, "grad_norm": 0.6102669202268484, "learning_rate": 3.872051597018944e-06, "loss": 1.3359, "step": 8993 }, { "epoch": 0.6883514465023726, "grad_norm": 0.6001398497439447, "learning_rate": 3.871212100684222e-06, "loss": 1.5248, "step": 8994 }, { "epoch": 0.6884279810194398, "grad_norm": 0.9720268541423743, "learning_rate": 3.870372735148591e-06, "loss": 1.1143, "step": 8995 }, { "epoch": 0.6885045155365069, "grad_norm": 0.569542437149209, "learning_rate": 3.869533500463626e-06, "loss": 1.3297, "step": 8996 }, { "epoch": 0.6885810500535742, "grad_norm": 0.5314704987315786, "learning_rate": 3.8686943966808895e-06, "loss": 1.2731, "step": 8997 }, { "epoch": 0.6886575845706414, "grad_norm": 0.5155661270281928, "learning_rate": 3.867855423851944e-06, "loss": 1.173, "step": 8998 }, { "epoch": 0.6887341190877085, "grad_norm": 0.563063803955956, "learning_rate": 3.867016582028333e-06, "loss": 1.3865, "step": 8999 }, { "epoch": 0.6888106536047758, "grad_norm": 0.6203788894828545, "learning_rate": 3.866177871261601e-06, "loss": 1.552, "step": 9000 }, { "epoch": 0.6888871881218429, "grad_norm": 0.6865791167263571, "learning_rate": 3.865339291603279e-06, "loss": 1.5089, "step": 9001 }, { "epoch": 0.6889637226389101, "grad_norm": 0.8869124519641248, "learning_rate": 3.864500843104892e-06, "loss": 1.455, "step": 9002 }, { "epoch": 0.6890402571559774, "grad_norm": 0.583661410965773, "learning_rate": 3.863662525817957e-06, "loss": 1.3913, "step": 9003 }, { "epoch": 0.6891167916730445, "grad_norm": 0.7319844422820138, "learning_rate": 3.862824339793983e-06, "loss": 1.5258, "step": 9004 }, { "epoch": 0.6891933261901118, "grad_norm": 0.7868477368742384, "learning_rate": 3.861986285084472e-06, "loss": 1.337, "step": 9005 }, { "epoch": 0.6892698607071789, "grad_norm": 0.7378313130927089, "learning_rate": 3.861148361740914e-06, "loss": 1.6842, "step": 9006 }, { "epoch": 0.6893463952242461, "grad_norm": 0.5399288245615889, "learning_rate": 3.860310569814795e-06, "loss": 1.1002, "step": 9007 }, { "epoch": 0.6894229297413134, "grad_norm": 0.5528786346479969, "learning_rate": 3.8594729093575885e-06, "loss": 1.2116, "step": 9008 }, { "epoch": 0.6894994642583805, "grad_norm": 0.6254549323590394, "learning_rate": 3.858635380420768e-06, "loss": 1.1991, "step": 9009 }, { "epoch": 0.6895759987754477, "grad_norm": 0.8010258411158707, "learning_rate": 3.85779798305579e-06, "loss": 1.9544, "step": 9010 }, { "epoch": 0.689652533292515, "grad_norm": 0.5155557347299723, "learning_rate": 3.856960717314109e-06, "loss": 1.1938, "step": 9011 }, { "epoch": 0.6897290678095821, "grad_norm": 0.5563122688567494, "learning_rate": 3.8561235832471675e-06, "loss": 1.1699, "step": 9012 }, { "epoch": 0.6898056023266493, "grad_norm": 0.6873561456288677, "learning_rate": 3.8552865809064025e-06, "loss": 1.7987, "step": 9013 }, { "epoch": 0.6898821368437165, "grad_norm": 0.6422108969442093, "learning_rate": 3.854449710343241e-06, "loss": 1.6588, "step": 9014 }, { "epoch": 0.6899586713607837, "grad_norm": 0.5668454531058865, "learning_rate": 3.8536129716091035e-06, "loss": 1.306, "step": 9015 }, { "epoch": 0.6900352058778509, "grad_norm": 0.7462554311824622, "learning_rate": 3.852776364755401e-06, "loss": 1.7638, "step": 9016 }, { "epoch": 0.6901117403949181, "grad_norm": 0.551021239405211, "learning_rate": 3.851939889833538e-06, "loss": 1.3518, "step": 9017 }, { "epoch": 0.6901882749119853, "grad_norm": 0.6048212329362748, "learning_rate": 3.85110354689491e-06, "loss": 1.596, "step": 9018 }, { "epoch": 0.6902648094290526, "grad_norm": 0.57059996115502, "learning_rate": 3.850267335990902e-06, "loss": 1.2393, "step": 9019 }, { "epoch": 0.6903413439461197, "grad_norm": 0.5787636213824666, "learning_rate": 3.849431257172897e-06, "loss": 1.3951, "step": 9020 }, { "epoch": 0.6904178784631869, "grad_norm": 0.6599812668504745, "learning_rate": 3.848595310492264e-06, "loss": 1.4473, "step": 9021 }, { "epoch": 0.690494412980254, "grad_norm": 0.5322341394549136, "learning_rate": 3.847759496000369e-06, "loss": 1.0297, "step": 9022 }, { "epoch": 0.6905709474973213, "grad_norm": 0.6149551487001262, "learning_rate": 3.846923813748562e-06, "loss": 1.435, "step": 9023 }, { "epoch": 0.6906474820143885, "grad_norm": 0.710418842995192, "learning_rate": 3.846088263788193e-06, "loss": 1.4486, "step": 9024 }, { "epoch": 0.6907240165314557, "grad_norm": 0.585108914504839, "learning_rate": 3.845252846170598e-06, "loss": 1.4384, "step": 9025 }, { "epoch": 0.6908005510485229, "grad_norm": 0.6146055293918421, "learning_rate": 3.844417560947113e-06, "loss": 1.5933, "step": 9026 }, { "epoch": 0.6908770855655901, "grad_norm": 0.5679004548400561, "learning_rate": 3.8435824081690556e-06, "loss": 1.3697, "step": 9027 }, { "epoch": 0.6909536200826573, "grad_norm": 0.5249168490072259, "learning_rate": 3.842747387887741e-06, "loss": 1.1868, "step": 9028 }, { "epoch": 0.6910301545997245, "grad_norm": 0.630161950383846, "learning_rate": 3.841912500154476e-06, "loss": 1.3881, "step": 9029 }, { "epoch": 0.6911066891167916, "grad_norm": 0.5999830011902146, "learning_rate": 3.841077745020558e-06, "loss": 1.3073, "step": 9030 }, { "epoch": 0.6911832236338589, "grad_norm": 0.5217324655137086, "learning_rate": 3.8402431225372785e-06, "loss": 1.0249, "step": 9031 }, { "epoch": 0.6912597581509261, "grad_norm": 0.698196519510088, "learning_rate": 3.839408632755918e-06, "loss": 1.309, "step": 9032 }, { "epoch": 0.6913362926679932, "grad_norm": 0.5412246790455346, "learning_rate": 3.83857427572775e-06, "loss": 1.2078, "step": 9033 }, { "epoch": 0.6914128271850605, "grad_norm": 0.595041908205269, "learning_rate": 3.837740051504039e-06, "loss": 1.5804, "step": 9034 }, { "epoch": 0.6914893617021277, "grad_norm": 0.9360929532447493, "learning_rate": 3.836905960136044e-06, "loss": 1.676, "step": 9035 }, { "epoch": 0.6915658962191948, "grad_norm": 0.5127691345018386, "learning_rate": 3.836072001675013e-06, "loss": 1.1444, "step": 9036 }, { "epoch": 0.6916424307362621, "grad_norm": 1.1816610283527025, "learning_rate": 3.835238176172188e-06, "loss": 1.409, "step": 9037 }, { "epoch": 0.6917189652533292, "grad_norm": 0.7040155801335677, "learning_rate": 3.8344044836788e-06, "loss": 1.6571, "step": 9038 }, { "epoch": 0.6917954997703964, "grad_norm": 0.6617771984640806, "learning_rate": 3.833570924246074e-06, "loss": 1.5099, "step": 9039 }, { "epoch": 0.6918720342874637, "grad_norm": 0.6200658157261213, "learning_rate": 3.832737497925227e-06, "loss": 1.4482, "step": 9040 }, { "epoch": 0.6919485688045308, "grad_norm": 0.5341822743563607, "learning_rate": 3.831904204767466e-06, "loss": 1.2183, "step": 9041 }, { "epoch": 0.6920251033215981, "grad_norm": 0.6791464057808706, "learning_rate": 3.831071044823994e-06, "loss": 1.7782, "step": 9042 }, { "epoch": 0.6921016378386652, "grad_norm": 0.5487510672130125, "learning_rate": 3.8302380181459995e-06, "loss": 1.0409, "step": 9043 }, { "epoch": 0.6921781723557324, "grad_norm": 0.6016355138261188, "learning_rate": 3.829405124784669e-06, "loss": 1.4782, "step": 9044 }, { "epoch": 0.6922547068727997, "grad_norm": 1.0137394143918843, "learning_rate": 3.8285723647911756e-06, "loss": 1.5045, "step": 9045 }, { "epoch": 0.6923312413898668, "grad_norm": 0.54099232335714, "learning_rate": 3.827739738216687e-06, "loss": 1.1709, "step": 9046 }, { "epoch": 0.692407775906934, "grad_norm": 0.5804143666421769, "learning_rate": 3.826907245112362e-06, "loss": 1.3634, "step": 9047 }, { "epoch": 0.6924843104240013, "grad_norm": 0.5200608501048143, "learning_rate": 3.826074885529354e-06, "loss": 1.1385, "step": 9048 }, { "epoch": 0.6925608449410684, "grad_norm": 0.5253972858579231, "learning_rate": 3.825242659518803e-06, "loss": 1.2617, "step": 9049 }, { "epoch": 0.6926373794581356, "grad_norm": 0.9396706526602863, "learning_rate": 3.824410567131844e-06, "loss": 1.8864, "step": 9050 }, { "epoch": 0.6927139139752028, "grad_norm": 0.6127623157440063, "learning_rate": 3.8235786084196045e-06, "loss": 1.4495, "step": 9051 }, { "epoch": 0.69279044849227, "grad_norm": 0.7098327558011249, "learning_rate": 3.8227467834331996e-06, "loss": 1.8877, "step": 9052 }, { "epoch": 0.6928669830093372, "grad_norm": 0.6461460088841383, "learning_rate": 3.821915092223743e-06, "loss": 1.6754, "step": 9053 }, { "epoch": 0.6929435175264044, "grad_norm": 0.5746949320811175, "learning_rate": 3.821083534842334e-06, "loss": 1.3473, "step": 9054 }, { "epoch": 0.6930200520434716, "grad_norm": 0.5502657916840924, "learning_rate": 3.820252111340066e-06, "loss": 1.3162, "step": 9055 }, { "epoch": 0.6930965865605389, "grad_norm": 1.4055171512122293, "learning_rate": 3.819420821768025e-06, "loss": 1.2455, "step": 9056 }, { "epoch": 0.693173121077606, "grad_norm": 0.5980845341071164, "learning_rate": 3.818589666177288e-06, "loss": 1.4083, "step": 9057 }, { "epoch": 0.6932496555946732, "grad_norm": 0.6316369517380173, "learning_rate": 3.8177586446189215e-06, "loss": 1.4474, "step": 9058 }, { "epoch": 0.6933261901117403, "grad_norm": 0.6979454713564993, "learning_rate": 3.816927757143989e-06, "loss": 1.4566, "step": 9059 }, { "epoch": 0.6934027246288076, "grad_norm": 0.8709536315299393, "learning_rate": 3.81609700380354e-06, "loss": 1.429, "step": 9060 }, { "epoch": 0.6934792591458748, "grad_norm": 0.6207051904853947, "learning_rate": 3.81526638464862e-06, "loss": 1.4002, "step": 9061 }, { "epoch": 0.693555793662942, "grad_norm": 0.58275347979991, "learning_rate": 3.8144358997302654e-06, "loss": 1.502, "step": 9062 }, { "epoch": 0.6936323281800092, "grad_norm": 0.5509092151194319, "learning_rate": 3.8136055490995004e-06, "loss": 1.2646, "step": 9063 }, { "epoch": 0.6937088626970764, "grad_norm": 0.7408734667994605, "learning_rate": 3.8127753328073482e-06, "loss": 1.5611, "step": 9064 }, { "epoch": 0.6937853972141436, "grad_norm": 0.6923251853147963, "learning_rate": 3.8119452509048183e-06, "loss": 1.5988, "step": 9065 }, { "epoch": 0.6938619317312108, "grad_norm": 0.5253094113339893, "learning_rate": 3.8111153034429126e-06, "loss": 1.2925, "step": 9066 }, { "epoch": 0.6939384662482779, "grad_norm": 0.6305142334630934, "learning_rate": 3.8102854904726273e-06, "loss": 1.4246, "step": 9067 }, { "epoch": 0.6940150007653452, "grad_norm": 0.6869203227124505, "learning_rate": 3.8094558120449464e-06, "loss": 1.5871, "step": 9068 }, { "epoch": 0.6940915352824124, "grad_norm": 0.57274951251568, "learning_rate": 3.8086262682108475e-06, "loss": 1.2765, "step": 9069 }, { "epoch": 0.6941680697994795, "grad_norm": 0.6353347317062101, "learning_rate": 3.8077968590213037e-06, "loss": 1.659, "step": 9070 }, { "epoch": 0.6942446043165468, "grad_norm": 0.8641616081577619, "learning_rate": 3.806967584527274e-06, "loss": 1.6124, "step": 9071 }, { "epoch": 0.6943211388336139, "grad_norm": 0.6329209367607217, "learning_rate": 3.8061384447797116e-06, "loss": 1.5607, "step": 9072 }, { "epoch": 0.6943976733506811, "grad_norm": 0.5813780113136221, "learning_rate": 3.805309439829562e-06, "loss": 1.3904, "step": 9073 }, { "epoch": 0.6944742078677484, "grad_norm": 0.714982678569841, "learning_rate": 3.80448056972776e-06, "loss": 1.5762, "step": 9074 }, { "epoch": 0.6945507423848155, "grad_norm": 0.5774206966073591, "learning_rate": 3.803651834525236e-06, "loss": 1.3005, "step": 9075 }, { "epoch": 0.6946272769018828, "grad_norm": 0.6395042981017736, "learning_rate": 3.8028232342729094e-06, "loss": 1.5234, "step": 9076 }, { "epoch": 0.69470381141895, "grad_norm": 0.8685452615428451, "learning_rate": 3.801994769021692e-06, "loss": 1.5683, "step": 9077 }, { "epoch": 0.6947803459360171, "grad_norm": 0.6036210169405132, "learning_rate": 3.8011664388224857e-06, "loss": 1.2556, "step": 9078 }, { "epoch": 0.6948568804530844, "grad_norm": 0.5344421872417757, "learning_rate": 3.800338243726187e-06, "loss": 1.2972, "step": 9079 }, { "epoch": 0.6949334149701515, "grad_norm": 0.6659605517704614, "learning_rate": 3.799510183783683e-06, "loss": 1.772, "step": 9080 }, { "epoch": 0.6950099494872187, "grad_norm": 0.665412125276059, "learning_rate": 3.798682259045851e-06, "loss": 1.3472, "step": 9081 }, { "epoch": 0.695086484004286, "grad_norm": 0.6914180502196652, "learning_rate": 3.7978544695635624e-06, "loss": 1.5939, "step": 9082 }, { "epoch": 0.6951630185213531, "grad_norm": 0.6219686602907634, "learning_rate": 3.797026815387678e-06, "loss": 1.4553, "step": 9083 }, { "epoch": 0.6952395530384203, "grad_norm": 0.6280117613455044, "learning_rate": 3.796199296569051e-06, "loss": 1.3197, "step": 9084 }, { "epoch": 0.6953160875554876, "grad_norm": 0.6199818055403185, "learning_rate": 3.7953719131585266e-06, "loss": 1.6569, "step": 9085 }, { "epoch": 0.6953926220725547, "grad_norm": 0.5831972744576986, "learning_rate": 3.794544665206943e-06, "loss": 1.3429, "step": 9086 }, { "epoch": 0.6954691565896219, "grad_norm": 0.5343584852858967, "learning_rate": 3.7937175527651294e-06, "loss": 1.2832, "step": 9087 }, { "epoch": 0.6955456911066891, "grad_norm": 0.8477272678420985, "learning_rate": 3.7928905758839042e-06, "loss": 1.7714, "step": 9088 }, { "epoch": 0.6956222256237563, "grad_norm": 0.6396419784642526, "learning_rate": 3.79206373461408e-06, "loss": 1.5383, "step": 9089 }, { "epoch": 0.6956987601408235, "grad_norm": 0.6123575801981952, "learning_rate": 3.7912370290064592e-06, "loss": 1.4493, "step": 9090 }, { "epoch": 0.6957752946578907, "grad_norm": 0.5952726372715827, "learning_rate": 3.7904104591118398e-06, "loss": 1.3518, "step": 9091 }, { "epoch": 0.6958518291749579, "grad_norm": 0.6024946740883719, "learning_rate": 3.7895840249810075e-06, "loss": 1.4431, "step": 9092 }, { "epoch": 0.6959283636920252, "grad_norm": 0.541437192459582, "learning_rate": 3.7887577266647402e-06, "loss": 1.0828, "step": 9093 }, { "epoch": 0.6960048982090923, "grad_norm": 0.7970181067666314, "learning_rate": 3.787931564213808e-06, "loss": 1.59, "step": 9094 }, { "epoch": 0.6960814327261595, "grad_norm": 0.6422663561618461, "learning_rate": 3.7871055376789738e-06, "loss": 1.5656, "step": 9095 }, { "epoch": 0.6961579672432266, "grad_norm": 0.5213528402149034, "learning_rate": 3.7862796471109897e-06, "loss": 1.0082, "step": 9096 }, { "epoch": 0.6962345017602939, "grad_norm": 0.5455650541785827, "learning_rate": 3.785453892560602e-06, "loss": 1.3156, "step": 9097 }, { "epoch": 0.6963110362773611, "grad_norm": 0.4754177442678969, "learning_rate": 3.7846282740785488e-06, "loss": 0.8809, "step": 9098 }, { "epoch": 0.6963875707944283, "grad_norm": 0.5796216030206854, "learning_rate": 3.7838027917155562e-06, "loss": 1.3467, "step": 9099 }, { "epoch": 0.6964641053114955, "grad_norm": 0.5636068976678434, "learning_rate": 3.782977445522346e-06, "loss": 1.2829, "step": 9100 }, { "epoch": 0.6965406398285627, "grad_norm": 0.6746810432109074, "learning_rate": 3.7821522355496287e-06, "loss": 1.5007, "step": 9101 }, { "epoch": 0.6966171743456299, "grad_norm": 0.6148612233993178, "learning_rate": 3.7813271618481085e-06, "loss": 1.5788, "step": 9102 }, { "epoch": 0.6966937088626971, "grad_norm": 0.5172357840729952, "learning_rate": 3.7805022244684796e-06, "loss": 1.1561, "step": 9103 }, { "epoch": 0.6967702433797642, "grad_norm": 0.5658688071633736, "learning_rate": 3.7796774234614296e-06, "loss": 1.2868, "step": 9104 }, { "epoch": 0.6968467778968315, "grad_norm": 0.6399834585233201, "learning_rate": 3.7788527588776357e-06, "loss": 1.5664, "step": 9105 }, { "epoch": 0.6969233124138987, "grad_norm": 0.8141535547128175, "learning_rate": 3.7780282307677687e-06, "loss": 1.3886, "step": 9106 }, { "epoch": 0.6969998469309658, "grad_norm": 0.5608829944930571, "learning_rate": 3.777203839182488e-06, "loss": 1.2781, "step": 9107 }, { "epoch": 0.6970763814480331, "grad_norm": 0.5626341384141891, "learning_rate": 3.776379584172449e-06, "loss": 1.1582, "step": 9108 }, { "epoch": 0.6971529159651002, "grad_norm": 0.5065962474540864, "learning_rate": 3.7755554657882963e-06, "loss": 1.1561, "step": 9109 }, { "epoch": 0.6972294504821674, "grad_norm": 0.5700318578901119, "learning_rate": 3.774731484080666e-06, "loss": 1.4039, "step": 9110 }, { "epoch": 0.6973059849992347, "grad_norm": 0.5924703579390427, "learning_rate": 3.773907639100185e-06, "loss": 1.3943, "step": 9111 }, { "epoch": 0.6973825195163018, "grad_norm": 0.4930980176703767, "learning_rate": 3.773083930897472e-06, "loss": 1.1361, "step": 9112 }, { "epoch": 0.697459054033369, "grad_norm": 0.554271983091455, "learning_rate": 3.772260359523142e-06, "loss": 1.2685, "step": 9113 }, { "epoch": 0.6975355885504363, "grad_norm": 0.6148973870031007, "learning_rate": 3.7714369250277937e-06, "loss": 1.4244, "step": 9114 }, { "epoch": 0.6976121230675034, "grad_norm": 0.6777523963214159, "learning_rate": 3.7706136274620242e-06, "loss": 1.5711, "step": 9115 }, { "epoch": 0.6976886575845707, "grad_norm": 0.6680381647787295, "learning_rate": 3.769790466876417e-06, "loss": 1.537, "step": 9116 }, { "epoch": 0.6977651921016378, "grad_norm": 0.5352045718752424, "learning_rate": 3.768967443321551e-06, "loss": 1.1121, "step": 9117 }, { "epoch": 0.697841726618705, "grad_norm": 0.5477560645455708, "learning_rate": 3.7681445568479937e-06, "loss": 1.2071, "step": 9118 }, { "epoch": 0.6979182611357723, "grad_norm": 0.596161018213261, "learning_rate": 3.767321807506307e-06, "loss": 1.397, "step": 9119 }, { "epoch": 0.6979947956528394, "grad_norm": 0.6500351396695984, "learning_rate": 3.7664991953470442e-06, "loss": 1.5519, "step": 9120 }, { "epoch": 0.6980713301699066, "grad_norm": 0.6891239345076335, "learning_rate": 3.765676720420747e-06, "loss": 1.4526, "step": 9121 }, { "epoch": 0.6981478646869739, "grad_norm": 0.6570606988531685, "learning_rate": 3.7648543827779515e-06, "loss": 1.4422, "step": 9122 }, { "epoch": 0.698224399204041, "grad_norm": 0.9750413624564611, "learning_rate": 3.764032182469185e-06, "loss": 2.0192, "step": 9123 }, { "epoch": 0.6983009337211082, "grad_norm": 0.664708339309831, "learning_rate": 3.7632101195449655e-06, "loss": 1.2012, "step": 9124 }, { "epoch": 0.6983774682381754, "grad_norm": 0.6783968581562765, "learning_rate": 3.762388194055803e-06, "loss": 1.5709, "step": 9125 }, { "epoch": 0.6984540027552426, "grad_norm": 0.5839682188233201, "learning_rate": 3.7615664060521984e-06, "loss": 1.183, "step": 9126 }, { "epoch": 0.6985305372723098, "grad_norm": 0.624646027110552, "learning_rate": 3.7607447555846455e-06, "loss": 1.4688, "step": 9127 }, { "epoch": 0.698607071789377, "grad_norm": 0.5103381359827839, "learning_rate": 3.759923242703629e-06, "loss": 1.1412, "step": 9128 }, { "epoch": 0.6986836063064442, "grad_norm": 0.4957961494025573, "learning_rate": 3.759101867459624e-06, "loss": 0.9447, "step": 9129 }, { "epoch": 0.6987601408235115, "grad_norm": 0.872838936464633, "learning_rate": 3.7582806299031006e-06, "loss": 1.817, "step": 9130 }, { "epoch": 0.6988366753405786, "grad_norm": 1.3436474980888584, "learning_rate": 3.757459530084517e-06, "loss": 1.3174, "step": 9131 }, { "epoch": 0.6989132098576458, "grad_norm": 0.6494730708805266, "learning_rate": 3.756638568054324e-06, "loss": 1.4447, "step": 9132 }, { "epoch": 0.698989744374713, "grad_norm": 0.6830977379333046, "learning_rate": 3.755817743862963e-06, "loss": 1.6247, "step": 9133 }, { "epoch": 0.6990662788917802, "grad_norm": 0.5706569368890013, "learning_rate": 3.754997057560867e-06, "loss": 1.4742, "step": 9134 }, { "epoch": 0.6991428134088474, "grad_norm": 0.5938205876445017, "learning_rate": 3.754176509198465e-06, "loss": 1.3106, "step": 9135 }, { "epoch": 0.6992193479259146, "grad_norm": 1.3599388704285944, "learning_rate": 3.753356098826172e-06, "loss": 1.3991, "step": 9136 }, { "epoch": 0.6992958824429818, "grad_norm": 0.5717111231680685, "learning_rate": 3.752535826494396e-06, "loss": 1.3234, "step": 9137 }, { "epoch": 0.699372416960049, "grad_norm": 0.7266514291633938, "learning_rate": 3.751715692253538e-06, "loss": 1.3825, "step": 9138 }, { "epoch": 0.6994489514771162, "grad_norm": 0.6590947523476416, "learning_rate": 3.750895696153989e-06, "loss": 1.5153, "step": 9139 }, { "epoch": 0.6995254859941834, "grad_norm": 0.5502756841448688, "learning_rate": 3.75007583824613e-06, "loss": 1.3222, "step": 9140 }, { "epoch": 0.6996020205112505, "grad_norm": 0.5789769823415142, "learning_rate": 3.7492561185803396e-06, "loss": 1.3208, "step": 9141 }, { "epoch": 0.6996785550283178, "grad_norm": 0.4772681520832474, "learning_rate": 3.748436537206982e-06, "loss": 1.1382, "step": 9142 }, { "epoch": 0.699755089545385, "grad_norm": 0.5501389213984609, "learning_rate": 3.7476170941764135e-06, "loss": 1.2692, "step": 9143 }, { "epoch": 0.6998316240624521, "grad_norm": 0.5968851215800856, "learning_rate": 3.7467977895389838e-06, "loss": 1.3857, "step": 9144 }, { "epoch": 0.6999081585795194, "grad_norm": 0.5258599241351215, "learning_rate": 3.7459786233450345e-06, "loss": 1.1404, "step": 9145 }, { "epoch": 0.6999846930965865, "grad_norm": 0.5813453694853905, "learning_rate": 3.7451595956448972e-06, "loss": 1.4132, "step": 9146 }, { "epoch": 0.7000612276136537, "grad_norm": 0.5486117167694502, "learning_rate": 3.7443407064888947e-06, "loss": 1.3648, "step": 9147 }, { "epoch": 0.700137762130721, "grad_norm": 0.5110926395200375, "learning_rate": 3.7435219559273418e-06, "loss": 1.0462, "step": 9148 }, { "epoch": 0.7002142966477881, "grad_norm": 0.6437746935529469, "learning_rate": 3.742703344010545e-06, "loss": 1.6656, "step": 9149 }, { "epoch": 0.7002908311648554, "grad_norm": 0.5448962838315686, "learning_rate": 3.7418848707888046e-06, "loss": 1.2923, "step": 9150 }, { "epoch": 0.7003673656819226, "grad_norm": 0.7337687956758812, "learning_rate": 3.7410665363124055e-06, "loss": 1.5264, "step": 9151 }, { "epoch": 0.7004439001989897, "grad_norm": 0.5468000592419039, "learning_rate": 3.740248340631633e-06, "loss": 1.2315, "step": 9152 }, { "epoch": 0.700520434716057, "grad_norm": 0.5392682164460401, "learning_rate": 3.7394302837967577e-06, "loss": 1.1847, "step": 9153 }, { "epoch": 0.7005969692331241, "grad_norm": 0.593061272682138, "learning_rate": 3.7386123658580443e-06, "loss": 1.3954, "step": 9154 }, { "epoch": 0.7006735037501913, "grad_norm": 0.6279074927472456, "learning_rate": 3.7377945868657474e-06, "loss": 1.5263, "step": 9155 }, { "epoch": 0.7007500382672586, "grad_norm": 0.788930725649073, "learning_rate": 3.736976946870111e-06, "loss": 1.5897, "step": 9156 }, { "epoch": 0.7008265727843257, "grad_norm": 0.5770396190466291, "learning_rate": 3.7361594459213784e-06, "loss": 1.3194, "step": 9157 }, { "epoch": 0.7009031073013929, "grad_norm": 0.5458970561340941, "learning_rate": 3.7353420840697775e-06, "loss": 1.2834, "step": 9158 }, { "epoch": 0.7009796418184602, "grad_norm": 0.616304321217551, "learning_rate": 3.7345248613655285e-06, "loss": 1.2225, "step": 9159 }, { "epoch": 0.7010561763355273, "grad_norm": 0.5730172597195454, "learning_rate": 3.7337077778588436e-06, "loss": 1.2652, "step": 9160 }, { "epoch": 0.7011327108525945, "grad_norm": 0.6138599303921844, "learning_rate": 3.7328908335999293e-06, "loss": 1.5781, "step": 9161 }, { "epoch": 0.7012092453696617, "grad_norm": 0.9395485481140048, "learning_rate": 3.732074028638978e-06, "loss": 1.6455, "step": 9162 }, { "epoch": 0.7012857798867289, "grad_norm": 0.5457040675947523, "learning_rate": 3.731257363026178e-06, "loss": 1.2786, "step": 9163 }, { "epoch": 0.7013623144037961, "grad_norm": 1.3910686772138317, "learning_rate": 3.7304408368117094e-06, "loss": 1.701, "step": 9164 }, { "epoch": 0.7014388489208633, "grad_norm": 0.5258832180183507, "learning_rate": 3.72962445004574e-06, "loss": 1.317, "step": 9165 }, { "epoch": 0.7015153834379305, "grad_norm": 0.5741529108584655, "learning_rate": 3.7288082027784323e-06, "loss": 1.2358, "step": 9166 }, { "epoch": 0.7015919179549978, "grad_norm": 0.5655119300670275, "learning_rate": 3.727992095059937e-06, "loss": 1.2672, "step": 9167 }, { "epoch": 0.7016684524720649, "grad_norm": 0.748427635022125, "learning_rate": 3.7271761269403994e-06, "loss": 1.5439, "step": 9168 }, { "epoch": 0.7017449869891321, "grad_norm": 0.5028581884333759, "learning_rate": 3.7263602984699565e-06, "loss": 0.9785, "step": 9169 }, { "epoch": 0.7018215215061993, "grad_norm": 0.5826960177166769, "learning_rate": 3.7255446096987324e-06, "loss": 1.4366, "step": 9170 }, { "epoch": 0.7018980560232665, "grad_norm": 0.5445441937151727, "learning_rate": 3.724729060676847e-06, "loss": 1.2291, "step": 9171 }, { "epoch": 0.7019745905403337, "grad_norm": 0.5913966785221817, "learning_rate": 3.7239136514544104e-06, "loss": 1.5028, "step": 9172 }, { "epoch": 0.7020511250574009, "grad_norm": 0.5262773973465896, "learning_rate": 3.7230983820815212e-06, "loss": 1.2249, "step": 9173 }, { "epoch": 0.7021276595744681, "grad_norm": 0.630799635091758, "learning_rate": 3.7222832526082753e-06, "loss": 1.5418, "step": 9174 }, { "epoch": 0.7022041940915352, "grad_norm": 0.5931804098246793, "learning_rate": 3.7214682630847567e-06, "loss": 1.3933, "step": 9175 }, { "epoch": 0.7022807286086025, "grad_norm": 0.6633412842629184, "learning_rate": 3.7206534135610383e-06, "loss": 1.2884, "step": 9176 }, { "epoch": 0.7023572631256697, "grad_norm": 0.6200758968768527, "learning_rate": 3.7198387040871886e-06, "loss": 1.5263, "step": 9177 }, { "epoch": 0.7024337976427368, "grad_norm": 0.5557583974752491, "learning_rate": 3.719024134713264e-06, "loss": 1.2228, "step": 9178 }, { "epoch": 0.7025103321598041, "grad_norm": 0.6688074544625064, "learning_rate": 3.7182097054893163e-06, "loss": 1.2379, "step": 9179 }, { "epoch": 0.7025868666768713, "grad_norm": 0.5957458002511801, "learning_rate": 3.717395416465386e-06, "loss": 1.4485, "step": 9180 }, { "epoch": 0.7026634011939384, "grad_norm": 0.848330929258682, "learning_rate": 3.7165812676915046e-06, "loss": 1.7726, "step": 9181 }, { "epoch": 0.7027399357110057, "grad_norm": 0.8917121370731464, "learning_rate": 3.7157672592176964e-06, "loss": 1.4408, "step": 9182 }, { "epoch": 0.7028164702280728, "grad_norm": 0.6062003990434336, "learning_rate": 3.714953391093976e-06, "loss": 1.4131, "step": 9183 }, { "epoch": 0.70289300474514, "grad_norm": 0.5079245256089814, "learning_rate": 3.7141396633703496e-06, "loss": 1.0924, "step": 9184 }, { "epoch": 0.7029695392622073, "grad_norm": 0.606589987864665, "learning_rate": 3.713326076096817e-06, "loss": 1.1481, "step": 9185 }, { "epoch": 0.7030460737792744, "grad_norm": 0.8322089063472526, "learning_rate": 3.7125126293233654e-06, "loss": 1.2765, "step": 9186 }, { "epoch": 0.7031226082963417, "grad_norm": 0.6749430789073239, "learning_rate": 3.711699323099977e-06, "loss": 1.3678, "step": 9187 }, { "epoch": 0.7031991428134089, "grad_norm": 0.5034176612460651, "learning_rate": 3.7108861574766225e-06, "loss": 1.0524, "step": 9188 }, { "epoch": 0.703275677330476, "grad_norm": 0.6284813116140257, "learning_rate": 3.7100731325032657e-06, "loss": 1.4746, "step": 9189 }, { "epoch": 0.7033522118475433, "grad_norm": 0.667848340881402, "learning_rate": 3.709260248229861e-06, "loss": 1.4295, "step": 9190 }, { "epoch": 0.7034287463646104, "grad_norm": 0.5940425253350939, "learning_rate": 3.708447504706355e-06, "loss": 1.1311, "step": 9191 }, { "epoch": 0.7035052808816776, "grad_norm": 0.6680160421694565, "learning_rate": 3.707634901982685e-06, "loss": 1.6163, "step": 9192 }, { "epoch": 0.7035818153987449, "grad_norm": 0.5225292342502337, "learning_rate": 3.706822440108779e-06, "loss": 1.1035, "step": 9193 }, { "epoch": 0.703658349915812, "grad_norm": 0.5498076794688117, "learning_rate": 3.7060101191345577e-06, "loss": 1.3247, "step": 9194 }, { "epoch": 0.7037348844328792, "grad_norm": 0.596773445563037, "learning_rate": 3.70519793910993e-06, "loss": 1.3931, "step": 9195 }, { "epoch": 0.7038114189499465, "grad_norm": 0.5041937537250043, "learning_rate": 3.704385900084804e-06, "loss": 0.9817, "step": 9196 }, { "epoch": 0.7038879534670136, "grad_norm": 0.5684597841299536, "learning_rate": 3.7035740021090707e-06, "loss": 1.2549, "step": 9197 }, { "epoch": 0.7039644879840808, "grad_norm": 0.5433518214601477, "learning_rate": 3.7027622452326152e-06, "loss": 1.2957, "step": 9198 }, { "epoch": 0.704041022501148, "grad_norm": 0.5622742056648714, "learning_rate": 3.7019506295053147e-06, "loss": 1.2892, "step": 9199 }, { "epoch": 0.7041175570182152, "grad_norm": 0.665540552569054, "learning_rate": 3.701139154977036e-06, "loss": 1.6477, "step": 9200 }, { "epoch": 0.7041940915352825, "grad_norm": 0.6611010449446456, "learning_rate": 3.7003278216976417e-06, "loss": 1.5233, "step": 9201 }, { "epoch": 0.7042706260523496, "grad_norm": 0.5965439374234021, "learning_rate": 3.6995166297169803e-06, "loss": 1.228, "step": 9202 }, { "epoch": 0.7043471605694168, "grad_norm": 0.5712133861500576, "learning_rate": 3.6987055790848946e-06, "loss": 1.3945, "step": 9203 }, { "epoch": 0.7044236950864841, "grad_norm": 1.2223603138810584, "learning_rate": 3.697894669851218e-06, "loss": 1.52, "step": 9204 }, { "epoch": 0.7045002296035512, "grad_norm": 0.5754276376068319, "learning_rate": 3.6970839020657746e-06, "loss": 1.4126, "step": 9205 }, { "epoch": 0.7045767641206184, "grad_norm": 0.6509667032983498, "learning_rate": 3.696273275778379e-06, "loss": 1.7281, "step": 9206 }, { "epoch": 0.7046532986376856, "grad_norm": 0.9098420126234883, "learning_rate": 3.695462791038842e-06, "loss": 1.208, "step": 9207 }, { "epoch": 0.7047298331547528, "grad_norm": 0.5867842854242824, "learning_rate": 3.6946524478969604e-06, "loss": 1.2098, "step": 9208 }, { "epoch": 0.70480636767182, "grad_norm": 0.5518024725689197, "learning_rate": 3.6938422464025234e-06, "loss": 1.3038, "step": 9209 }, { "epoch": 0.7048829021888872, "grad_norm": 0.6245485822596466, "learning_rate": 3.693032186605313e-06, "loss": 1.5111, "step": 9210 }, { "epoch": 0.7049594367059544, "grad_norm": 0.6976302377075969, "learning_rate": 3.692222268555101e-06, "loss": 1.5196, "step": 9211 }, { "epoch": 0.7050359712230215, "grad_norm": 0.5454929748866477, "learning_rate": 3.6914124923016538e-06, "loss": 1.3638, "step": 9212 }, { "epoch": 0.7051125057400888, "grad_norm": 0.6518832260988916, "learning_rate": 3.690602857894723e-06, "loss": 1.6476, "step": 9213 }, { "epoch": 0.705189040257156, "grad_norm": 0.7223596746454536, "learning_rate": 3.689793365384057e-06, "loss": 1.2549, "step": 9214 }, { "epoch": 0.7052655747742231, "grad_norm": 0.6588674245034366, "learning_rate": 3.688984014819392e-06, "loss": 1.5832, "step": 9215 }, { "epoch": 0.7053421092912904, "grad_norm": 0.6200675474952422, "learning_rate": 3.6881748062504587e-06, "loss": 1.2493, "step": 9216 }, { "epoch": 0.7054186438083576, "grad_norm": 0.516901062405459, "learning_rate": 3.6873657397269745e-06, "loss": 1.0841, "step": 9217 }, { "epoch": 0.7054951783254247, "grad_norm": 0.5829168066090928, "learning_rate": 3.6865568152986542e-06, "loss": 1.4618, "step": 9218 }, { "epoch": 0.705571712842492, "grad_norm": 0.6702596938273709, "learning_rate": 3.6857480330151985e-06, "loss": 1.543, "step": 9219 }, { "epoch": 0.7056482473595591, "grad_norm": 0.4895381212890864, "learning_rate": 3.6849393929263028e-06, "loss": 1.0546, "step": 9220 }, { "epoch": 0.7057247818766264, "grad_norm": 0.6187627318844949, "learning_rate": 3.6841308950816514e-06, "loss": 1.6615, "step": 9221 }, { "epoch": 0.7058013163936936, "grad_norm": 0.5613066228807558, "learning_rate": 3.683322539530919e-06, "loss": 1.4708, "step": 9222 }, { "epoch": 0.7058778509107607, "grad_norm": 0.6271361746309653, "learning_rate": 3.682514326323777e-06, "loss": 1.6182, "step": 9223 }, { "epoch": 0.705954385427828, "grad_norm": 0.7686285929676914, "learning_rate": 3.6817062555098827e-06, "loss": 1.7898, "step": 9224 }, { "epoch": 0.7060309199448952, "grad_norm": 0.6131924381099199, "learning_rate": 3.6808983271388865e-06, "loss": 1.5897, "step": 9225 }, { "epoch": 0.7061074544619623, "grad_norm": 0.5052808645530295, "learning_rate": 3.680090541260429e-06, "loss": 1.0893, "step": 9226 }, { "epoch": 0.7061839889790296, "grad_norm": 0.8719436677443362, "learning_rate": 3.6792828979241436e-06, "loss": 1.4429, "step": 9227 }, { "epoch": 0.7062605234960967, "grad_norm": 0.6040507462243875, "learning_rate": 3.678475397179656e-06, "loss": 1.1962, "step": 9228 }, { "epoch": 0.7063370580131639, "grad_norm": 0.5833347386339593, "learning_rate": 3.6776680390765796e-06, "loss": 1.3885, "step": 9229 }, { "epoch": 0.7064135925302312, "grad_norm": 0.6024768543601625, "learning_rate": 3.676860823664521e-06, "loss": 1.4851, "step": 9230 }, { "epoch": 0.7064901270472983, "grad_norm": 0.5660031664506063, "learning_rate": 3.676053750993079e-06, "loss": 1.3453, "step": 9231 }, { "epoch": 0.7065666615643655, "grad_norm": 0.49641900942037315, "learning_rate": 3.6752468211118413e-06, "loss": 1.0818, "step": 9232 }, { "epoch": 0.7066431960814328, "grad_norm": 0.568567732219471, "learning_rate": 3.6744400340703877e-06, "loss": 1.319, "step": 9233 }, { "epoch": 0.7067197305984999, "grad_norm": 0.6407319644360999, "learning_rate": 3.673633389918292e-06, "loss": 1.5705, "step": 9234 }, { "epoch": 0.7067962651155671, "grad_norm": 0.654046306357474, "learning_rate": 3.672826888705114e-06, "loss": 1.6197, "step": 9235 }, { "epoch": 0.7068727996326343, "grad_norm": 0.57673171402852, "learning_rate": 3.6720205304804103e-06, "loss": 1.327, "step": 9236 }, { "epoch": 0.7069493341497015, "grad_norm": 0.5769776038534349, "learning_rate": 3.671214315293724e-06, "loss": 1.3394, "step": 9237 }, { "epoch": 0.7070258686667688, "grad_norm": 0.6653592076370427, "learning_rate": 3.670408243194592e-06, "loss": 1.5147, "step": 9238 }, { "epoch": 0.7071024031838359, "grad_norm": 0.5798893598196344, "learning_rate": 3.669602314232542e-06, "loss": 1.4233, "step": 9239 }, { "epoch": 0.7071789377009031, "grad_norm": 0.5858153262902254, "learning_rate": 3.668796528457092e-06, "loss": 1.3192, "step": 9240 }, { "epoch": 0.7072554722179702, "grad_norm": 0.5260572049494906, "learning_rate": 3.6679908859177527e-06, "loss": 1.1198, "step": 9241 }, { "epoch": 0.7073320067350375, "grad_norm": 0.6278983588030205, "learning_rate": 3.667185386664024e-06, "loss": 1.4479, "step": 9242 }, { "epoch": 0.7074085412521047, "grad_norm": 1.6121426971496395, "learning_rate": 3.666380030745399e-06, "loss": 1.687, "step": 9243 }, { "epoch": 0.7074850757691719, "grad_norm": 0.6008069140778987, "learning_rate": 3.6655748182113606e-06, "loss": 1.5286, "step": 9244 }, { "epoch": 0.7075616102862391, "grad_norm": 0.8809833163524311, "learning_rate": 3.6647697491113843e-06, "loss": 1.8646, "step": 9245 }, { "epoch": 0.7076381448033063, "grad_norm": 0.5437558959676914, "learning_rate": 3.663964823494936e-06, "loss": 1.2154, "step": 9246 }, { "epoch": 0.7077146793203735, "grad_norm": 0.6382583859714235, "learning_rate": 3.6631600414114726e-06, "loss": 1.4371, "step": 9247 }, { "epoch": 0.7077912138374407, "grad_norm": 0.8130952763462055, "learning_rate": 3.662355402910442e-06, "loss": 1.8808, "step": 9248 }, { "epoch": 0.7078677483545078, "grad_norm": 0.6999114999568896, "learning_rate": 3.661550908041282e-06, "loss": 1.7144, "step": 9249 }, { "epoch": 0.7079442828715751, "grad_norm": 0.6063441813407714, "learning_rate": 3.660746556853426e-06, "loss": 1.4227, "step": 9250 }, { "epoch": 0.7080208173886423, "grad_norm": 0.615135653361097, "learning_rate": 3.659942349396296e-06, "loss": 1.5131, "step": 9251 }, { "epoch": 0.7080973519057094, "grad_norm": 0.5637501148848311, "learning_rate": 3.6591382857193026e-06, "loss": 1.2201, "step": 9252 }, { "epoch": 0.7081738864227767, "grad_norm": 0.5413623821811058, "learning_rate": 3.6583343658718496e-06, "loss": 1.1798, "step": 9253 }, { "epoch": 0.7082504209398439, "grad_norm": 0.5462335401543201, "learning_rate": 3.657530589903335e-06, "loss": 1.1863, "step": 9254 }, { "epoch": 0.708326955456911, "grad_norm": 0.6626954174641139, "learning_rate": 3.6567269578631413e-06, "loss": 1.6984, "step": 9255 }, { "epoch": 0.7084034899739783, "grad_norm": 0.5491307556418296, "learning_rate": 3.6559234698006506e-06, "loss": 1.197, "step": 9256 }, { "epoch": 0.7084800244910454, "grad_norm": 0.6379952387151285, "learning_rate": 3.655120125765229e-06, "loss": 1.5594, "step": 9257 }, { "epoch": 0.7085565590081127, "grad_norm": 0.65933792101956, "learning_rate": 3.6543169258062362e-06, "loss": 1.4426, "step": 9258 }, { "epoch": 0.7086330935251799, "grad_norm": 0.5120756116173105, "learning_rate": 3.6535138699730245e-06, "loss": 1.2293, "step": 9259 }, { "epoch": 0.708709628042247, "grad_norm": 0.619563931133094, "learning_rate": 3.6527109583149352e-06, "loss": 1.3038, "step": 9260 }, { "epoch": 0.7087861625593143, "grad_norm": 0.5560266183393655, "learning_rate": 3.6519081908813013e-06, "loss": 1.2561, "step": 9261 }, { "epoch": 0.7088626970763815, "grad_norm": 0.5528985636321566, "learning_rate": 3.651105567721448e-06, "loss": 1.4006, "step": 9262 }, { "epoch": 0.7089392315934486, "grad_norm": 0.5894633468830114, "learning_rate": 3.65030308888469e-06, "loss": 1.4892, "step": 9263 }, { "epoch": 0.7090157661105159, "grad_norm": 0.5279871160294279, "learning_rate": 3.649500754420334e-06, "loss": 1.2324, "step": 9264 }, { "epoch": 0.709092300627583, "grad_norm": 0.8956193219559065, "learning_rate": 3.6486985643776795e-06, "loss": 1.4364, "step": 9265 }, { "epoch": 0.7091688351446502, "grad_norm": 0.5625903684878029, "learning_rate": 3.647896518806013e-06, "loss": 1.197, "step": 9266 }, { "epoch": 0.7092453696617175, "grad_norm": 0.5485357573181349, "learning_rate": 3.647094617754616e-06, "loss": 1.2505, "step": 9267 }, { "epoch": 0.7093219041787846, "grad_norm": 0.582693631535858, "learning_rate": 3.646292861272761e-06, "loss": 1.4225, "step": 9268 }, { "epoch": 0.7093984386958518, "grad_norm": 0.4748499025749142, "learning_rate": 3.6454912494097073e-06, "loss": 0.9112, "step": 9269 }, { "epoch": 0.7094749732129191, "grad_norm": 0.6120650971432938, "learning_rate": 3.644689782214711e-06, "loss": 1.6831, "step": 9270 }, { "epoch": 0.7095515077299862, "grad_norm": 1.76946263193683, "learning_rate": 3.643888459737014e-06, "loss": 1.2058, "step": 9271 }, { "epoch": 0.7096280422470534, "grad_norm": 0.5802681089174909, "learning_rate": 3.643087282025854e-06, "loss": 1.309, "step": 9272 }, { "epoch": 0.7097045767641206, "grad_norm": 0.5798871632023425, "learning_rate": 3.6422862491304577e-06, "loss": 1.4943, "step": 9273 }, { "epoch": 0.7097811112811878, "grad_norm": 1.2804170615842738, "learning_rate": 3.641485361100043e-06, "loss": 1.577, "step": 9274 }, { "epoch": 0.709857645798255, "grad_norm": 0.5880226945069492, "learning_rate": 3.6406846179838175e-06, "loss": 1.4565, "step": 9275 }, { "epoch": 0.7099341803153222, "grad_norm": 0.8459700951112941, "learning_rate": 3.6398840198309828e-06, "loss": 1.7209, "step": 9276 }, { "epoch": 0.7100107148323894, "grad_norm": 0.6646481050670959, "learning_rate": 3.6390835666907278e-06, "loss": 1.751, "step": 9277 }, { "epoch": 0.7100872493494566, "grad_norm": 0.6096500362845996, "learning_rate": 3.6382832586122375e-06, "loss": 1.4132, "step": 9278 }, { "epoch": 0.7101637838665238, "grad_norm": 0.6082112211663068, "learning_rate": 3.6374830956446836e-06, "loss": 1.5622, "step": 9279 }, { "epoch": 0.710240318383591, "grad_norm": 0.6488263892655478, "learning_rate": 3.636683077837231e-06, "loss": 1.2338, "step": 9280 }, { "epoch": 0.7103168529006582, "grad_norm": 0.5622071433419003, "learning_rate": 3.635883205239036e-06, "loss": 1.2361, "step": 9281 }, { "epoch": 0.7103933874177254, "grad_norm": 0.6537297353812168, "learning_rate": 3.6350834778992426e-06, "loss": 1.1877, "step": 9282 }, { "epoch": 0.7104699219347926, "grad_norm": 0.5637113420927945, "learning_rate": 3.6342838958669914e-06, "loss": 1.1885, "step": 9283 }, { "epoch": 0.7105464564518598, "grad_norm": 0.5606010981081025, "learning_rate": 3.633484459191409e-06, "loss": 1.3037, "step": 9284 }, { "epoch": 0.710622990968927, "grad_norm": 0.6517102842802305, "learning_rate": 3.632685167921617e-06, "loss": 1.2411, "step": 9285 }, { "epoch": 0.7106995254859941, "grad_norm": 0.626607045878895, "learning_rate": 3.6318860221067243e-06, "loss": 1.5732, "step": 9286 }, { "epoch": 0.7107760600030614, "grad_norm": 0.6509152950627839, "learning_rate": 3.6310870217958337e-06, "loss": 1.532, "step": 9287 }, { "epoch": 0.7108525945201286, "grad_norm": 0.8069368235073797, "learning_rate": 3.6302881670380374e-06, "loss": 1.3477, "step": 9288 }, { "epoch": 0.7109291290371957, "grad_norm": 0.670749810219987, "learning_rate": 3.6294894578824215e-06, "loss": 1.5737, "step": 9289 }, { "epoch": 0.711005663554263, "grad_norm": 0.5656110789057531, "learning_rate": 3.628690894378059e-06, "loss": 1.1954, "step": 9290 }, { "epoch": 0.7110821980713302, "grad_norm": 0.7828026114037446, "learning_rate": 3.6278924765740176e-06, "loss": 1.2487, "step": 9291 }, { "epoch": 0.7111587325883973, "grad_norm": 0.550746504988256, "learning_rate": 3.6270942045193537e-06, "loss": 1.2546, "step": 9292 }, { "epoch": 0.7112352671054646, "grad_norm": 0.7343486382735794, "learning_rate": 3.6262960782631133e-06, "loss": 1.5916, "step": 9293 }, { "epoch": 0.7113118016225317, "grad_norm": 0.9058595946478054, "learning_rate": 3.6254980978543403e-06, "loss": 1.3531, "step": 9294 }, { "epoch": 0.711388336139599, "grad_norm": 0.5700565478304651, "learning_rate": 3.624700263342062e-06, "loss": 1.2651, "step": 9295 }, { "epoch": 0.7114648706566662, "grad_norm": 0.6006244557038838, "learning_rate": 3.623902574775301e-06, "loss": 1.297, "step": 9296 }, { "epoch": 0.7115414051737333, "grad_norm": 0.6694474142469189, "learning_rate": 3.6231050322030682e-06, "loss": 1.1353, "step": 9297 }, { "epoch": 0.7116179396908006, "grad_norm": 0.6165988794415479, "learning_rate": 3.622307635674368e-06, "loss": 1.4574, "step": 9298 }, { "epoch": 0.7116944742078678, "grad_norm": 0.5894859411677752, "learning_rate": 3.6215103852381927e-06, "loss": 1.3034, "step": 9299 }, { "epoch": 0.7117710087249349, "grad_norm": 0.6020744413880422, "learning_rate": 3.620713280943532e-06, "loss": 1.3597, "step": 9300 }, { "epoch": 0.7118475432420022, "grad_norm": 0.6458348142224124, "learning_rate": 3.6199163228393597e-06, "loss": 1.4644, "step": 9301 }, { "epoch": 0.7119240777590693, "grad_norm": 0.6167506662050803, "learning_rate": 3.6191195109746436e-06, "loss": 1.3758, "step": 9302 }, { "epoch": 0.7120006122761365, "grad_norm": 0.7171932845380274, "learning_rate": 3.618322845398342e-06, "loss": 1.4417, "step": 9303 }, { "epoch": 0.7120771467932038, "grad_norm": 0.7223925516192683, "learning_rate": 3.617526326159404e-06, "loss": 1.404, "step": 9304 }, { "epoch": 0.7121536813102709, "grad_norm": 0.6562277544983772, "learning_rate": 3.616729953306772e-06, "loss": 1.3252, "step": 9305 }, { "epoch": 0.7122302158273381, "grad_norm": 0.5919745557384923, "learning_rate": 3.615933726889376e-06, "loss": 1.3454, "step": 9306 }, { "epoch": 0.7123067503444054, "grad_norm": 0.8811326256620147, "learning_rate": 3.6151376469561383e-06, "loss": 1.6894, "step": 9307 }, { "epoch": 0.7123832848614725, "grad_norm": 0.5816143236810657, "learning_rate": 3.6143417135559732e-06, "loss": 1.4133, "step": 9308 }, { "epoch": 0.7124598193785397, "grad_norm": 0.657578251970754, "learning_rate": 3.6135459267377858e-06, "loss": 1.638, "step": 9309 }, { "epoch": 0.7125363538956069, "grad_norm": 0.5944944920724322, "learning_rate": 3.6127502865504683e-06, "loss": 1.3712, "step": 9310 }, { "epoch": 0.7126128884126741, "grad_norm": 0.6882214312882394, "learning_rate": 3.6119547930429112e-06, "loss": 1.7139, "step": 9311 }, { "epoch": 0.7126894229297414, "grad_norm": 0.5561230519447639, "learning_rate": 3.611159446263991e-06, "loss": 1.0673, "step": 9312 }, { "epoch": 0.7127659574468085, "grad_norm": 0.511159748164786, "learning_rate": 3.6103642462625754e-06, "loss": 1.1587, "step": 9313 }, { "epoch": 0.7128424919638757, "grad_norm": 0.8089369826333673, "learning_rate": 3.6095691930875244e-06, "loss": 1.8998, "step": 9314 }, { "epoch": 0.7129190264809429, "grad_norm": 0.8404073582342007, "learning_rate": 3.6087742867876875e-06, "loss": 1.2102, "step": 9315 }, { "epoch": 0.7129955609980101, "grad_norm": 0.4899606281735853, "learning_rate": 3.6079795274119066e-06, "loss": 1.0238, "step": 9316 }, { "epoch": 0.7130720955150773, "grad_norm": 0.5445869724707799, "learning_rate": 3.6071849150090154e-06, "loss": 1.1342, "step": 9317 }, { "epoch": 0.7131486300321445, "grad_norm": 0.7775040777689999, "learning_rate": 3.6063904496278355e-06, "loss": 1.6643, "step": 9318 }, { "epoch": 0.7132251645492117, "grad_norm": 1.8247237247110917, "learning_rate": 3.605596131317183e-06, "loss": 1.6433, "step": 9319 }, { "epoch": 0.7133016990662789, "grad_norm": 0.5665704470476276, "learning_rate": 3.604801960125861e-06, "loss": 1.2921, "step": 9320 }, { "epoch": 0.7133782335833461, "grad_norm": 0.5640867944338653, "learning_rate": 3.604007936102666e-06, "loss": 1.3478, "step": 9321 }, { "epoch": 0.7134547681004133, "grad_norm": 0.47560397635470747, "learning_rate": 3.603214059296387e-06, "loss": 0.9791, "step": 9322 }, { "epoch": 0.7135313026174804, "grad_norm": 0.5813096757411818, "learning_rate": 3.6024203297558015e-06, "loss": 1.3546, "step": 9323 }, { "epoch": 0.7136078371345477, "grad_norm": 0.5210001285171861, "learning_rate": 3.6016267475296786e-06, "loss": 1.1716, "step": 9324 }, { "epoch": 0.7136843716516149, "grad_norm": 0.6331286150239083, "learning_rate": 3.6008333126667776e-06, "loss": 1.3267, "step": 9325 }, { "epoch": 0.713760906168682, "grad_norm": 0.7680874861025787, "learning_rate": 3.6000400252158508e-06, "loss": 1.6965, "step": 9326 }, { "epoch": 0.7138374406857493, "grad_norm": 0.5429278141463034, "learning_rate": 3.599246885225639e-06, "loss": 1.135, "step": 9327 }, { "epoch": 0.7139139752028165, "grad_norm": 0.4998451685683802, "learning_rate": 3.598453892744875e-06, "loss": 1.0469, "step": 9328 }, { "epoch": 0.7139905097198836, "grad_norm": 0.5738002470491858, "learning_rate": 3.5976610478222828e-06, "loss": 1.4542, "step": 9329 }, { "epoch": 0.7140670442369509, "grad_norm": 0.5238025352718774, "learning_rate": 3.596868350506578e-06, "loss": 1.046, "step": 9330 }, { "epoch": 0.714143578754018, "grad_norm": 0.9010024379706463, "learning_rate": 3.596075800846466e-06, "loss": 1.5523, "step": 9331 }, { "epoch": 0.7142201132710853, "grad_norm": 0.5553834181228975, "learning_rate": 3.5952833988906423e-06, "loss": 1.2991, "step": 9332 }, { "epoch": 0.7142966477881525, "grad_norm": 0.566834282541184, "learning_rate": 3.594491144687795e-06, "loss": 1.1996, "step": 9333 }, { "epoch": 0.7143731823052196, "grad_norm": 0.5730548233797312, "learning_rate": 3.5936990382866043e-06, "loss": 1.2976, "step": 9334 }, { "epoch": 0.7144497168222869, "grad_norm": 0.6248027838359571, "learning_rate": 3.592907079735738e-06, "loss": 1.4709, "step": 9335 }, { "epoch": 0.7145262513393541, "grad_norm": 0.6410337267476828, "learning_rate": 3.5921152690838555e-06, "loss": 1.5002, "step": 9336 }, { "epoch": 0.7146027858564212, "grad_norm": 0.6091117267164365, "learning_rate": 3.5913236063796085e-06, "loss": 1.5587, "step": 9337 }, { "epoch": 0.7146793203734885, "grad_norm": 0.6483039395833906, "learning_rate": 3.5905320916716415e-06, "loss": 1.656, "step": 9338 }, { "epoch": 0.7147558548905556, "grad_norm": 1.114843024871846, "learning_rate": 3.5897407250085846e-06, "loss": 1.4394, "step": 9339 }, { "epoch": 0.7148323894076228, "grad_norm": 0.8453464393272213, "learning_rate": 3.588949506439063e-06, "loss": 1.8279, "step": 9340 }, { "epoch": 0.7149089239246901, "grad_norm": 0.5656610739209803, "learning_rate": 3.5881584360116927e-06, "loss": 1.1858, "step": 9341 }, { "epoch": 0.7149854584417572, "grad_norm": 0.5553816323450873, "learning_rate": 3.5873675137750774e-06, "loss": 1.4774, "step": 9342 }, { "epoch": 0.7150619929588244, "grad_norm": 0.5238559215894736, "learning_rate": 3.5865767397778127e-06, "loss": 1.1412, "step": 9343 }, { "epoch": 0.7151385274758916, "grad_norm": 0.6471666574785356, "learning_rate": 3.5857861140684897e-06, "loss": 1.5814, "step": 9344 }, { "epoch": 0.7152150619929588, "grad_norm": 0.7024964277465923, "learning_rate": 3.584995636695685e-06, "loss": 1.6031, "step": 9345 }, { "epoch": 0.715291596510026, "grad_norm": 0.6495866247077193, "learning_rate": 3.5842053077079673e-06, "loss": 1.5155, "step": 9346 }, { "epoch": 0.7153681310270932, "grad_norm": 0.577872671351687, "learning_rate": 3.5834151271538976e-06, "loss": 1.285, "step": 9347 }, { "epoch": 0.7154446655441604, "grad_norm": 0.9486302481922215, "learning_rate": 3.5826250950820274e-06, "loss": 1.6801, "step": 9348 }, { "epoch": 0.7155212000612277, "grad_norm": 0.5673597208463154, "learning_rate": 3.581835211540897e-06, "loss": 1.2474, "step": 9349 }, { "epoch": 0.7155977345782948, "grad_norm": 0.5029368413117616, "learning_rate": 3.581045476579041e-06, "loss": 0.9796, "step": 9350 }, { "epoch": 0.715674269095362, "grad_norm": 0.6367669575925621, "learning_rate": 3.580255890244981e-06, "loss": 1.5435, "step": 9351 }, { "epoch": 0.7157508036124292, "grad_norm": 0.5943485088410104, "learning_rate": 3.5794664525872336e-06, "loss": 1.5028, "step": 9352 }, { "epoch": 0.7158273381294964, "grad_norm": 0.5475884312960139, "learning_rate": 3.5786771636543037e-06, "loss": 1.3212, "step": 9353 }, { "epoch": 0.7159038726465636, "grad_norm": 0.5372635450468257, "learning_rate": 3.5778880234946854e-06, "loss": 1.342, "step": 9354 }, { "epoch": 0.7159804071636308, "grad_norm": 0.6593175171153306, "learning_rate": 3.5770990321568687e-06, "loss": 1.5634, "step": 9355 }, { "epoch": 0.716056941680698, "grad_norm": 0.6489216911311766, "learning_rate": 3.576310189689331e-06, "loss": 1.2341, "step": 9356 }, { "epoch": 0.7161334761977652, "grad_norm": 0.5295107219305043, "learning_rate": 3.5755214961405405e-06, "loss": 1.2944, "step": 9357 }, { "epoch": 0.7162100107148324, "grad_norm": 0.6933779821781996, "learning_rate": 3.574732951558958e-06, "loss": 1.5204, "step": 9358 }, { "epoch": 0.7162865452318996, "grad_norm": 0.6049621735936515, "learning_rate": 3.5739445559930307e-06, "loss": 1.4698, "step": 9359 }, { "epoch": 0.7163630797489667, "grad_norm": 0.6106038513733448, "learning_rate": 3.5731563094912035e-06, "loss": 1.0603, "step": 9360 }, { "epoch": 0.716439614266034, "grad_norm": 0.4662628390962993, "learning_rate": 3.572368212101908e-06, "loss": 1.0374, "step": 9361 }, { "epoch": 0.7165161487831012, "grad_norm": 0.5736041994607162, "learning_rate": 3.5715802638735674e-06, "loss": 1.2935, "step": 9362 }, { "epoch": 0.7165926833001683, "grad_norm": 1.2085477347133133, "learning_rate": 3.5707924648545944e-06, "loss": 1.5289, "step": 9363 }, { "epoch": 0.7166692178172356, "grad_norm": 0.5711507160428925, "learning_rate": 3.570004815093394e-06, "loss": 1.3977, "step": 9364 }, { "epoch": 0.7167457523343028, "grad_norm": 2.042031584334144, "learning_rate": 3.5692173146383613e-06, "loss": 1.4332, "step": 9365 }, { "epoch": 0.71682228685137, "grad_norm": 0.6515339438031319, "learning_rate": 3.5684299635378845e-06, "loss": 1.4951, "step": 9366 }, { "epoch": 0.7168988213684372, "grad_norm": 0.8010726560649658, "learning_rate": 3.5676427618403396e-06, "loss": 1.6091, "step": 9367 }, { "epoch": 0.7169753558855043, "grad_norm": 0.5907092028359527, "learning_rate": 3.5668557095940948e-06, "loss": 1.2894, "step": 9368 }, { "epoch": 0.7170518904025716, "grad_norm": 0.9317928337325098, "learning_rate": 3.566068806847509e-06, "loss": 1.2298, "step": 9369 }, { "epoch": 0.7171284249196388, "grad_norm": 0.6136579334681811, "learning_rate": 3.5652820536489323e-06, "loss": 1.3361, "step": 9370 }, { "epoch": 0.7172049594367059, "grad_norm": 0.5359728505084882, "learning_rate": 3.5644954500467043e-06, "loss": 1.3355, "step": 9371 }, { "epoch": 0.7172814939537732, "grad_norm": 0.5844713078511219, "learning_rate": 3.5637089960891565e-06, "loss": 1.2552, "step": 9372 }, { "epoch": 0.7173580284708404, "grad_norm": 0.6004311227569096, "learning_rate": 3.5629226918246108e-06, "loss": 1.4282, "step": 9373 }, { "epoch": 0.7174345629879075, "grad_norm": 0.7320837691835236, "learning_rate": 3.562136537301381e-06, "loss": 1.4712, "step": 9374 }, { "epoch": 0.7175110975049748, "grad_norm": 0.614228814508482, "learning_rate": 3.5613505325677683e-06, "loss": 1.4102, "step": 9375 }, { "epoch": 0.7175876320220419, "grad_norm": 0.5875547178056278, "learning_rate": 3.560564677672071e-06, "loss": 1.4358, "step": 9376 }, { "epoch": 0.7176641665391091, "grad_norm": 0.5736680485507478, "learning_rate": 3.559778972662572e-06, "loss": 1.3557, "step": 9377 }, { "epoch": 0.7177407010561764, "grad_norm": 0.6013167021691601, "learning_rate": 3.558993417587548e-06, "loss": 1.3989, "step": 9378 }, { "epoch": 0.7178172355732435, "grad_norm": 0.7401953466477609, "learning_rate": 3.5582080124952656e-06, "loss": 1.6307, "step": 9379 }, { "epoch": 0.7178937700903107, "grad_norm": 0.5777266390486365, "learning_rate": 3.5574227574339827e-06, "loss": 1.5345, "step": 9380 }, { "epoch": 0.7179703046073779, "grad_norm": 0.8022580442753356, "learning_rate": 3.556637652451945e-06, "loss": 1.3659, "step": 9381 }, { "epoch": 0.7180468391244451, "grad_norm": 0.5129299903046332, "learning_rate": 3.5558526975973966e-06, "loss": 1.2272, "step": 9382 }, { "epoch": 0.7181233736415124, "grad_norm": 0.5920511695739953, "learning_rate": 3.5550678929185645e-06, "loss": 1.4081, "step": 9383 }, { "epoch": 0.7181999081585795, "grad_norm": 0.5905235685198554, "learning_rate": 3.5542832384636705e-06, "loss": 1.4979, "step": 9384 }, { "epoch": 0.7182764426756467, "grad_norm": 0.6685133429414202, "learning_rate": 3.553498734280926e-06, "loss": 1.6537, "step": 9385 }, { "epoch": 0.718352977192714, "grad_norm": 0.5359935258158118, "learning_rate": 3.5527143804185304e-06, "loss": 1.1633, "step": 9386 }, { "epoch": 0.7184295117097811, "grad_norm": 0.5464999601050342, "learning_rate": 3.5519301769246817e-06, "loss": 1.1427, "step": 9387 }, { "epoch": 0.7185060462268483, "grad_norm": 0.4944159137447779, "learning_rate": 3.551146123847561e-06, "loss": 1.1587, "step": 9388 }, { "epoch": 0.7185825807439155, "grad_norm": 0.6335511228493607, "learning_rate": 3.5503622212353426e-06, "loss": 1.4469, "step": 9389 }, { "epoch": 0.7186591152609827, "grad_norm": 0.7347800646137872, "learning_rate": 3.5495784691361936e-06, "loss": 1.3219, "step": 9390 }, { "epoch": 0.7187356497780499, "grad_norm": 0.5546414701689746, "learning_rate": 3.5487948675982673e-06, "loss": 1.0072, "step": 9391 }, { "epoch": 0.7188121842951171, "grad_norm": 0.7097403028323043, "learning_rate": 3.548011416669712e-06, "loss": 1.4202, "step": 9392 }, { "epoch": 0.7188887188121843, "grad_norm": 0.7633820270530712, "learning_rate": 3.5472281163986667e-06, "loss": 1.4877, "step": 9393 }, { "epoch": 0.7189652533292515, "grad_norm": 0.8435639195149728, "learning_rate": 3.5464449668332575e-06, "loss": 1.4865, "step": 9394 }, { "epoch": 0.7190417878463187, "grad_norm": 0.5596922860690479, "learning_rate": 3.5456619680216043e-06, "loss": 1.3109, "step": 9395 }, { "epoch": 0.7191183223633859, "grad_norm": 0.536731380484106, "learning_rate": 3.544879120011817e-06, "loss": 1.3268, "step": 9396 }, { "epoch": 0.719194856880453, "grad_norm": 0.5532460551003351, "learning_rate": 3.544096422851995e-06, "loss": 1.3686, "step": 9397 }, { "epoch": 0.7192713913975203, "grad_norm": 0.5248389444586502, "learning_rate": 3.5433138765902307e-06, "loss": 1.0166, "step": 9398 }, { "epoch": 0.7193479259145875, "grad_norm": 0.4792414664726994, "learning_rate": 3.542531481274607e-06, "loss": 0.9521, "step": 9399 }, { "epoch": 0.7194244604316546, "grad_norm": 0.5900341797630291, "learning_rate": 3.541749236953195e-06, "loss": 1.445, "step": 9400 }, { "epoch": 0.7195009949487219, "grad_norm": 0.5126383524317132, "learning_rate": 3.540967143674059e-06, "loss": 1.1362, "step": 9401 }, { "epoch": 0.7195775294657891, "grad_norm": 0.5706154147942034, "learning_rate": 3.540185201485253e-06, "loss": 1.4014, "step": 9402 }, { "epoch": 0.7196540639828563, "grad_norm": 0.7176655290662773, "learning_rate": 3.5394034104348195e-06, "loss": 1.4511, "step": 9403 }, { "epoch": 0.7197305984999235, "grad_norm": 0.6061634745872645, "learning_rate": 3.538621770570798e-06, "loss": 1.4314, "step": 9404 }, { "epoch": 0.7198071330169906, "grad_norm": 1.249431448694394, "learning_rate": 3.5378402819412134e-06, "loss": 1.6313, "step": 9405 }, { "epoch": 0.7198836675340579, "grad_norm": 0.5732751149695736, "learning_rate": 3.5370589445940824e-06, "loss": 1.2198, "step": 9406 }, { "epoch": 0.7199602020511251, "grad_norm": 0.5629281030489144, "learning_rate": 3.536277758577412e-06, "loss": 1.2423, "step": 9407 }, { "epoch": 0.7200367365681922, "grad_norm": 0.736360012526064, "learning_rate": 3.5354967239392002e-06, "loss": 1.5258, "step": 9408 }, { "epoch": 0.7201132710852595, "grad_norm": 0.5908526913564398, "learning_rate": 3.5347158407274383e-06, "loss": 1.321, "step": 9409 }, { "epoch": 0.7201898056023267, "grad_norm": 0.5213100810272456, "learning_rate": 3.5339351089901053e-06, "loss": 1.1825, "step": 9410 }, { "epoch": 0.7202663401193938, "grad_norm": 0.5592348119869628, "learning_rate": 3.5331545287751712e-06, "loss": 1.3885, "step": 9411 }, { "epoch": 0.7203428746364611, "grad_norm": 0.6766831939504447, "learning_rate": 3.532374100130597e-06, "loss": 1.6412, "step": 9412 }, { "epoch": 0.7204194091535282, "grad_norm": 0.61454134001658, "learning_rate": 3.531593823104335e-06, "loss": 1.396, "step": 9413 }, { "epoch": 0.7204959436705954, "grad_norm": 0.6121949326936427, "learning_rate": 3.530813697744328e-06, "loss": 1.2717, "step": 9414 }, { "epoch": 0.7205724781876627, "grad_norm": 0.5815063848668941, "learning_rate": 3.5300337240985083e-06, "loss": 1.3536, "step": 9415 }, { "epoch": 0.7206490127047298, "grad_norm": 0.5769952907930622, "learning_rate": 3.529253902214801e-06, "loss": 1.4244, "step": 9416 }, { "epoch": 0.720725547221797, "grad_norm": 0.628012147243779, "learning_rate": 3.5284742321411202e-06, "loss": 1.5418, "step": 9417 }, { "epoch": 0.7208020817388642, "grad_norm": 0.6317591584075165, "learning_rate": 3.5276947139253703e-06, "loss": 1.3324, "step": 9418 }, { "epoch": 0.7208786162559314, "grad_norm": 0.6476891985804313, "learning_rate": 3.5269153476154473e-06, "loss": 1.7989, "step": 9419 }, { "epoch": 0.7209551507729987, "grad_norm": 0.5579447630634611, "learning_rate": 3.526136133259239e-06, "loss": 1.3094, "step": 9420 }, { "epoch": 0.7210316852900658, "grad_norm": 0.7558845327342045, "learning_rate": 3.5253570709046224e-06, "loss": 1.7112, "step": 9421 }, { "epoch": 0.721108219807133, "grad_norm": 0.5147848961702787, "learning_rate": 3.5245781605994656e-06, "loss": 1.119, "step": 9422 }, { "epoch": 0.7211847543242003, "grad_norm": 0.6922360571739498, "learning_rate": 3.5237994023916267e-06, "loss": 1.4647, "step": 9423 }, { "epoch": 0.7212612888412674, "grad_norm": 0.56013092960327, "learning_rate": 3.523020796328954e-06, "loss": 1.1827, "step": 9424 }, { "epoch": 0.7213378233583346, "grad_norm": 0.6535162427420675, "learning_rate": 3.522242342459288e-06, "loss": 1.2244, "step": 9425 }, { "epoch": 0.7214143578754018, "grad_norm": 0.5370667289983555, "learning_rate": 3.52146404083046e-06, "loss": 1.3465, "step": 9426 }, { "epoch": 0.721490892392469, "grad_norm": 0.6630333980940701, "learning_rate": 3.520685891490292e-06, "loss": 1.5648, "step": 9427 }, { "epoch": 0.7215674269095362, "grad_norm": 0.5451284343555685, "learning_rate": 3.5199078944865938e-06, "loss": 1.367, "step": 9428 }, { "epoch": 0.7216439614266034, "grad_norm": 0.5142426473995919, "learning_rate": 3.5191300498671686e-06, "loss": 1.0474, "step": 9429 }, { "epoch": 0.7217204959436706, "grad_norm": 0.6337947090384379, "learning_rate": 3.5183523576798082e-06, "loss": 1.4744, "step": 9430 }, { "epoch": 0.7217970304607378, "grad_norm": 0.6153549822870219, "learning_rate": 3.517574817972299e-06, "loss": 1.6718, "step": 9431 }, { "epoch": 0.721873564977805, "grad_norm": 0.5906591783769163, "learning_rate": 3.5167974307924147e-06, "loss": 1.3617, "step": 9432 }, { "epoch": 0.7219500994948722, "grad_norm": 0.5214315819945231, "learning_rate": 3.516020196187919e-06, "loss": 1.0053, "step": 9433 }, { "epoch": 0.7220266340119393, "grad_norm": 0.7048641368112443, "learning_rate": 3.5152431142065678e-06, "loss": 1.6791, "step": 9434 }, { "epoch": 0.7221031685290066, "grad_norm": 0.5602879961788682, "learning_rate": 3.514466184896108e-06, "loss": 1.2252, "step": 9435 }, { "epoch": 0.7221797030460738, "grad_norm": 0.862252953728888, "learning_rate": 3.513689408304275e-06, "loss": 1.4464, "step": 9436 }, { "epoch": 0.722256237563141, "grad_norm": 0.5374476648487485, "learning_rate": 3.5129127844788003e-06, "loss": 1.2922, "step": 9437 }, { "epoch": 0.7223327720802082, "grad_norm": 0.5130971854627893, "learning_rate": 3.5121363134673967e-06, "loss": 1.0973, "step": 9438 }, { "epoch": 0.7224093065972754, "grad_norm": 0.49765686845361834, "learning_rate": 3.5113599953177756e-06, "loss": 1.1153, "step": 9439 }, { "epoch": 0.7224858411143426, "grad_norm": 0.5439756879740932, "learning_rate": 3.5105838300776372e-06, "loss": 1.2286, "step": 9440 }, { "epoch": 0.7225623756314098, "grad_norm": 0.5555766120952548, "learning_rate": 3.509807817794668e-06, "loss": 1.1358, "step": 9441 }, { "epoch": 0.7226389101484769, "grad_norm": 0.6464466275990529, "learning_rate": 3.5090319585165523e-06, "loss": 1.6086, "step": 9442 }, { "epoch": 0.7227154446655442, "grad_norm": 0.5910848960539798, "learning_rate": 3.5082562522909593e-06, "loss": 1.3209, "step": 9443 }, { "epoch": 0.7227919791826114, "grad_norm": 0.49380073011113834, "learning_rate": 3.5074806991655507e-06, "loss": 0.9618, "step": 9444 }, { "epoch": 0.7228685136996785, "grad_norm": 0.558518504534099, "learning_rate": 3.50670529918798e-06, "loss": 1.2134, "step": 9445 }, { "epoch": 0.7229450482167458, "grad_norm": 0.5812192198035229, "learning_rate": 3.5059300524058892e-06, "loss": 1.3095, "step": 9446 }, { "epoch": 0.7230215827338129, "grad_norm": 0.5467617739860844, "learning_rate": 3.5051549588669096e-06, "loss": 1.231, "step": 9447 }, { "epoch": 0.7230981172508801, "grad_norm": 0.5718092187619803, "learning_rate": 3.5043800186186695e-06, "loss": 1.3135, "step": 9448 }, { "epoch": 0.7231746517679474, "grad_norm": 0.5836495473749908, "learning_rate": 3.503605231708782e-06, "loss": 1.2552, "step": 9449 }, { "epoch": 0.7232511862850145, "grad_norm": 0.6944175319887042, "learning_rate": 3.502830598184851e-06, "loss": 1.8443, "step": 9450 }, { "epoch": 0.7233277208020817, "grad_norm": 0.5896489162999419, "learning_rate": 3.5020561180944736e-06, "loss": 1.4437, "step": 9451 }, { "epoch": 0.723404255319149, "grad_norm": 0.6079314452459569, "learning_rate": 3.5012817914852346e-06, "loss": 1.5175, "step": 9452 }, { "epoch": 0.7234807898362161, "grad_norm": 0.6803933765320043, "learning_rate": 3.500507618404713e-06, "loss": 1.7319, "step": 9453 }, { "epoch": 0.7235573243532833, "grad_norm": 0.518784844417935, "learning_rate": 3.4997335989004755e-06, "loss": 1.0629, "step": 9454 }, { "epoch": 0.7236338588703505, "grad_norm": 0.6344332366037968, "learning_rate": 3.4989597330200796e-06, "loss": 1.5355, "step": 9455 }, { "epoch": 0.7237103933874177, "grad_norm": 0.7741017442914265, "learning_rate": 3.498186020811075e-06, "loss": 1.5361, "step": 9456 }, { "epoch": 0.723786927904485, "grad_norm": 0.4435692386986925, "learning_rate": 3.4974124623210004e-06, "loss": 0.9254, "step": 9457 }, { "epoch": 0.7238634624215521, "grad_norm": 0.5869831936928047, "learning_rate": 3.496639057597384e-06, "loss": 1.2949, "step": 9458 }, { "epoch": 0.7239399969386193, "grad_norm": 0.5690528689397782, "learning_rate": 3.495865806687748e-06, "loss": 1.3166, "step": 9459 }, { "epoch": 0.7240165314556866, "grad_norm": 0.599476296001274, "learning_rate": 3.495092709639604e-06, "loss": 1.2985, "step": 9460 }, { "epoch": 0.7240930659727537, "grad_norm": 1.0254142113123312, "learning_rate": 3.4943197665004513e-06, "loss": 1.3364, "step": 9461 }, { "epoch": 0.7241696004898209, "grad_norm": 0.6081196689833167, "learning_rate": 3.493546977317783e-06, "loss": 1.32, "step": 9462 }, { "epoch": 0.7242461350068881, "grad_norm": 0.5328290585313389, "learning_rate": 3.4927743421390813e-06, "loss": 1.1811, "step": 9463 }, { "epoch": 0.7243226695239553, "grad_norm": 0.5600700738006348, "learning_rate": 3.4920018610118196e-06, "loss": 1.2505, "step": 9464 }, { "epoch": 0.7243992040410225, "grad_norm": 0.7226234032051169, "learning_rate": 3.491229533983461e-06, "loss": 1.2782, "step": 9465 }, { "epoch": 0.7244757385580897, "grad_norm": 0.6513786524095878, "learning_rate": 3.490457361101459e-06, "loss": 1.615, "step": 9466 }, { "epoch": 0.7245522730751569, "grad_norm": 0.5179618101629945, "learning_rate": 3.489685342413259e-06, "loss": 1.1199, "step": 9467 }, { "epoch": 0.7246288075922241, "grad_norm": 0.636075994552531, "learning_rate": 3.488913477966296e-06, "loss": 1.4858, "step": 9468 }, { "epoch": 0.7247053421092913, "grad_norm": 0.6303481531316611, "learning_rate": 3.488141767807994e-06, "loss": 1.4429, "step": 9469 }, { "epoch": 0.7247818766263585, "grad_norm": 0.6122219959947525, "learning_rate": 3.4873702119857734e-06, "loss": 1.5146, "step": 9470 }, { "epoch": 0.7248584111434256, "grad_norm": 0.5532232357854555, "learning_rate": 3.486598810547037e-06, "loss": 1.1831, "step": 9471 }, { "epoch": 0.7249349456604929, "grad_norm": 0.49642630578562646, "learning_rate": 3.4858275635391838e-06, "loss": 1.101, "step": 9472 }, { "epoch": 0.7250114801775601, "grad_norm": 0.7339174531225526, "learning_rate": 3.485056471009601e-06, "loss": 1.5585, "step": 9473 }, { "epoch": 0.7250880146946272, "grad_norm": 1.8210329331511421, "learning_rate": 3.4842855330056656e-06, "loss": 1.9443, "step": 9474 }, { "epoch": 0.7251645492116945, "grad_norm": 0.62582731727602, "learning_rate": 3.483514749574749e-06, "loss": 1.5765, "step": 9475 }, { "epoch": 0.7252410837287617, "grad_norm": 0.57748409493134, "learning_rate": 3.482744120764209e-06, "loss": 1.482, "step": 9476 }, { "epoch": 0.7253176182458289, "grad_norm": 0.6071256294347402, "learning_rate": 3.481973646621395e-06, "loss": 1.3895, "step": 9477 }, { "epoch": 0.7253941527628961, "grad_norm": 0.5537896219294942, "learning_rate": 3.4812033271936484e-06, "loss": 1.1027, "step": 9478 }, { "epoch": 0.7254706872799632, "grad_norm": 0.6613175010273312, "learning_rate": 3.4804331625282982e-06, "loss": 1.6011, "step": 9479 }, { "epoch": 0.7255472217970305, "grad_norm": 0.6005564015184214, "learning_rate": 3.4796631526726666e-06, "loss": 1.2721, "step": 9480 }, { "epoch": 0.7256237563140977, "grad_norm": 0.5634028472105526, "learning_rate": 3.4788932976740654e-06, "loss": 1.3802, "step": 9481 }, { "epoch": 0.7257002908311648, "grad_norm": 0.7120006229014301, "learning_rate": 3.4781235975797967e-06, "loss": 1.7498, "step": 9482 }, { "epoch": 0.7257768253482321, "grad_norm": 0.8835194198067768, "learning_rate": 3.4773540524371534e-06, "loss": 1.4767, "step": 9483 }, { "epoch": 0.7258533598652992, "grad_norm": 0.6161885646280909, "learning_rate": 3.4765846622934185e-06, "loss": 1.6223, "step": 9484 }, { "epoch": 0.7259298943823664, "grad_norm": 0.531859150208693, "learning_rate": 3.4758154271958653e-06, "loss": 1.0906, "step": 9485 }, { "epoch": 0.7260064288994337, "grad_norm": 0.8953229694170574, "learning_rate": 3.4750463471917584e-06, "loss": 1.4441, "step": 9486 }, { "epoch": 0.7260829634165008, "grad_norm": 0.6101160515015935, "learning_rate": 3.4742774223283527e-06, "loss": 1.4561, "step": 9487 }, { "epoch": 0.726159497933568, "grad_norm": 0.5626199592240309, "learning_rate": 3.4735086526528916e-06, "loss": 1.2639, "step": 9488 }, { "epoch": 0.7262360324506353, "grad_norm": 0.5138241776582158, "learning_rate": 3.472740038212612e-06, "loss": 1.0078, "step": 9489 }, { "epoch": 0.7263125669677024, "grad_norm": 0.5901975931274672, "learning_rate": 3.4719715790547394e-06, "loss": 1.3646, "step": 9490 }, { "epoch": 0.7263891014847697, "grad_norm": 0.504578605793934, "learning_rate": 3.471203275226489e-06, "loss": 1.0896, "step": 9491 }, { "epoch": 0.7264656360018368, "grad_norm": 0.7575204729031143, "learning_rate": 3.4704351267750703e-06, "loss": 1.2958, "step": 9492 }, { "epoch": 0.726542170518904, "grad_norm": 0.572962135164322, "learning_rate": 3.4696671337476795e-06, "loss": 1.3305, "step": 9493 }, { "epoch": 0.7266187050359713, "grad_norm": 0.6443387148746826, "learning_rate": 3.4688992961915048e-06, "loss": 1.5722, "step": 9494 }, { "epoch": 0.7266952395530384, "grad_norm": 0.7225495083057589, "learning_rate": 3.4681316141537238e-06, "loss": 1.4666, "step": 9495 }, { "epoch": 0.7267717740701056, "grad_norm": 0.5481660957680522, "learning_rate": 3.4673640876815027e-06, "loss": 1.3162, "step": 9496 }, { "epoch": 0.7268483085871729, "grad_norm": 0.7723737702922556, "learning_rate": 3.4665967168220056e-06, "loss": 1.3116, "step": 9497 }, { "epoch": 0.72692484310424, "grad_norm": 0.5138082705632391, "learning_rate": 3.465829501622379e-06, "loss": 1.0999, "step": 9498 }, { "epoch": 0.7270013776213072, "grad_norm": 0.61178588613488, "learning_rate": 3.465062442129764e-06, "loss": 1.4729, "step": 9499 }, { "epoch": 0.7270779121383744, "grad_norm": 0.5206262179922398, "learning_rate": 3.464295538391291e-06, "loss": 1.2501, "step": 9500 }, { "epoch": 0.7271544466554416, "grad_norm": 0.5852486756228485, "learning_rate": 3.46352879045408e-06, "loss": 1.3878, "step": 9501 }, { "epoch": 0.7272309811725088, "grad_norm": 0.5831586062149507, "learning_rate": 3.4627621983652415e-06, "loss": 1.3329, "step": 9502 }, { "epoch": 0.727307515689576, "grad_norm": 0.5632939920906092, "learning_rate": 3.4619957621718792e-06, "loss": 1.1734, "step": 9503 }, { "epoch": 0.7273840502066432, "grad_norm": 0.6348596903145385, "learning_rate": 3.4612294819210845e-06, "loss": 1.5505, "step": 9504 }, { "epoch": 0.7274605847237104, "grad_norm": 0.6134051885814807, "learning_rate": 3.4604633576599413e-06, "loss": 1.4011, "step": 9505 }, { "epoch": 0.7275371192407776, "grad_norm": 0.5638562126591788, "learning_rate": 3.45969738943552e-06, "loss": 1.3143, "step": 9506 }, { "epoch": 0.7276136537578448, "grad_norm": 0.5800643166710805, "learning_rate": 3.458931577294885e-06, "loss": 1.377, "step": 9507 }, { "epoch": 0.7276901882749119, "grad_norm": 0.663549396487392, "learning_rate": 3.4581659212850902e-06, "loss": 1.472, "step": 9508 }, { "epoch": 0.7277667227919792, "grad_norm": 0.7466865559676554, "learning_rate": 3.4574004214531807e-06, "loss": 1.3994, "step": 9509 }, { "epoch": 0.7278432573090464, "grad_norm": 0.5930819104659925, "learning_rate": 3.45663507784619e-06, "loss": 1.421, "step": 9510 }, { "epoch": 0.7279197918261135, "grad_norm": 0.6076657046187345, "learning_rate": 3.4558698905111425e-06, "loss": 1.4717, "step": 9511 }, { "epoch": 0.7279963263431808, "grad_norm": 0.6509594494491888, "learning_rate": 3.4551048594950553e-06, "loss": 1.4995, "step": 9512 }, { "epoch": 0.7280728608602479, "grad_norm": 0.575284426295873, "learning_rate": 3.4543399848449318e-06, "loss": 1.3577, "step": 9513 }, { "epoch": 0.7281493953773152, "grad_norm": 0.6123312771987521, "learning_rate": 3.453575266607771e-06, "loss": 1.4094, "step": 9514 }, { "epoch": 0.7282259298943824, "grad_norm": 0.5552844747241015, "learning_rate": 3.4528107048305582e-06, "loss": 1.3388, "step": 9515 }, { "epoch": 0.7283024644114495, "grad_norm": 0.5367614114606373, "learning_rate": 3.45204629956027e-06, "loss": 1.2054, "step": 9516 }, { "epoch": 0.7283789989285168, "grad_norm": 0.5857937698795981, "learning_rate": 3.4512820508438737e-06, "loss": 1.377, "step": 9517 }, { "epoch": 0.728455533445584, "grad_norm": 0.6119916672256653, "learning_rate": 3.450517958728327e-06, "loss": 1.4954, "step": 9518 }, { "epoch": 0.7285320679626511, "grad_norm": 0.7915103575702812, "learning_rate": 3.4497540232605797e-06, "loss": 1.4397, "step": 9519 }, { "epoch": 0.7286086024797184, "grad_norm": 0.6341756088292672, "learning_rate": 3.448990244487569e-06, "loss": 1.5235, "step": 9520 }, { "epoch": 0.7286851369967855, "grad_norm": 1.2044682669916063, "learning_rate": 3.4482266224562235e-06, "loss": 1.6527, "step": 9521 }, { "epoch": 0.7287616715138527, "grad_norm": 0.5485952555708474, "learning_rate": 3.447463157213462e-06, "loss": 1.2169, "step": 9522 }, { "epoch": 0.72883820603092, "grad_norm": 0.6733115530083982, "learning_rate": 3.4466998488061944e-06, "loss": 1.4939, "step": 9523 }, { "epoch": 0.7289147405479871, "grad_norm": 0.6644685924463143, "learning_rate": 3.4459366972813224e-06, "loss": 1.3782, "step": 9524 }, { "epoch": 0.7289912750650543, "grad_norm": 0.6688196479128544, "learning_rate": 3.4451737026857348e-06, "loss": 1.2738, "step": 9525 }, { "epoch": 0.7290678095821216, "grad_norm": 0.7676560313362712, "learning_rate": 3.4444108650663116e-06, "loss": 1.526, "step": 9526 }, { "epoch": 0.7291443440991887, "grad_norm": 0.546164158826789, "learning_rate": 3.4436481844699252e-06, "loss": 1.2786, "step": 9527 }, { "epoch": 0.729220878616256, "grad_norm": 0.5993114152530433, "learning_rate": 3.4428856609434356e-06, "loss": 1.454, "step": 9528 }, { "epoch": 0.7292974131333231, "grad_norm": 0.7644050439166014, "learning_rate": 3.442123294533697e-06, "loss": 1.8478, "step": 9529 }, { "epoch": 0.7293739476503903, "grad_norm": 0.5316248144963444, "learning_rate": 3.4413610852875486e-06, "loss": 1.1483, "step": 9530 }, { "epoch": 0.7294504821674576, "grad_norm": 0.6056131363835653, "learning_rate": 3.4405990332518245e-06, "loss": 1.3776, "step": 9531 }, { "epoch": 0.7295270166845247, "grad_norm": 0.5262737303858043, "learning_rate": 3.4398371384733465e-06, "loss": 1.1489, "step": 9532 }, { "epoch": 0.7296035512015919, "grad_norm": 0.5429513872430943, "learning_rate": 3.439075400998928e-06, "loss": 1.3336, "step": 9533 }, { "epoch": 0.7296800857186592, "grad_norm": 0.6349369750501055, "learning_rate": 3.438313820875371e-06, "loss": 1.6668, "step": 9534 }, { "epoch": 0.7297566202357263, "grad_norm": 0.6509420221415004, "learning_rate": 3.4375523981494728e-06, "loss": 1.4773, "step": 9535 }, { "epoch": 0.7298331547527935, "grad_norm": 0.5808377524706546, "learning_rate": 3.436791132868016e-06, "loss": 1.3292, "step": 9536 }, { "epoch": 0.7299096892698607, "grad_norm": 0.5199540093154812, "learning_rate": 3.436030025077774e-06, "loss": 1.1473, "step": 9537 }, { "epoch": 0.7299862237869279, "grad_norm": 0.5473006053593494, "learning_rate": 3.4352690748255113e-06, "loss": 1.1221, "step": 9538 }, { "epoch": 0.7300627583039951, "grad_norm": 0.6734788372576708, "learning_rate": 3.4345082821579844e-06, "loss": 1.1843, "step": 9539 }, { "epoch": 0.7301392928210623, "grad_norm": 0.6797158610230158, "learning_rate": 3.4337476471219376e-06, "loss": 1.7038, "step": 9540 }, { "epoch": 0.7302158273381295, "grad_norm": 0.5765536095775863, "learning_rate": 3.432987169764107e-06, "loss": 1.2911, "step": 9541 }, { "epoch": 0.7302923618551967, "grad_norm": 0.6210811214731724, "learning_rate": 3.432226850131219e-06, "loss": 1.2519, "step": 9542 }, { "epoch": 0.7303688963722639, "grad_norm": 0.7126521796183384, "learning_rate": 3.43146668826999e-06, "loss": 1.9119, "step": 9543 }, { "epoch": 0.7304454308893311, "grad_norm": 0.5670898681083759, "learning_rate": 3.4307066842271252e-06, "loss": 1.3666, "step": 9544 }, { "epoch": 0.7305219654063982, "grad_norm": 0.6031284261688907, "learning_rate": 3.4299468380493217e-06, "loss": 1.4367, "step": 9545 }, { "epoch": 0.7305984999234655, "grad_norm": 0.4752667382761016, "learning_rate": 3.4291871497832683e-06, "loss": 0.9517, "step": 9546 }, { "epoch": 0.7306750344405327, "grad_norm": 0.7658757842075775, "learning_rate": 3.428427619475643e-06, "loss": 1.3939, "step": 9547 }, { "epoch": 0.7307515689575999, "grad_norm": 1.7713839536679492, "learning_rate": 3.4276682471731106e-06, "loss": 1.4686, "step": 9548 }, { "epoch": 0.7308281034746671, "grad_norm": 0.5674386275824304, "learning_rate": 3.426909032922332e-06, "loss": 1.3584, "step": 9549 }, { "epoch": 0.7309046379917342, "grad_norm": 1.4760297715606427, "learning_rate": 3.426149976769954e-06, "loss": 1.5451, "step": 9550 }, { "epoch": 0.7309811725088015, "grad_norm": 0.7557444392843464, "learning_rate": 3.425391078762616e-06, "loss": 1.5659, "step": 9551 }, { "epoch": 0.7310577070258687, "grad_norm": 0.47527416734486405, "learning_rate": 3.4246323389469464e-06, "loss": 1.0667, "step": 9552 }, { "epoch": 0.7311342415429358, "grad_norm": 0.7234807441263045, "learning_rate": 3.423873757369566e-06, "loss": 1.5498, "step": 9553 }, { "epoch": 0.7312107760600031, "grad_norm": 0.6436817455127747, "learning_rate": 3.423115334077082e-06, "loss": 1.5258, "step": 9554 }, { "epoch": 0.7312873105770703, "grad_norm": 0.6012527865437585, "learning_rate": 3.4223570691160944e-06, "loss": 1.4027, "step": 9555 }, { "epoch": 0.7313638450941374, "grad_norm": 0.560079347222745, "learning_rate": 3.421598962533194e-06, "loss": 1.0976, "step": 9556 }, { "epoch": 0.7314403796112047, "grad_norm": 0.86543861610411, "learning_rate": 3.4208410143749627e-06, "loss": 1.4408, "step": 9557 }, { "epoch": 0.7315169141282718, "grad_norm": 0.8657393751318023, "learning_rate": 3.42008322468797e-06, "loss": 1.5213, "step": 9558 }, { "epoch": 0.731593448645339, "grad_norm": 0.5930267811367237, "learning_rate": 3.419325593518775e-06, "loss": 1.5079, "step": 9559 }, { "epoch": 0.7316699831624063, "grad_norm": 0.7696276342463498, "learning_rate": 3.4185681209139313e-06, "loss": 1.8412, "step": 9560 }, { "epoch": 0.7317465176794734, "grad_norm": 0.5993333200124012, "learning_rate": 3.41781080691998e-06, "loss": 1.4058, "step": 9561 }, { "epoch": 0.7318230521965406, "grad_norm": 0.5310608535290349, "learning_rate": 3.4170536515834494e-06, "loss": 1.1213, "step": 9562 }, { "epoch": 0.7318995867136079, "grad_norm": 0.620779258445796, "learning_rate": 3.416296654950866e-06, "loss": 1.4334, "step": 9563 }, { "epoch": 0.731976121230675, "grad_norm": 0.5690245352549986, "learning_rate": 3.4155398170687393e-06, "loss": 1.2833, "step": 9564 }, { "epoch": 0.7320526557477423, "grad_norm": 0.6232634916102352, "learning_rate": 3.4147831379835726e-06, "loss": 1.4754, "step": 9565 }, { "epoch": 0.7321291902648094, "grad_norm": 0.6140368239322889, "learning_rate": 3.4140266177418587e-06, "loss": 1.3793, "step": 9566 }, { "epoch": 0.7322057247818766, "grad_norm": 0.6086819449820295, "learning_rate": 3.4132702563900787e-06, "loss": 1.321, "step": 9567 }, { "epoch": 0.7322822592989439, "grad_norm": 0.5915872911360184, "learning_rate": 3.412514053974707e-06, "loss": 1.4855, "step": 9568 }, { "epoch": 0.732358793816011, "grad_norm": 0.5932725794003371, "learning_rate": 3.4117580105422087e-06, "loss": 1.4208, "step": 9569 }, { "epoch": 0.7324353283330782, "grad_norm": 0.5664965589246732, "learning_rate": 3.411002126139035e-06, "loss": 1.3469, "step": 9570 }, { "epoch": 0.7325118628501455, "grad_norm": 0.6729353987904789, "learning_rate": 3.4102464008116292e-06, "loss": 1.6482, "step": 9571 }, { "epoch": 0.7325883973672126, "grad_norm": 0.5494246850549831, "learning_rate": 3.4094908346064274e-06, "loss": 1.3085, "step": 9572 }, { "epoch": 0.7326649318842798, "grad_norm": 0.6575302784738124, "learning_rate": 3.4087354275698532e-06, "loss": 1.6887, "step": 9573 }, { "epoch": 0.732741466401347, "grad_norm": 0.6186178651454087, "learning_rate": 3.40798017974832e-06, "loss": 1.0184, "step": 9574 }, { "epoch": 0.7328180009184142, "grad_norm": 1.0672958902179646, "learning_rate": 3.407225091188233e-06, "loss": 1.4282, "step": 9575 }, { "epoch": 0.7328945354354814, "grad_norm": 0.5944218123253944, "learning_rate": 3.4064701619359888e-06, "loss": 1.4836, "step": 9576 }, { "epoch": 0.7329710699525486, "grad_norm": 0.5427366532017203, "learning_rate": 3.4057153920379693e-06, "loss": 1.2864, "step": 9577 }, { "epoch": 0.7330476044696158, "grad_norm": 1.0133590574073537, "learning_rate": 3.404960781540551e-06, "loss": 1.54, "step": 9578 }, { "epoch": 0.733124138986683, "grad_norm": 0.5902717139689284, "learning_rate": 3.404206330490102e-06, "loss": 1.4176, "step": 9579 }, { "epoch": 0.7332006735037502, "grad_norm": 0.5840502443789289, "learning_rate": 3.403452038932975e-06, "loss": 1.3844, "step": 9580 }, { "epoch": 0.7332772080208174, "grad_norm": 0.49150035928685587, "learning_rate": 3.402697906915517e-06, "loss": 1.104, "step": 9581 }, { "epoch": 0.7333537425378845, "grad_norm": 0.8258624665590064, "learning_rate": 3.4019439344840644e-06, "loss": 1.5746, "step": 9582 }, { "epoch": 0.7334302770549518, "grad_norm": 0.5809127280848918, "learning_rate": 3.401190121684943e-06, "loss": 1.3215, "step": 9583 }, { "epoch": 0.733506811572019, "grad_norm": 0.9785248917413251, "learning_rate": 3.4004364685644685e-06, "loss": 1.3461, "step": 9584 }, { "epoch": 0.7335833460890862, "grad_norm": 1.5886574402043494, "learning_rate": 3.399682975168949e-06, "loss": 1.5924, "step": 9585 }, { "epoch": 0.7336598806061534, "grad_norm": 0.6204312259499962, "learning_rate": 3.3989296415446816e-06, "loss": 1.5383, "step": 9586 }, { "epoch": 0.7337364151232205, "grad_norm": 0.5497558739129818, "learning_rate": 3.3981764677379527e-06, "loss": 1.3118, "step": 9587 }, { "epoch": 0.7338129496402878, "grad_norm": 0.6388080733140223, "learning_rate": 3.39742345379504e-06, "loss": 1.5755, "step": 9588 }, { "epoch": 0.733889484157355, "grad_norm": 0.5672323707064072, "learning_rate": 3.3966705997622083e-06, "loss": 1.1513, "step": 9589 }, { "epoch": 0.7339660186744221, "grad_norm": 0.5731970093259715, "learning_rate": 3.3959179056857193e-06, "loss": 1.34, "step": 9590 }, { "epoch": 0.7340425531914894, "grad_norm": 0.5891896301184211, "learning_rate": 3.3951653716118195e-06, "loss": 1.5491, "step": 9591 }, { "epoch": 0.7341190877085566, "grad_norm": 0.5403398387738805, "learning_rate": 3.3944129975867457e-06, "loss": 1.2297, "step": 9592 }, { "epoch": 0.7341956222256237, "grad_norm": 0.497676169009842, "learning_rate": 3.3936607836567258e-06, "loss": 1.1178, "step": 9593 }, { "epoch": 0.734272156742691, "grad_norm": 0.5678203659265124, "learning_rate": 3.3929087298679796e-06, "loss": 1.4824, "step": 9594 }, { "epoch": 0.7343486912597581, "grad_norm": 0.6558184020341196, "learning_rate": 3.3921568362667146e-06, "loss": 1.583, "step": 9595 }, { "epoch": 0.7344252257768253, "grad_norm": 0.6283022795002463, "learning_rate": 3.3914051028991296e-06, "loss": 1.5673, "step": 9596 }, { "epoch": 0.7345017602938926, "grad_norm": 0.5923324265640604, "learning_rate": 3.3906535298114134e-06, "loss": 1.4542, "step": 9597 }, { "epoch": 0.7345782948109597, "grad_norm": 0.7676580129684236, "learning_rate": 3.389902117049745e-06, "loss": 1.6025, "step": 9598 }, { "epoch": 0.734654829328027, "grad_norm": 0.584615106282105, "learning_rate": 3.389150864660293e-06, "loss": 1.3543, "step": 9599 }, { "epoch": 0.7347313638450942, "grad_norm": 0.5379729918907307, "learning_rate": 3.3883997726892147e-06, "loss": 1.2464, "step": 9600 }, { "epoch": 0.7348078983621613, "grad_norm": 0.6503207738225772, "learning_rate": 3.387648841182664e-06, "loss": 1.4189, "step": 9601 }, { "epoch": 0.7348844328792286, "grad_norm": 0.7080911905990575, "learning_rate": 3.3868980701867772e-06, "loss": 1.3406, "step": 9602 }, { "epoch": 0.7349609673962957, "grad_norm": 0.4775268665851697, "learning_rate": 3.3861474597476853e-06, "loss": 0.9789, "step": 9603 }, { "epoch": 0.7350375019133629, "grad_norm": 0.52678089535283, "learning_rate": 3.3853970099115073e-06, "loss": 1.255, "step": 9604 }, { "epoch": 0.7351140364304302, "grad_norm": 0.642977718110194, "learning_rate": 3.3846467207243527e-06, "loss": 1.6045, "step": 9605 }, { "epoch": 0.7351905709474973, "grad_norm": 0.6600313436925235, "learning_rate": 3.3838965922323206e-06, "loss": 1.4613, "step": 9606 }, { "epoch": 0.7352671054645645, "grad_norm": 0.6440831414810613, "learning_rate": 3.3831466244815048e-06, "loss": 1.5372, "step": 9607 }, { "epoch": 0.7353436399816318, "grad_norm": 0.6636064366702061, "learning_rate": 3.382396817517982e-06, "loss": 1.5009, "step": 9608 }, { "epoch": 0.7354201744986989, "grad_norm": 0.5419099807499939, "learning_rate": 3.381647171387825e-06, "loss": 1.3393, "step": 9609 }, { "epoch": 0.7354967090157661, "grad_norm": 0.6682239469757735, "learning_rate": 3.3808976861370923e-06, "loss": 1.6377, "step": 9610 }, { "epoch": 0.7355732435328333, "grad_norm": 0.6913970143100279, "learning_rate": 3.3801483618118345e-06, "loss": 1.4162, "step": 9611 }, { "epoch": 0.7356497780499005, "grad_norm": 0.5618962556154716, "learning_rate": 3.3793991984580954e-06, "loss": 1.2929, "step": 9612 }, { "epoch": 0.7357263125669677, "grad_norm": 0.6096908231094724, "learning_rate": 3.378650196121902e-06, "loss": 1.3301, "step": 9613 }, { "epoch": 0.7358028470840349, "grad_norm": 0.581793534263774, "learning_rate": 3.377901354849278e-06, "loss": 1.4978, "step": 9614 }, { "epoch": 0.7358793816011021, "grad_norm": 0.7019036720589734, "learning_rate": 3.3771526746862336e-06, "loss": 1.4757, "step": 9615 }, { "epoch": 0.7359559161181692, "grad_norm": 0.5351844975349637, "learning_rate": 3.3764041556787685e-06, "loss": 1.2491, "step": 9616 }, { "epoch": 0.7360324506352365, "grad_norm": 0.5417440136457148, "learning_rate": 3.375655797872876e-06, "loss": 1.1378, "step": 9617 }, { "epoch": 0.7361089851523037, "grad_norm": 0.8421464088367109, "learning_rate": 3.3749076013145364e-06, "loss": 1.7267, "step": 9618 }, { "epoch": 0.7361855196693708, "grad_norm": 0.5392313471401072, "learning_rate": 3.374159566049721e-06, "loss": 1.149, "step": 9619 }, { "epoch": 0.7362620541864381, "grad_norm": 1.0125212428900146, "learning_rate": 3.373411692124392e-06, "loss": 1.8829, "step": 9620 }, { "epoch": 0.7363385887035053, "grad_norm": 0.5134251832181447, "learning_rate": 3.372663979584501e-06, "loss": 1.0579, "step": 9621 }, { "epoch": 0.7364151232205725, "grad_norm": 0.5609483356044945, "learning_rate": 3.3719164284759874e-06, "loss": 1.2547, "step": 9622 }, { "epoch": 0.7364916577376397, "grad_norm": 1.348629569929068, "learning_rate": 3.371169038844786e-06, "loss": 1.7771, "step": 9623 }, { "epoch": 0.7365681922547068, "grad_norm": 0.49810711038846267, "learning_rate": 3.370421810736818e-06, "loss": 1.0823, "step": 9624 }, { "epoch": 0.7366447267717741, "grad_norm": 0.6516108514511235, "learning_rate": 3.3696747441979948e-06, "loss": 1.7611, "step": 9625 }, { "epoch": 0.7367212612888413, "grad_norm": 0.542672682943723, "learning_rate": 3.368927839274219e-06, "loss": 1.1592, "step": 9626 }, { "epoch": 0.7367977958059084, "grad_norm": 0.5529004721445319, "learning_rate": 3.3681810960113814e-06, "loss": 1.2389, "step": 9627 }, { "epoch": 0.7368743303229757, "grad_norm": 0.6240519405478088, "learning_rate": 3.3674345144553634e-06, "loss": 1.427, "step": 9628 }, { "epoch": 0.7369508648400429, "grad_norm": 0.6731436590011225, "learning_rate": 3.366688094652041e-06, "loss": 1.5492, "step": 9629 }, { "epoch": 0.73702739935711, "grad_norm": 0.7661787700322691, "learning_rate": 3.365941836647274e-06, "loss": 1.6024, "step": 9630 }, { "epoch": 0.7371039338741773, "grad_norm": 0.5872380176104725, "learning_rate": 3.365195740486914e-06, "loss": 1.2491, "step": 9631 }, { "epoch": 0.7371804683912444, "grad_norm": 0.7877378600775885, "learning_rate": 3.3644498062168052e-06, "loss": 1.6478, "step": 9632 }, { "epoch": 0.7372570029083116, "grad_norm": 0.6151528824395788, "learning_rate": 3.363704033882777e-06, "loss": 1.4835, "step": 9633 }, { "epoch": 0.7373335374253789, "grad_norm": 0.6399408305865428, "learning_rate": 3.3629584235306554e-06, "loss": 1.5081, "step": 9634 }, { "epoch": 0.737410071942446, "grad_norm": 0.5860307465358169, "learning_rate": 3.362212975206253e-06, "loss": 1.5321, "step": 9635 }, { "epoch": 0.7374866064595132, "grad_norm": 0.5797337894581877, "learning_rate": 3.361467688955369e-06, "loss": 1.4199, "step": 9636 }, { "epoch": 0.7375631409765805, "grad_norm": 0.5364467153226252, "learning_rate": 3.3607225648237984e-06, "loss": 1.3569, "step": 9637 }, { "epoch": 0.7376396754936476, "grad_norm": 0.7450236546087867, "learning_rate": 3.359977602857324e-06, "loss": 1.784, "step": 9638 }, { "epoch": 0.7377162100107149, "grad_norm": 0.5325271352930986, "learning_rate": 3.359232803101717e-06, "loss": 1.1192, "step": 9639 }, { "epoch": 0.737792744527782, "grad_norm": 0.6676642865156537, "learning_rate": 3.3584881656027413e-06, "loss": 1.3503, "step": 9640 }, { "epoch": 0.7378692790448492, "grad_norm": 0.6619982572266648, "learning_rate": 3.3577436904061493e-06, "loss": 1.4543, "step": 9641 }, { "epoch": 0.7379458135619165, "grad_norm": 0.5353708676489237, "learning_rate": 3.3569993775576835e-06, "loss": 1.0662, "step": 9642 }, { "epoch": 0.7380223480789836, "grad_norm": 0.6523308505400395, "learning_rate": 3.356255227103078e-06, "loss": 1.5454, "step": 9643 }, { "epoch": 0.7380988825960508, "grad_norm": 0.5615061321097112, "learning_rate": 3.355511239088053e-06, "loss": 1.2681, "step": 9644 }, { "epoch": 0.7381754171131181, "grad_norm": 0.5575457386437325, "learning_rate": 3.3547674135583238e-06, "loss": 1.3, "step": 9645 }, { "epoch": 0.7382519516301852, "grad_norm": 0.833836214162437, "learning_rate": 3.354023750559593e-06, "loss": 1.4737, "step": 9646 }, { "epoch": 0.7383284861472524, "grad_norm": 0.6691152476346657, "learning_rate": 3.3532802501375517e-06, "loss": 1.474, "step": 9647 }, { "epoch": 0.7384050206643196, "grad_norm": 0.4935570936754594, "learning_rate": 3.352536912337885e-06, "loss": 1.0641, "step": 9648 }, { "epoch": 0.7384815551813868, "grad_norm": 0.8904361948661369, "learning_rate": 3.3517937372062647e-06, "loss": 1.3237, "step": 9649 }, { "epoch": 0.738558089698454, "grad_norm": 0.5981955347316387, "learning_rate": 3.351050724788353e-06, "loss": 1.389, "step": 9650 }, { "epoch": 0.7386346242155212, "grad_norm": 0.5531679802503656, "learning_rate": 3.3503078751298046e-06, "loss": 1.2937, "step": 9651 }, { "epoch": 0.7387111587325884, "grad_norm": 0.5968568217435599, "learning_rate": 3.3495651882762615e-06, "loss": 1.3697, "step": 9652 }, { "epoch": 0.7387876932496555, "grad_norm": 0.5204536103565582, "learning_rate": 3.348822664273356e-06, "loss": 1.1087, "step": 9653 }, { "epoch": 0.7388642277667228, "grad_norm": 0.635237780299962, "learning_rate": 3.3480803031667117e-06, "loss": 1.2563, "step": 9654 }, { "epoch": 0.73894076228379, "grad_norm": 0.7032773458516598, "learning_rate": 3.3473381050019403e-06, "loss": 1.5911, "step": 9655 }, { "epoch": 0.7390172968008571, "grad_norm": 1.300538026027432, "learning_rate": 3.346596069824647e-06, "loss": 1.6495, "step": 9656 }, { "epoch": 0.7390938313179244, "grad_norm": 0.6262914611412449, "learning_rate": 3.3458541976804226e-06, "loss": 1.3943, "step": 9657 }, { "epoch": 0.7391703658349916, "grad_norm": 0.7542706379045054, "learning_rate": 3.345112488614851e-06, "loss": 1.7251, "step": 9658 }, { "epoch": 0.7392469003520588, "grad_norm": 0.5344016474389232, "learning_rate": 3.344370942673505e-06, "loss": 1.2162, "step": 9659 }, { "epoch": 0.739323434869126, "grad_norm": 0.6109593507942936, "learning_rate": 3.3436295599019476e-06, "loss": 1.5319, "step": 9660 }, { "epoch": 0.7393999693861931, "grad_norm": 0.6918165936060218, "learning_rate": 3.3428883403457303e-06, "loss": 1.6392, "step": 9661 }, { "epoch": 0.7394765039032604, "grad_norm": 0.6307346376636391, "learning_rate": 3.3421472840503965e-06, "loss": 1.6295, "step": 9662 }, { "epoch": 0.7395530384203276, "grad_norm": 1.2276511820855234, "learning_rate": 3.34140639106148e-06, "loss": 1.5738, "step": 9663 }, { "epoch": 0.7396295729373947, "grad_norm": 0.6579929202102073, "learning_rate": 3.3406656614245015e-06, "loss": 1.46, "step": 9664 }, { "epoch": 0.739706107454462, "grad_norm": 0.5577948136773571, "learning_rate": 3.3399250951849747e-06, "loss": 1.3076, "step": 9665 }, { "epoch": 0.7397826419715292, "grad_norm": 0.785358569089768, "learning_rate": 3.339184692388402e-06, "loss": 1.7979, "step": 9666 }, { "epoch": 0.7398591764885963, "grad_norm": 0.650396917763141, "learning_rate": 3.338444453080276e-06, "loss": 1.3728, "step": 9667 }, { "epoch": 0.7399357110056636, "grad_norm": 1.0194163323768346, "learning_rate": 3.3377043773060804e-06, "loss": 1.2568, "step": 9668 }, { "epoch": 0.7400122455227307, "grad_norm": 0.8104389075475023, "learning_rate": 3.3369644651112863e-06, "loss": 1.4168, "step": 9669 }, { "epoch": 0.7400887800397979, "grad_norm": 0.6147662456914865, "learning_rate": 3.3362247165413563e-06, "loss": 1.3826, "step": 9670 }, { "epoch": 0.7401653145568652, "grad_norm": 0.5955051669592827, "learning_rate": 3.3354851316417423e-06, "loss": 1.4082, "step": 9671 }, { "epoch": 0.7402418490739323, "grad_norm": 0.5919198740516999, "learning_rate": 3.3347457104578866e-06, "loss": 1.1538, "step": 9672 }, { "epoch": 0.7403183835909996, "grad_norm": 0.49855881772717836, "learning_rate": 3.334006453035223e-06, "loss": 1.0553, "step": 9673 }, { "epoch": 0.7403949181080668, "grad_norm": 0.5858628244183226, "learning_rate": 3.3332673594191735e-06, "loss": 1.3467, "step": 9674 }, { "epoch": 0.7404714526251339, "grad_norm": 0.5581152499201233, "learning_rate": 3.3325284296551486e-06, "loss": 1.4226, "step": 9675 }, { "epoch": 0.7405479871422012, "grad_norm": 0.540627583541716, "learning_rate": 3.3317896637885516e-06, "loss": 1.2269, "step": 9676 }, { "epoch": 0.7406245216592683, "grad_norm": 0.7092367143560029, "learning_rate": 3.3310510618647728e-06, "loss": 1.6462, "step": 9677 }, { "epoch": 0.7407010561763355, "grad_norm": 0.7452998731824508, "learning_rate": 3.3303126239291967e-06, "loss": 1.3002, "step": 9678 }, { "epoch": 0.7407775906934028, "grad_norm": 0.5346712145066969, "learning_rate": 3.3295743500271938e-06, "loss": 1.2022, "step": 9679 }, { "epoch": 0.7408541252104699, "grad_norm": 0.9445347144682489, "learning_rate": 3.328836240204125e-06, "loss": 1.4932, "step": 9680 }, { "epoch": 0.7409306597275371, "grad_norm": 0.5678391538659092, "learning_rate": 3.328098294505344e-06, "loss": 1.4293, "step": 9681 }, { "epoch": 0.7410071942446043, "grad_norm": 0.5780922016033778, "learning_rate": 3.32736051297619e-06, "loss": 1.3197, "step": 9682 }, { "epoch": 0.7410837287616715, "grad_norm": 0.6512209559095905, "learning_rate": 3.3266228956619967e-06, "loss": 1.779, "step": 9683 }, { "epoch": 0.7411602632787387, "grad_norm": 0.5413529772831511, "learning_rate": 3.3258854426080842e-06, "loss": 1.3308, "step": 9684 }, { "epoch": 0.7412367977958059, "grad_norm": 0.671468886757582, "learning_rate": 3.3251481538597645e-06, "loss": 1.4888, "step": 9685 }, { "epoch": 0.7413133323128731, "grad_norm": 0.5140220801527212, "learning_rate": 3.324411029462339e-06, "loss": 1.0334, "step": 9686 }, { "epoch": 0.7413898668299403, "grad_norm": 0.5915168532186043, "learning_rate": 3.3236740694610975e-06, "loss": 1.422, "step": 9687 }, { "epoch": 0.7414664013470075, "grad_norm": 0.5734031419680926, "learning_rate": 3.3229372739013222e-06, "loss": 1.447, "step": 9688 }, { "epoch": 0.7415429358640747, "grad_norm": 0.6002373725898564, "learning_rate": 3.322200642828284e-06, "loss": 1.3923, "step": 9689 }, { "epoch": 0.7416194703811418, "grad_norm": 0.9362054714577104, "learning_rate": 3.321464176287243e-06, "loss": 1.5533, "step": 9690 }, { "epoch": 0.7416960048982091, "grad_norm": 0.6403009223524568, "learning_rate": 3.3207278743234505e-06, "loss": 1.6225, "step": 9691 }, { "epoch": 0.7417725394152763, "grad_norm": 0.5393864901206641, "learning_rate": 3.3199917369821466e-06, "loss": 1.2195, "step": 9692 }, { "epoch": 0.7418490739323435, "grad_norm": 0.6552267240315792, "learning_rate": 3.319255764308561e-06, "loss": 1.4766, "step": 9693 }, { "epoch": 0.7419256084494107, "grad_norm": 0.5260253805288271, "learning_rate": 3.3185199563479163e-06, "loss": 1.1581, "step": 9694 }, { "epoch": 0.7420021429664779, "grad_norm": 0.5866273172877119, "learning_rate": 3.3177843131454224e-06, "loss": 1.3496, "step": 9695 }, { "epoch": 0.7420786774835451, "grad_norm": 0.6161664802460475, "learning_rate": 3.317048834746278e-06, "loss": 1.4094, "step": 9696 }, { "epoch": 0.7421552120006123, "grad_norm": 0.7553483017224694, "learning_rate": 3.3163135211956732e-06, "loss": 1.5382, "step": 9697 }, { "epoch": 0.7422317465176794, "grad_norm": 0.5461188705516863, "learning_rate": 3.315578372538789e-06, "loss": 1.1581, "step": 9698 }, { "epoch": 0.7423082810347467, "grad_norm": 0.6816871862435137, "learning_rate": 3.314843388820793e-06, "loss": 1.2791, "step": 9699 }, { "epoch": 0.7423848155518139, "grad_norm": 0.5642870125809453, "learning_rate": 3.314108570086847e-06, "loss": 1.3198, "step": 9700 }, { "epoch": 0.742461350068881, "grad_norm": 0.6076575169665878, "learning_rate": 3.313373916382101e-06, "loss": 1.4733, "step": 9701 }, { "epoch": 0.7425378845859483, "grad_norm": 0.611036209756144, "learning_rate": 3.312639427751693e-06, "loss": 1.3583, "step": 9702 }, { "epoch": 0.7426144191030155, "grad_norm": 0.5662827727325745, "learning_rate": 3.3119051042407508e-06, "loss": 1.3199, "step": 9703 }, { "epoch": 0.7426909536200826, "grad_norm": 0.6577488934134471, "learning_rate": 3.311170945894394e-06, "loss": 1.4616, "step": 9704 }, { "epoch": 0.7427674881371499, "grad_norm": 0.6266911344043355, "learning_rate": 3.310436952757734e-06, "loss": 1.0121, "step": 9705 }, { "epoch": 0.742844022654217, "grad_norm": 0.5767354786713519, "learning_rate": 3.3097031248758682e-06, "loss": 1.3021, "step": 9706 }, { "epoch": 0.7429205571712842, "grad_norm": 0.6484375734251028, "learning_rate": 3.3089694622938844e-06, "loss": 1.4211, "step": 9707 }, { "epoch": 0.7429970916883515, "grad_norm": 0.5251924268737357, "learning_rate": 3.308235965056862e-06, "loss": 1.064, "step": 9708 }, { "epoch": 0.7430736262054186, "grad_norm": 0.5700445730352648, "learning_rate": 3.307502633209868e-06, "loss": 1.1522, "step": 9709 }, { "epoch": 0.7431501607224859, "grad_norm": 0.5958337604476229, "learning_rate": 3.306769466797962e-06, "loss": 1.4979, "step": 9710 }, { "epoch": 0.7432266952395531, "grad_norm": 0.6128816190138676, "learning_rate": 3.306036465866191e-06, "loss": 1.5636, "step": 9711 }, { "epoch": 0.7433032297566202, "grad_norm": 1.2821840760495582, "learning_rate": 3.3053036304595926e-06, "loss": 1.7083, "step": 9712 }, { "epoch": 0.7433797642736875, "grad_norm": 0.5810915471375641, "learning_rate": 3.3045709606231945e-06, "loss": 1.2866, "step": 9713 }, { "epoch": 0.7434562987907546, "grad_norm": 0.560687623339346, "learning_rate": 3.3038384564020156e-06, "loss": 1.2811, "step": 9714 }, { "epoch": 0.7435328333078218, "grad_norm": 0.6180850465886628, "learning_rate": 3.3031061178410595e-06, "loss": 1.3843, "step": 9715 }, { "epoch": 0.7436093678248891, "grad_norm": 0.8050781909453145, "learning_rate": 3.3023739449853274e-06, "loss": 1.478, "step": 9716 }, { "epoch": 0.7436859023419562, "grad_norm": 0.5745503328320819, "learning_rate": 3.301641937879804e-06, "loss": 1.4666, "step": 9717 }, { "epoch": 0.7437624368590234, "grad_norm": 0.6429961521250769, "learning_rate": 3.3009100965694673e-06, "loss": 1.5769, "step": 9718 }, { "epoch": 0.7438389713760906, "grad_norm": 0.6060413994706852, "learning_rate": 3.3001784210992826e-06, "loss": 1.3301, "step": 9719 }, { "epoch": 0.7439155058931578, "grad_norm": 0.5990584217016971, "learning_rate": 3.2994469115142067e-06, "loss": 1.3292, "step": 9720 }, { "epoch": 0.743992040410225, "grad_norm": 0.6120397109781989, "learning_rate": 3.298715567859184e-06, "loss": 1.4742, "step": 9721 }, { "epoch": 0.7440685749272922, "grad_norm": 0.6320196042035094, "learning_rate": 3.297984390179154e-06, "loss": 1.4685, "step": 9722 }, { "epoch": 0.7441451094443594, "grad_norm": 0.5366098957903663, "learning_rate": 3.2972533785190404e-06, "loss": 1.2518, "step": 9723 }, { "epoch": 0.7442216439614266, "grad_norm": 0.5195953964079243, "learning_rate": 3.296522532923758e-06, "loss": 1.1762, "step": 9724 }, { "epoch": 0.7442981784784938, "grad_norm": 0.5295931734829691, "learning_rate": 3.2957918534382145e-06, "loss": 1.293, "step": 9725 }, { "epoch": 0.744374712995561, "grad_norm": 0.5361928922364928, "learning_rate": 3.2950613401073017e-06, "loss": 1.1876, "step": 9726 }, { "epoch": 0.7444512475126281, "grad_norm": 0.609001687559032, "learning_rate": 3.294330992975907e-06, "loss": 1.4824, "step": 9727 }, { "epoch": 0.7445277820296954, "grad_norm": 0.5975868775251134, "learning_rate": 3.2936008120889055e-06, "loss": 1.2722, "step": 9728 }, { "epoch": 0.7446043165467626, "grad_norm": 1.031164857559144, "learning_rate": 3.2928707974911606e-06, "loss": 1.1547, "step": 9729 }, { "epoch": 0.7446808510638298, "grad_norm": 0.5795729936319912, "learning_rate": 3.292140949227526e-06, "loss": 1.3442, "step": 9730 }, { "epoch": 0.744757385580897, "grad_norm": 0.542160828780783, "learning_rate": 3.291411267342847e-06, "loss": 1.2787, "step": 9731 }, { "epoch": 0.7448339200979642, "grad_norm": 0.726453454952795, "learning_rate": 3.2906817518819566e-06, "loss": 1.3976, "step": 9732 }, { "epoch": 0.7449104546150314, "grad_norm": 0.6068169036330083, "learning_rate": 3.289952402889679e-06, "loss": 1.519, "step": 9733 }, { "epoch": 0.7449869891320986, "grad_norm": 0.5656688426395814, "learning_rate": 3.2892232204108264e-06, "loss": 1.2098, "step": 9734 }, { "epoch": 0.7450635236491657, "grad_norm": 0.5741827227972103, "learning_rate": 3.288494204490204e-06, "loss": 1.4021, "step": 9735 }, { "epoch": 0.745140058166233, "grad_norm": 0.5735259603923814, "learning_rate": 3.287765355172603e-06, "loss": 1.3753, "step": 9736 }, { "epoch": 0.7452165926833002, "grad_norm": 0.6724997391241216, "learning_rate": 3.287036672502805e-06, "loss": 1.7867, "step": 9737 }, { "epoch": 0.7452931272003673, "grad_norm": 0.6893106395657207, "learning_rate": 3.286308156525586e-06, "loss": 1.863, "step": 9738 }, { "epoch": 0.7453696617174346, "grad_norm": 0.5520926368727582, "learning_rate": 3.285579807285706e-06, "loss": 1.2093, "step": 9739 }, { "epoch": 0.7454461962345018, "grad_norm": 0.7937662957038616, "learning_rate": 3.284851624827917e-06, "loss": 1.4091, "step": 9740 }, { "epoch": 0.7455227307515689, "grad_norm": 0.5320959232313557, "learning_rate": 3.284123609196961e-06, "loss": 1.1441, "step": 9741 }, { "epoch": 0.7455992652686362, "grad_norm": 0.7321407448445171, "learning_rate": 3.28339576043757e-06, "loss": 1.3983, "step": 9742 }, { "epoch": 0.7456757997857033, "grad_norm": 0.5793557203999634, "learning_rate": 3.282668078594462e-06, "loss": 1.385, "step": 9743 }, { "epoch": 0.7457523343027705, "grad_norm": 0.5984408187467931, "learning_rate": 3.281940563712353e-06, "loss": 1.2857, "step": 9744 }, { "epoch": 0.7458288688198378, "grad_norm": 0.6088882794924944, "learning_rate": 3.2812132158359404e-06, "loss": 1.4103, "step": 9745 }, { "epoch": 0.7459054033369049, "grad_norm": 0.5090428568158677, "learning_rate": 3.2804860350099167e-06, "loss": 1.1123, "step": 9746 }, { "epoch": 0.7459819378539722, "grad_norm": 0.59293339323847, "learning_rate": 3.27975902127896e-06, "loss": 1.3094, "step": 9747 }, { "epoch": 0.7460584723710394, "grad_norm": 0.6189578314777943, "learning_rate": 3.279032174687739e-06, "loss": 1.3764, "step": 9748 }, { "epoch": 0.7461350068881065, "grad_norm": 0.6417920794195167, "learning_rate": 3.2783054952809174e-06, "loss": 1.4155, "step": 9749 }, { "epoch": 0.7462115414051738, "grad_norm": 0.5927930573865833, "learning_rate": 3.277578983103142e-06, "loss": 1.4541, "step": 9750 }, { "epoch": 0.7462880759222409, "grad_norm": 0.6907864564031413, "learning_rate": 3.276852638199053e-06, "loss": 1.6887, "step": 9751 }, { "epoch": 0.7463646104393081, "grad_norm": 0.5273955980184503, "learning_rate": 3.2761264606132782e-06, "loss": 1.1613, "step": 9752 }, { "epoch": 0.7464411449563754, "grad_norm": 0.5294952115478956, "learning_rate": 3.275400450390437e-06, "loss": 1.1145, "step": 9753 }, { "epoch": 0.7465176794734425, "grad_norm": 0.675578759369621, "learning_rate": 3.2746746075751363e-06, "loss": 1.5445, "step": 9754 }, { "epoch": 0.7465942139905097, "grad_norm": 0.6474045636133489, "learning_rate": 3.2739489322119757e-06, "loss": 1.3437, "step": 9755 }, { "epoch": 0.7466707485075769, "grad_norm": 0.6793966374368569, "learning_rate": 3.2732234243455413e-06, "loss": 1.5584, "step": 9756 }, { "epoch": 0.7467472830246441, "grad_norm": 0.6964278544986456, "learning_rate": 3.272498084020411e-06, "loss": 1.4173, "step": 9757 }, { "epoch": 0.7468238175417113, "grad_norm": 0.5228325964864902, "learning_rate": 3.2717729112811525e-06, "loss": 1.3346, "step": 9758 }, { "epoch": 0.7469003520587785, "grad_norm": 0.5795624875486002, "learning_rate": 3.2710479061723217e-06, "loss": 1.3235, "step": 9759 }, { "epoch": 0.7469768865758457, "grad_norm": 0.5380311777634321, "learning_rate": 3.270323068738466e-06, "loss": 1.3019, "step": 9760 }, { "epoch": 0.747053421092913, "grad_norm": 0.5213681412853661, "learning_rate": 3.2695983990241208e-06, "loss": 0.9989, "step": 9761 }, { "epoch": 0.7471299556099801, "grad_norm": 0.9897231509936983, "learning_rate": 3.2688738970738132e-06, "loss": 1.4598, "step": 9762 }, { "epoch": 0.7472064901270473, "grad_norm": 0.526815304452774, "learning_rate": 3.2681495629320582e-06, "loss": 1.1868, "step": 9763 }, { "epoch": 0.7472830246441144, "grad_norm": 1.1322116499768715, "learning_rate": 3.2674253966433593e-06, "loss": 1.6825, "step": 9764 }, { "epoch": 0.7473595591611817, "grad_norm": 0.8846829557993934, "learning_rate": 3.2667013982522123e-06, "loss": 1.466, "step": 9765 }, { "epoch": 0.7474360936782489, "grad_norm": 0.7371793250536243, "learning_rate": 3.2659775678031035e-06, "loss": 1.8854, "step": 9766 }, { "epoch": 0.747512628195316, "grad_norm": 0.6037926521679449, "learning_rate": 3.2652539053405063e-06, "loss": 1.1093, "step": 9767 }, { "epoch": 0.7475891627123833, "grad_norm": 0.5300650249820366, "learning_rate": 3.264530410908885e-06, "loss": 1.1156, "step": 9768 }, { "epoch": 0.7476656972294505, "grad_norm": 1.0662192783598885, "learning_rate": 3.263807084552692e-06, "loss": 1.7339, "step": 9769 }, { "epoch": 0.7477422317465177, "grad_norm": 0.5430172085259409, "learning_rate": 3.2630839263163707e-06, "loss": 1.2455, "step": 9770 }, { "epoch": 0.7478187662635849, "grad_norm": 0.6272757102899859, "learning_rate": 3.2623609362443555e-06, "loss": 1.4352, "step": 9771 }, { "epoch": 0.747895300780652, "grad_norm": 0.5833362248350228, "learning_rate": 3.261638114381069e-06, "loss": 1.3978, "step": 9772 }, { "epoch": 0.7479718352977193, "grad_norm": 0.6789256931304679, "learning_rate": 3.2609154607709225e-06, "loss": 1.3719, "step": 9773 }, { "epoch": 0.7480483698147865, "grad_norm": 0.4966363244188992, "learning_rate": 3.260192975458318e-06, "loss": 1.0679, "step": 9774 }, { "epoch": 0.7481249043318536, "grad_norm": 0.6297909023311236, "learning_rate": 3.259470658487648e-06, "loss": 1.2539, "step": 9775 }, { "epoch": 0.7482014388489209, "grad_norm": 0.74036566614543, "learning_rate": 3.2587485099032932e-06, "loss": 1.2214, "step": 9776 }, { "epoch": 0.7482779733659881, "grad_norm": 0.6967989148528104, "learning_rate": 3.258026529749625e-06, "loss": 1.3953, "step": 9777 }, { "epoch": 0.7483545078830552, "grad_norm": 0.5848264996082644, "learning_rate": 3.257304718071003e-06, "loss": 1.4649, "step": 9778 }, { "epoch": 0.7484310424001225, "grad_norm": 0.49143923361088954, "learning_rate": 3.256583074911779e-06, "loss": 1.0066, "step": 9779 }, { "epoch": 0.7485075769171896, "grad_norm": 1.2743958621837619, "learning_rate": 3.2558616003162923e-06, "loss": 1.5711, "step": 9780 }, { "epoch": 0.7485841114342568, "grad_norm": 0.5424668608912318, "learning_rate": 3.2551402943288705e-06, "loss": 1.2508, "step": 9781 }, { "epoch": 0.7486606459513241, "grad_norm": 0.6102239422143175, "learning_rate": 3.2544191569938354e-06, "loss": 1.4944, "step": 9782 }, { "epoch": 0.7487371804683912, "grad_norm": 0.5706173082250737, "learning_rate": 3.2536981883554964e-06, "loss": 1.3056, "step": 9783 }, { "epoch": 0.7488137149854585, "grad_norm": 0.6422521969548667, "learning_rate": 3.2529773884581493e-06, "loss": 1.4687, "step": 9784 }, { "epoch": 0.7488902495025256, "grad_norm": 0.580597713807191, "learning_rate": 3.252256757346084e-06, "loss": 1.4147, "step": 9785 }, { "epoch": 0.7489667840195928, "grad_norm": 0.8064289748309691, "learning_rate": 3.2515362950635776e-06, "loss": 1.5371, "step": 9786 }, { "epoch": 0.7490433185366601, "grad_norm": 0.5936736450724731, "learning_rate": 3.250816001654897e-06, "loss": 1.3487, "step": 9787 }, { "epoch": 0.7491198530537272, "grad_norm": 0.5717545610224282, "learning_rate": 3.2500958771643005e-06, "loss": 1.3508, "step": 9788 }, { "epoch": 0.7491963875707944, "grad_norm": 0.6359107118838833, "learning_rate": 3.249375921636033e-06, "loss": 1.4788, "step": 9789 }, { "epoch": 0.7492729220878617, "grad_norm": 0.6000586664233557, "learning_rate": 3.248656135114333e-06, "loss": 1.4018, "step": 9790 }, { "epoch": 0.7493494566049288, "grad_norm": 0.8159354518957928, "learning_rate": 3.2479365176434254e-06, "loss": 1.726, "step": 9791 }, { "epoch": 0.749425991121996, "grad_norm": 0.7510021852558242, "learning_rate": 3.2472170692675226e-06, "loss": 1.9751, "step": 9792 }, { "epoch": 0.7495025256390632, "grad_norm": 0.5581134839461236, "learning_rate": 3.246497790030835e-06, "loss": 1.2609, "step": 9793 }, { "epoch": 0.7495790601561304, "grad_norm": 0.6418518813384234, "learning_rate": 3.2457786799775533e-06, "loss": 1.5928, "step": 9794 }, { "epoch": 0.7496555946731976, "grad_norm": 0.5849274229845612, "learning_rate": 3.2450597391518635e-06, "loss": 1.286, "step": 9795 }, { "epoch": 0.7497321291902648, "grad_norm": 0.590922327444732, "learning_rate": 3.244340967597939e-06, "loss": 1.4479, "step": 9796 }, { "epoch": 0.749808663707332, "grad_norm": 1.477877738890031, "learning_rate": 3.243622365359943e-06, "loss": 1.7951, "step": 9797 }, { "epoch": 0.7498851982243993, "grad_norm": 0.516432303173067, "learning_rate": 3.242903932482029e-06, "loss": 1.2001, "step": 9798 }, { "epoch": 0.7499617327414664, "grad_norm": 0.6749346718199084, "learning_rate": 3.24218566900834e-06, "loss": 1.5135, "step": 9799 }, { "epoch": 0.7500382672585336, "grad_norm": 0.5300126663669829, "learning_rate": 3.2414675749830077e-06, "loss": 1.23, "step": 9800 }, { "epoch": 0.7501148017756007, "grad_norm": 0.5364791835962969, "learning_rate": 3.2407496504501536e-06, "loss": 1.2684, "step": 9801 }, { "epoch": 0.750191336292668, "grad_norm": 0.7099005123464862, "learning_rate": 3.24003189545389e-06, "loss": 1.3859, "step": 9802 }, { "epoch": 0.7502678708097352, "grad_norm": 0.5982966950508283, "learning_rate": 3.239314310038316e-06, "loss": 1.3389, "step": 9803 }, { "epoch": 0.7503444053268024, "grad_norm": 0.621571304708494, "learning_rate": 3.238596894247525e-06, "loss": 1.4632, "step": 9804 }, { "epoch": 0.7504209398438696, "grad_norm": 0.5851722972807339, "learning_rate": 3.2378796481255957e-06, "loss": 1.4116, "step": 9805 }, { "epoch": 0.7504974743609368, "grad_norm": 0.7973897859981706, "learning_rate": 3.2371625717165983e-06, "loss": 1.4115, "step": 9806 }, { "epoch": 0.750574008878004, "grad_norm": 0.6111996872176582, "learning_rate": 3.2364456650645913e-06, "loss": 1.388, "step": 9807 }, { "epoch": 0.7506505433950712, "grad_norm": 0.554824196365967, "learning_rate": 3.235728928213625e-06, "loss": 1.3124, "step": 9808 }, { "epoch": 0.7507270779121383, "grad_norm": 0.4949251219256888, "learning_rate": 3.2350123612077346e-06, "loss": 1.0356, "step": 9809 }, { "epoch": 0.7508036124292056, "grad_norm": 0.672353028024176, "learning_rate": 3.234295964090952e-06, "loss": 1.5174, "step": 9810 }, { "epoch": 0.7508801469462728, "grad_norm": 0.8002141263903074, "learning_rate": 3.2335797369072943e-06, "loss": 1.7079, "step": 9811 }, { "epoch": 0.7509566814633399, "grad_norm": 0.7040109773578083, "learning_rate": 3.232863679700766e-06, "loss": 1.4614, "step": 9812 }, { "epoch": 0.7510332159804072, "grad_norm": 0.7895605205371589, "learning_rate": 3.2321477925153667e-06, "loss": 1.2684, "step": 9813 }, { "epoch": 0.7511097504974744, "grad_norm": 0.5237935831619367, "learning_rate": 3.231432075395079e-06, "loss": 1.1875, "step": 9814 }, { "epoch": 0.7511862850145415, "grad_norm": 0.7323677918692155, "learning_rate": 3.230716528383883e-06, "loss": 1.6784, "step": 9815 }, { "epoch": 0.7512628195316088, "grad_norm": 0.8163191092606958, "learning_rate": 3.2300011515257424e-06, "loss": 1.4902, "step": 9816 }, { "epoch": 0.7513393540486759, "grad_norm": 0.549219491392109, "learning_rate": 3.229285944864611e-06, "loss": 1.402, "step": 9817 }, { "epoch": 0.7514158885657432, "grad_norm": 0.5470048290360977, "learning_rate": 3.2285709084444338e-06, "loss": 1.2125, "step": 9818 }, { "epoch": 0.7514924230828104, "grad_norm": 0.5839089231412633, "learning_rate": 3.2278560423091455e-06, "loss": 1.2256, "step": 9819 }, { "epoch": 0.7515689575998775, "grad_norm": 0.5659213238426111, "learning_rate": 3.227141346502669e-06, "loss": 1.3642, "step": 9820 }, { "epoch": 0.7516454921169448, "grad_norm": 0.6139366210373448, "learning_rate": 3.2264268210689175e-06, "loss": 1.4266, "step": 9821 }, { "epoch": 0.7517220266340119, "grad_norm": 0.43849832905889513, "learning_rate": 3.2257124660517935e-06, "loss": 0.8486, "step": 9822 }, { "epoch": 0.7517985611510791, "grad_norm": 0.4897373180884757, "learning_rate": 3.2249982814951895e-06, "loss": 1.065, "step": 9823 }, { "epoch": 0.7518750956681464, "grad_norm": 0.5433742550844538, "learning_rate": 3.2242842674429864e-06, "loss": 1.0137, "step": 9824 }, { "epoch": 0.7519516301852135, "grad_norm": 0.5120886780647532, "learning_rate": 3.223570423939054e-06, "loss": 1.161, "step": 9825 }, { "epoch": 0.7520281647022807, "grad_norm": 0.6909768660077834, "learning_rate": 3.2228567510272567e-06, "loss": 1.5571, "step": 9826 }, { "epoch": 0.752104699219348, "grad_norm": 0.564143412163001, "learning_rate": 3.2221432487514425e-06, "loss": 1.2622, "step": 9827 }, { "epoch": 0.7521812337364151, "grad_norm": 0.6555032770751655, "learning_rate": 3.2214299171554506e-06, "loss": 1.5858, "step": 9828 }, { "epoch": 0.7522577682534823, "grad_norm": 0.6412632268683987, "learning_rate": 3.220716756283112e-06, "loss": 1.5829, "step": 9829 }, { "epoch": 0.7523343027705495, "grad_norm": 0.6128701719747724, "learning_rate": 3.2200037661782424e-06, "loss": 1.53, "step": 9830 }, { "epoch": 0.7524108372876167, "grad_norm": 0.8145026884540061, "learning_rate": 3.219290946884654e-06, "loss": 1.7207, "step": 9831 }, { "epoch": 0.752487371804684, "grad_norm": 0.5494291022712197, "learning_rate": 3.218578298446142e-06, "loss": 1.1779, "step": 9832 }, { "epoch": 0.7525639063217511, "grad_norm": 0.8494813700275263, "learning_rate": 3.217865820906494e-06, "loss": 1.8781, "step": 9833 }, { "epoch": 0.7526404408388183, "grad_norm": 0.7381250434000459, "learning_rate": 3.2171535143094872e-06, "loss": 1.5862, "step": 9834 }, { "epoch": 0.7527169753558856, "grad_norm": 0.6430231299007457, "learning_rate": 3.216441378698888e-06, "loss": 1.6321, "step": 9835 }, { "epoch": 0.7527935098729527, "grad_norm": 0.5471510088155652, "learning_rate": 3.21572941411845e-06, "loss": 1.2291, "step": 9836 }, { "epoch": 0.7528700443900199, "grad_norm": 0.604853432121799, "learning_rate": 3.2150176206119217e-06, "loss": 1.41, "step": 9837 }, { "epoch": 0.752946578907087, "grad_norm": 0.8288925213579535, "learning_rate": 3.2143059982230363e-06, "loss": 1.3305, "step": 9838 }, { "epoch": 0.7530231134241543, "grad_norm": 0.5403886207144117, "learning_rate": 3.2135945469955178e-06, "loss": 1.3284, "step": 9839 }, { "epoch": 0.7530996479412215, "grad_norm": 0.5939885392724542, "learning_rate": 3.21288326697308e-06, "loss": 1.3568, "step": 9840 }, { "epoch": 0.7531761824582887, "grad_norm": 0.5398605487928562, "learning_rate": 3.2121721581994262e-06, "loss": 1.4161, "step": 9841 }, { "epoch": 0.7532527169753559, "grad_norm": 0.6868953754712939, "learning_rate": 3.2114612207182492e-06, "loss": 1.68, "step": 9842 }, { "epoch": 0.7533292514924231, "grad_norm": 0.6086733966028036, "learning_rate": 3.2107504545732315e-06, "loss": 1.4654, "step": 9843 }, { "epoch": 0.7534057860094903, "grad_norm": 0.49845569818734375, "learning_rate": 3.2100398598080428e-06, "loss": 1.1965, "step": 9844 }, { "epoch": 0.7534823205265575, "grad_norm": 0.5947095983708351, "learning_rate": 3.2093294364663462e-06, "loss": 1.1855, "step": 9845 }, { "epoch": 0.7535588550436246, "grad_norm": 0.5571421884306688, "learning_rate": 3.208619184591792e-06, "loss": 1.3142, "step": 9846 }, { "epoch": 0.7536353895606919, "grad_norm": 0.6825345918130373, "learning_rate": 3.2079091042280185e-06, "loss": 1.2166, "step": 9847 }, { "epoch": 0.7537119240777591, "grad_norm": 0.6480713617759969, "learning_rate": 3.207199195418657e-06, "loss": 1.5488, "step": 9848 }, { "epoch": 0.7537884585948262, "grad_norm": 0.5599712810796065, "learning_rate": 3.206489458207326e-06, "loss": 1.3172, "step": 9849 }, { "epoch": 0.7538649931118935, "grad_norm": 0.5629997699741935, "learning_rate": 3.205779892637634e-06, "loss": 1.2182, "step": 9850 }, { "epoch": 0.7539415276289606, "grad_norm": 0.7664016617314169, "learning_rate": 3.205070498753178e-06, "loss": 1.4917, "step": 9851 }, { "epoch": 0.7540180621460278, "grad_norm": 0.7655523365741503, "learning_rate": 3.2043612765975452e-06, "loss": 1.3537, "step": 9852 }, { "epoch": 0.7540945966630951, "grad_norm": 0.58465684641732, "learning_rate": 3.2036522262143134e-06, "loss": 1.4849, "step": 9853 }, { "epoch": 0.7541711311801622, "grad_norm": 0.5135688840563817, "learning_rate": 3.202943347647049e-06, "loss": 1.2276, "step": 9854 }, { "epoch": 0.7542476656972295, "grad_norm": 0.52442730624799, "learning_rate": 3.202234640939307e-06, "loss": 1.3061, "step": 9855 }, { "epoch": 0.7543242002142967, "grad_norm": 0.5549680808951473, "learning_rate": 3.2015261061346315e-06, "loss": 1.301, "step": 9856 }, { "epoch": 0.7544007347313638, "grad_norm": 0.5206931981008874, "learning_rate": 3.2008177432765584e-06, "loss": 1.3385, "step": 9857 }, { "epoch": 0.7544772692484311, "grad_norm": 0.6003916953149648, "learning_rate": 3.20010955240861e-06, "loss": 1.4211, "step": 9858 }, { "epoch": 0.7545538037654982, "grad_norm": 0.6473488060524395, "learning_rate": 3.1994015335743017e-06, "loss": 1.4413, "step": 9859 }, { "epoch": 0.7546303382825654, "grad_norm": 0.5746784990480117, "learning_rate": 3.198693686817136e-06, "loss": 1.0621, "step": 9860 }, { "epoch": 0.7547068727996327, "grad_norm": 0.6078922090324871, "learning_rate": 3.1979860121806037e-06, "loss": 1.3097, "step": 9861 }, { "epoch": 0.7547834073166998, "grad_norm": 0.5605398936790811, "learning_rate": 3.1972785097081883e-06, "loss": 1.2559, "step": 9862 }, { "epoch": 0.754859941833767, "grad_norm": 0.718053639614418, "learning_rate": 3.1965711794433584e-06, "loss": 1.2679, "step": 9863 }, { "epoch": 0.7549364763508343, "grad_norm": 0.6180318322433278, "learning_rate": 3.195864021429577e-06, "loss": 1.3536, "step": 9864 }, { "epoch": 0.7550130108679014, "grad_norm": 0.5738705854343146, "learning_rate": 3.195157035710292e-06, "loss": 1.1987, "step": 9865 }, { "epoch": 0.7550895453849686, "grad_norm": 0.6559865076982282, "learning_rate": 3.194450222328944e-06, "loss": 1.4387, "step": 9866 }, { "epoch": 0.7551660799020358, "grad_norm": 0.5582900096269576, "learning_rate": 3.193743581328962e-06, "loss": 1.3312, "step": 9867 }, { "epoch": 0.755242614419103, "grad_norm": 0.6650367630167088, "learning_rate": 3.193037112753763e-06, "loss": 1.6283, "step": 9868 }, { "epoch": 0.7553191489361702, "grad_norm": 0.504702364310145, "learning_rate": 3.1923308166467536e-06, "loss": 1.1517, "step": 9869 }, { "epoch": 0.7553956834532374, "grad_norm": 0.7350238218758259, "learning_rate": 3.1916246930513338e-06, "loss": 1.3381, "step": 9870 }, { "epoch": 0.7554722179703046, "grad_norm": 0.7067831984702004, "learning_rate": 3.190918742010888e-06, "loss": 1.297, "step": 9871 }, { "epoch": 0.7555487524873719, "grad_norm": 0.6244847875972094, "learning_rate": 3.1902129635687924e-06, "loss": 1.6592, "step": 9872 }, { "epoch": 0.755625287004439, "grad_norm": 0.6233695344300269, "learning_rate": 3.1895073577684122e-06, "loss": 1.2669, "step": 9873 }, { "epoch": 0.7557018215215062, "grad_norm": 0.8534140651525823, "learning_rate": 3.1888019246530993e-06, "loss": 1.3293, "step": 9874 }, { "epoch": 0.7557783560385734, "grad_norm": 0.6350840764006692, "learning_rate": 3.1880966642662028e-06, "loss": 1.499, "step": 9875 }, { "epoch": 0.7558548905556406, "grad_norm": 0.5028915672284973, "learning_rate": 3.187391576651052e-06, "loss": 1.1427, "step": 9876 }, { "epoch": 0.7559314250727078, "grad_norm": 0.6669596069384611, "learning_rate": 3.1866866618509718e-06, "loss": 1.5912, "step": 9877 }, { "epoch": 0.756007959589775, "grad_norm": 0.5753447760708529, "learning_rate": 3.1859819199092723e-06, "loss": 1.4075, "step": 9878 }, { "epoch": 0.7560844941068422, "grad_norm": 0.49875348091980576, "learning_rate": 3.185277350869257e-06, "loss": 1.1052, "step": 9879 }, { "epoch": 0.7561610286239094, "grad_norm": 0.4797694461021805, "learning_rate": 3.184572954774214e-06, "loss": 0.9665, "step": 9880 }, { "epoch": 0.7562375631409766, "grad_norm": 0.5818531415835836, "learning_rate": 3.1838687316674267e-06, "loss": 1.3688, "step": 9881 }, { "epoch": 0.7563140976580438, "grad_norm": 0.7574552622693648, "learning_rate": 3.1831646815921624e-06, "loss": 1.5561, "step": 9882 }, { "epoch": 0.7563906321751109, "grad_norm": 0.6051187087985298, "learning_rate": 3.1824608045916815e-06, "loss": 1.1835, "step": 9883 }, { "epoch": 0.7564671666921782, "grad_norm": 0.6681835324510161, "learning_rate": 3.181757100709231e-06, "loss": 1.5862, "step": 9884 }, { "epoch": 0.7565437012092454, "grad_norm": 0.6475469486386187, "learning_rate": 3.1810535699880486e-06, "loss": 1.7669, "step": 9885 }, { "epoch": 0.7566202357263125, "grad_norm": 0.4702493866395275, "learning_rate": 3.1803502124713636e-06, "loss": 0.837, "step": 9886 }, { "epoch": 0.7566967702433798, "grad_norm": 0.5893453858229382, "learning_rate": 3.17964702820239e-06, "loss": 1.3649, "step": 9887 }, { "epoch": 0.7567733047604469, "grad_norm": 0.7248102824785071, "learning_rate": 3.178944017224334e-06, "loss": 1.5088, "step": 9888 }, { "epoch": 0.7568498392775141, "grad_norm": 0.5596177553515529, "learning_rate": 3.1782411795803915e-06, "loss": 1.2628, "step": 9889 }, { "epoch": 0.7569263737945814, "grad_norm": 0.6790657753858241, "learning_rate": 3.1775385153137462e-06, "loss": 1.6164, "step": 9890 }, { "epoch": 0.7570029083116485, "grad_norm": 0.5683028951465964, "learning_rate": 3.17683602446757e-06, "loss": 1.222, "step": 9891 }, { "epoch": 0.7570794428287158, "grad_norm": 0.5590912110071636, "learning_rate": 3.1761337070850293e-06, "loss": 1.1798, "step": 9892 }, { "epoch": 0.757155977345783, "grad_norm": 0.5587731383833904, "learning_rate": 3.175431563209277e-06, "loss": 1.2477, "step": 9893 }, { "epoch": 0.7572325118628501, "grad_norm": 0.5581909767745052, "learning_rate": 3.174729592883452e-06, "loss": 1.2668, "step": 9894 }, { "epoch": 0.7573090463799174, "grad_norm": 0.6118917989815714, "learning_rate": 3.1740277961506872e-06, "loss": 1.4236, "step": 9895 }, { "epoch": 0.7573855808969845, "grad_norm": 0.53529083962189, "learning_rate": 3.173326173054101e-06, "loss": 1.3682, "step": 9896 }, { "epoch": 0.7574621154140517, "grad_norm": 0.5919467711288747, "learning_rate": 3.172624723636807e-06, "loss": 1.3217, "step": 9897 }, { "epoch": 0.757538649931119, "grad_norm": 0.580770629970865, "learning_rate": 3.1719234479419016e-06, "loss": 1.3423, "step": 9898 }, { "epoch": 0.7576151844481861, "grad_norm": 0.52754402206214, "learning_rate": 3.1712223460124736e-06, "loss": 1.0752, "step": 9899 }, { "epoch": 0.7576917189652533, "grad_norm": 0.5871616561170583, "learning_rate": 3.170521417891602e-06, "loss": 1.4497, "step": 9900 }, { "epoch": 0.7577682534823206, "grad_norm": 0.5053604864129969, "learning_rate": 3.1698206636223518e-06, "loss": 0.8856, "step": 9901 }, { "epoch": 0.7578447879993877, "grad_norm": 0.6018287111468609, "learning_rate": 3.1691200832477804e-06, "loss": 1.4527, "step": 9902 }, { "epoch": 0.7579213225164549, "grad_norm": 0.584029809625164, "learning_rate": 3.1684196768109343e-06, "loss": 1.4391, "step": 9903 }, { "epoch": 0.7579978570335221, "grad_norm": 1.7456774596523912, "learning_rate": 3.1677194443548488e-06, "loss": 1.4716, "step": 9904 }, { "epoch": 0.7580743915505893, "grad_norm": 0.5369670264407369, "learning_rate": 3.1670193859225475e-06, "loss": 1.0321, "step": 9905 }, { "epoch": 0.7581509260676565, "grad_norm": 1.098544190273677, "learning_rate": 3.1663195015570436e-06, "loss": 1.3832, "step": 9906 }, { "epoch": 0.7582274605847237, "grad_norm": 0.5709083174713346, "learning_rate": 3.1656197913013397e-06, "loss": 1.4076, "step": 9907 }, { "epoch": 0.7583039951017909, "grad_norm": 0.6020271272199218, "learning_rate": 3.1649202551984305e-06, "loss": 1.348, "step": 9908 }, { "epoch": 0.7583805296188582, "grad_norm": 0.6575935132198371, "learning_rate": 3.1642208932912955e-06, "loss": 1.5297, "step": 9909 }, { "epoch": 0.7584570641359253, "grad_norm": 0.6052543991701971, "learning_rate": 3.163521705622907e-06, "loss": 1.4387, "step": 9910 }, { "epoch": 0.7585335986529925, "grad_norm": 0.5707732218849966, "learning_rate": 3.1628226922362237e-06, "loss": 1.3864, "step": 9911 }, { "epoch": 0.7586101331700597, "grad_norm": 0.5887297901303923, "learning_rate": 3.162123853174197e-06, "loss": 1.58, "step": 9912 }, { "epoch": 0.7586866676871269, "grad_norm": 0.5510915767034515, "learning_rate": 3.161425188479762e-06, "loss": 1.4891, "step": 9913 }, { "epoch": 0.7587632022041941, "grad_norm": 0.6066219495017435, "learning_rate": 3.1607266981958503e-06, "loss": 1.3479, "step": 9914 }, { "epoch": 0.7588397367212613, "grad_norm": 0.6003754800293628, "learning_rate": 3.1600283823653783e-06, "loss": 1.1419, "step": 9915 }, { "epoch": 0.7589162712383285, "grad_norm": 0.632874276769813, "learning_rate": 3.159330241031252e-06, "loss": 1.4186, "step": 9916 }, { "epoch": 0.7589928057553957, "grad_norm": 0.6253187753682224, "learning_rate": 3.1586322742363685e-06, "loss": 1.3453, "step": 9917 }, { "epoch": 0.7590693402724629, "grad_norm": 0.9396842453522256, "learning_rate": 3.15793448202361e-06, "loss": 1.6375, "step": 9918 }, { "epoch": 0.7591458747895301, "grad_norm": 0.6128562321850837, "learning_rate": 3.1572368644358547e-06, "loss": 1.6619, "step": 9919 }, { "epoch": 0.7592224093065972, "grad_norm": 0.5562269599025579, "learning_rate": 3.1565394215159645e-06, "loss": 1.3338, "step": 9920 }, { "epoch": 0.7592989438236645, "grad_norm": 0.5139588807022843, "learning_rate": 3.155842153306793e-06, "loss": 1.1709, "step": 9921 }, { "epoch": 0.7593754783407317, "grad_norm": 0.6305590642903344, "learning_rate": 3.1551450598511815e-06, "loss": 1.4341, "step": 9922 }, { "epoch": 0.7594520128577988, "grad_norm": 0.6709861573899434, "learning_rate": 3.1544481411919625e-06, "loss": 1.1429, "step": 9923 }, { "epoch": 0.7595285473748661, "grad_norm": 0.5338249242778969, "learning_rate": 3.1537513973719546e-06, "loss": 1.1901, "step": 9924 }, { "epoch": 0.7596050818919332, "grad_norm": 0.7304008229860924, "learning_rate": 3.1530548284339714e-06, "loss": 1.1086, "step": 9925 }, { "epoch": 0.7596816164090004, "grad_norm": 0.543393048294573, "learning_rate": 3.1523584344208097e-06, "loss": 1.2615, "step": 9926 }, { "epoch": 0.7597581509260677, "grad_norm": 0.5778346046656839, "learning_rate": 3.1516622153752586e-06, "loss": 1.409, "step": 9927 }, { "epoch": 0.7598346854431348, "grad_norm": 1.2021518882232711, "learning_rate": 3.150966171340096e-06, "loss": 1.2946, "step": 9928 }, { "epoch": 0.7599112199602021, "grad_norm": 0.6145734455766255, "learning_rate": 3.150270302358087e-06, "loss": 1.5258, "step": 9929 }, { "epoch": 0.7599877544772693, "grad_norm": 0.5713260455810373, "learning_rate": 3.1495746084719923e-06, "loss": 1.4563, "step": 9930 }, { "epoch": 0.7600642889943364, "grad_norm": 0.48820291664443394, "learning_rate": 3.148879089724554e-06, "loss": 0.998, "step": 9931 }, { "epoch": 0.7601408235114037, "grad_norm": 0.6091416676979913, "learning_rate": 3.148183746158508e-06, "loss": 1.3728, "step": 9932 }, { "epoch": 0.7602173580284708, "grad_norm": 0.5232612991392485, "learning_rate": 3.1474885778165778e-06, "loss": 1.1294, "step": 9933 }, { "epoch": 0.760293892545538, "grad_norm": 0.549282803683634, "learning_rate": 3.1467935847414767e-06, "loss": 1.0875, "step": 9934 }, { "epoch": 0.7603704270626053, "grad_norm": 0.5325235201807911, "learning_rate": 3.1460987669759067e-06, "loss": 1.2013, "step": 9935 }, { "epoch": 0.7604469615796724, "grad_norm": 0.6845951738348938, "learning_rate": 3.1454041245625604e-06, "loss": 1.7306, "step": 9936 }, { "epoch": 0.7605234960967396, "grad_norm": 1.5831249330991146, "learning_rate": 3.1447096575441182e-06, "loss": 1.2977, "step": 9937 }, { "epoch": 0.7606000306138069, "grad_norm": 0.7838598893532304, "learning_rate": 3.144015365963251e-06, "loss": 1.8158, "step": 9938 }, { "epoch": 0.760676565130874, "grad_norm": 0.6236901816542464, "learning_rate": 3.1433212498626166e-06, "loss": 1.7161, "step": 9939 }, { "epoch": 0.7607530996479412, "grad_norm": 0.596157998803819, "learning_rate": 3.1426273092848636e-06, "loss": 1.293, "step": 9940 }, { "epoch": 0.7608296341650084, "grad_norm": 0.5757994407698358, "learning_rate": 3.1419335442726324e-06, "loss": 1.3279, "step": 9941 }, { "epoch": 0.7609061686820756, "grad_norm": 0.5947912015113364, "learning_rate": 3.141239954868548e-06, "loss": 1.3789, "step": 9942 }, { "epoch": 0.7609827031991429, "grad_norm": 0.5983202092298688, "learning_rate": 3.1405465411152257e-06, "loss": 1.5015, "step": 9943 }, { "epoch": 0.76105923771621, "grad_norm": 0.5845224358404637, "learning_rate": 3.1398533030552735e-06, "loss": 1.3221, "step": 9944 }, { "epoch": 0.7611357722332772, "grad_norm": 1.1646704332072677, "learning_rate": 3.139160240731284e-06, "loss": 1.2753, "step": 9945 }, { "epoch": 0.7612123067503445, "grad_norm": 0.5561280672960933, "learning_rate": 3.1384673541858404e-06, "loss": 1.4211, "step": 9946 }, { "epoch": 0.7612888412674116, "grad_norm": 0.619331725353962, "learning_rate": 3.1377746434615176e-06, "loss": 1.5353, "step": 9947 }, { "epoch": 0.7613653757844788, "grad_norm": 0.7108896184904222, "learning_rate": 3.1370821086008774e-06, "loss": 1.3964, "step": 9948 }, { "epoch": 0.761441910301546, "grad_norm": 0.5055803415450212, "learning_rate": 3.1363897496464714e-06, "loss": 1.0433, "step": 9949 }, { "epoch": 0.7615184448186132, "grad_norm": 0.6015587134935393, "learning_rate": 3.1356975666408395e-06, "loss": 1.4094, "step": 9950 }, { "epoch": 0.7615949793356804, "grad_norm": 0.8280120968226095, "learning_rate": 3.1350055596265105e-06, "loss": 1.676, "step": 9951 }, { "epoch": 0.7616715138527476, "grad_norm": 0.6355497952592039, "learning_rate": 3.134313728646006e-06, "loss": 1.469, "step": 9952 }, { "epoch": 0.7617480483698148, "grad_norm": 1.712928815154997, "learning_rate": 3.1336220737418326e-06, "loss": 1.944, "step": 9953 }, { "epoch": 0.7618245828868819, "grad_norm": 0.605737551114495, "learning_rate": 3.132930594956488e-06, "loss": 1.5102, "step": 9954 }, { "epoch": 0.7619011174039492, "grad_norm": 0.6118618157647238, "learning_rate": 3.1322392923324584e-06, "loss": 1.4679, "step": 9955 }, { "epoch": 0.7619776519210164, "grad_norm": 0.5561903149437902, "learning_rate": 3.1315481659122195e-06, "loss": 1.3296, "step": 9956 }, { "epoch": 0.7620541864380835, "grad_norm": 0.5070169284219452, "learning_rate": 3.1308572157382365e-06, "loss": 1.0826, "step": 9957 }, { "epoch": 0.7621307209551508, "grad_norm": 0.5995349942488761, "learning_rate": 3.1301664418529633e-06, "loss": 1.5263, "step": 9958 }, { "epoch": 0.762207255472218, "grad_norm": 0.5562864782965816, "learning_rate": 3.1294758442988433e-06, "loss": 1.3585, "step": 9959 }, { "epoch": 0.7622837899892851, "grad_norm": 0.6029998011540204, "learning_rate": 3.128785423118308e-06, "loss": 1.2342, "step": 9960 }, { "epoch": 0.7623603245063524, "grad_norm": 0.5375803172198148, "learning_rate": 3.1280951783537805e-06, "loss": 1.1829, "step": 9961 }, { "epoch": 0.7624368590234195, "grad_norm": 0.5372673423022991, "learning_rate": 3.127405110047669e-06, "loss": 1.3226, "step": 9962 }, { "epoch": 0.7625133935404868, "grad_norm": 0.6557400829527956, "learning_rate": 3.1267152182423755e-06, "loss": 1.5462, "step": 9963 }, { "epoch": 0.762589928057554, "grad_norm": 0.5656099472174558, "learning_rate": 3.1260255029802892e-06, "loss": 1.4398, "step": 9964 }, { "epoch": 0.7626664625746211, "grad_norm": 1.2978130308223725, "learning_rate": 3.125335964303787e-06, "loss": 1.2624, "step": 9965 }, { "epoch": 0.7627429970916884, "grad_norm": 0.5052372950966352, "learning_rate": 3.124646602255238e-06, "loss": 1.0946, "step": 9966 }, { "epoch": 0.7628195316087556, "grad_norm": 0.5275955095898786, "learning_rate": 3.123957416876996e-06, "loss": 1.2403, "step": 9967 }, { "epoch": 0.7628960661258227, "grad_norm": 0.5989014923988433, "learning_rate": 3.123268408211407e-06, "loss": 1.5044, "step": 9968 }, { "epoch": 0.76297260064289, "grad_norm": 0.638651657302844, "learning_rate": 3.1225795763008093e-06, "loss": 1.1518, "step": 9969 }, { "epoch": 0.7630491351599571, "grad_norm": 0.5599129444224804, "learning_rate": 3.1218909211875226e-06, "loss": 1.353, "step": 9970 }, { "epoch": 0.7631256696770243, "grad_norm": 0.6366893023699546, "learning_rate": 3.121202442913863e-06, "loss": 1.5466, "step": 9971 }, { "epoch": 0.7632022041940916, "grad_norm": 0.5803435373869255, "learning_rate": 3.1205141415221306e-06, "loss": 1.3409, "step": 9972 }, { "epoch": 0.7632787387111587, "grad_norm": 0.8236080915324259, "learning_rate": 3.1198260170546163e-06, "loss": 1.416, "step": 9973 }, { "epoch": 0.7633552732282259, "grad_norm": 0.6268359501457795, "learning_rate": 3.1191380695536024e-06, "loss": 1.5028, "step": 9974 }, { "epoch": 0.7634318077452932, "grad_norm": 0.5683307187257259, "learning_rate": 3.1184502990613588e-06, "loss": 1.4589, "step": 9975 }, { "epoch": 0.7635083422623603, "grad_norm": 0.7466488630462877, "learning_rate": 3.1177627056201414e-06, "loss": 1.4263, "step": 9976 }, { "epoch": 0.7635848767794275, "grad_norm": 0.6500165487158734, "learning_rate": 3.117075289272201e-06, "loss": 1.5911, "step": 9977 }, { "epoch": 0.7636614112964947, "grad_norm": 0.5504502621415033, "learning_rate": 3.116388050059772e-06, "loss": 1.1554, "step": 9978 }, { "epoch": 0.7637379458135619, "grad_norm": 0.6702775284345623, "learning_rate": 3.115700988025082e-06, "loss": 1.4738, "step": 9979 }, { "epoch": 0.7638144803306292, "grad_norm": 0.567828790911515, "learning_rate": 3.1150141032103465e-06, "loss": 1.3187, "step": 9980 }, { "epoch": 0.7638910148476963, "grad_norm": 0.5791204821997605, "learning_rate": 3.114327395657768e-06, "loss": 1.2699, "step": 9981 }, { "epoch": 0.7639675493647635, "grad_norm": 0.5562883540712238, "learning_rate": 3.1136408654095413e-06, "loss": 1.3026, "step": 9982 }, { "epoch": 0.7640440838818308, "grad_norm": 0.6156039769524273, "learning_rate": 3.112954512507848e-06, "loss": 1.3345, "step": 9983 }, { "epoch": 0.7641206183988979, "grad_norm": 0.7016446591514178, "learning_rate": 3.112268336994859e-06, "loss": 1.7574, "step": 9984 }, { "epoch": 0.7641971529159651, "grad_norm": 0.666603422452789, "learning_rate": 3.1115823389127374e-06, "loss": 1.4384, "step": 9985 }, { "epoch": 0.7642736874330323, "grad_norm": 0.5688755376672844, "learning_rate": 3.1108965183036317e-06, "loss": 1.331, "step": 9986 }, { "epoch": 0.7643502219500995, "grad_norm": 0.6509621133448446, "learning_rate": 3.1102108752096805e-06, "loss": 1.7007, "step": 9987 }, { "epoch": 0.7644267564671667, "grad_norm": 0.7356769561896298, "learning_rate": 3.109525409673012e-06, "loss": 1.5812, "step": 9988 }, { "epoch": 0.7645032909842339, "grad_norm": 0.5460879595671925, "learning_rate": 3.108840121735742e-06, "loss": 1.2841, "step": 9989 }, { "epoch": 0.7645798255013011, "grad_norm": 0.6643434738135808, "learning_rate": 3.1081550114399787e-06, "loss": 1.5654, "step": 9990 }, { "epoch": 0.7646563600183682, "grad_norm": 0.6532448629700044, "learning_rate": 3.1074700788278167e-06, "loss": 1.3771, "step": 9991 }, { "epoch": 0.7647328945354355, "grad_norm": 0.5969230581034629, "learning_rate": 3.10678532394134e-06, "loss": 1.3351, "step": 9992 }, { "epoch": 0.7648094290525027, "grad_norm": 0.7386609978189339, "learning_rate": 3.1061007468226233e-06, "loss": 1.3326, "step": 9993 }, { "epoch": 0.7648859635695698, "grad_norm": 0.575363928236843, "learning_rate": 3.105416347513727e-06, "loss": 1.3481, "step": 9994 }, { "epoch": 0.7649624980866371, "grad_norm": 0.5300568685560187, "learning_rate": 3.1047321260567016e-06, "loss": 1.2135, "step": 9995 }, { "epoch": 0.7650390326037043, "grad_norm": 0.5649922235206017, "learning_rate": 3.104048082493591e-06, "loss": 1.2217, "step": 9996 }, { "epoch": 0.7651155671207714, "grad_norm": 0.6816593846000716, "learning_rate": 3.1033642168664237e-06, "loss": 1.6104, "step": 9997 }, { "epoch": 0.7651921016378387, "grad_norm": 0.595515328739395, "learning_rate": 3.102680529217218e-06, "loss": 1.3877, "step": 9998 }, { "epoch": 0.7652686361549058, "grad_norm": 0.513653309840621, "learning_rate": 3.1019970195879823e-06, "loss": 1.2077, "step": 9999 }, { "epoch": 0.765345170671973, "grad_norm": 0.9805023785492853, "learning_rate": 3.101313688020713e-06, "loss": 1.4071, "step": 10000 }, { "epoch": 0.765345170671973, "eval_loss": 1.3972890377044678, "eval_runtime": 648.9601, "eval_samples_per_second": 8.478, "eval_steps_per_second": 1.06, "step": 10000 }, { "epoch": 0.7654217051890403, "grad_norm": 0.504600435475537, "learning_rate": 3.100630534557396e-06, "loss": 1.2647, "step": 10001 }, { "epoch": 0.7654982397061074, "grad_norm": 0.8262497096551523, "learning_rate": 3.0999475592400064e-06, "loss": 1.4265, "step": 10002 }, { "epoch": 0.7655747742231747, "grad_norm": 1.0375727863938284, "learning_rate": 3.0992647621105086e-06, "loss": 1.9168, "step": 10003 }, { "epoch": 0.7656513087402419, "grad_norm": 0.518560127825003, "learning_rate": 3.098582143210855e-06, "loss": 1.213, "step": 10004 }, { "epoch": 0.765727843257309, "grad_norm": 0.5422652989533987, "learning_rate": 3.097899702582988e-06, "loss": 1.1974, "step": 10005 }, { "epoch": 0.7658043777743763, "grad_norm": 0.5615833433461604, "learning_rate": 3.097217440268838e-06, "loss": 1.2232, "step": 10006 }, { "epoch": 0.7658809122914434, "grad_norm": 0.7905554350185853, "learning_rate": 3.0965353563103273e-06, "loss": 1.7287, "step": 10007 }, { "epoch": 0.7659574468085106, "grad_norm": 0.8821281018145419, "learning_rate": 3.0958534507493643e-06, "loss": 1.5236, "step": 10008 }, { "epoch": 0.7660339813255779, "grad_norm": 0.6618833512648055, "learning_rate": 3.0951717236278467e-06, "loss": 1.6142, "step": 10009 }, { "epoch": 0.766110515842645, "grad_norm": 0.5855455093453047, "learning_rate": 3.094490174987662e-06, "loss": 1.422, "step": 10010 }, { "epoch": 0.7661870503597122, "grad_norm": 0.6434317082993877, "learning_rate": 3.0938088048706856e-06, "loss": 1.4531, "step": 10011 }, { "epoch": 0.7662635848767795, "grad_norm": 0.6873149311290522, "learning_rate": 3.093127613318786e-06, "loss": 1.7139, "step": 10012 }, { "epoch": 0.7663401193938466, "grad_norm": 0.680668388463945, "learning_rate": 3.0924466003738146e-06, "loss": 1.3428, "step": 10013 }, { "epoch": 0.7664166539109138, "grad_norm": 0.6228577769206904, "learning_rate": 3.091765766077617e-06, "loss": 1.5152, "step": 10014 }, { "epoch": 0.766493188427981, "grad_norm": 0.5888193665314926, "learning_rate": 3.0910851104720243e-06, "loss": 1.3518, "step": 10015 }, { "epoch": 0.7665697229450482, "grad_norm": 0.9188431442595694, "learning_rate": 3.090404633598858e-06, "loss": 1.4224, "step": 10016 }, { "epoch": 0.7666462574621155, "grad_norm": 0.5809611142492478, "learning_rate": 3.0897243354999283e-06, "loss": 1.3852, "step": 10017 }, { "epoch": 0.7667227919791826, "grad_norm": 0.7152742665020869, "learning_rate": 3.0890442162170364e-06, "loss": 1.2721, "step": 10018 }, { "epoch": 0.7667993264962498, "grad_norm": 0.5778239011381927, "learning_rate": 3.0883642757919704e-06, "loss": 1.3953, "step": 10019 }, { "epoch": 0.766875861013317, "grad_norm": 0.6076208860618524, "learning_rate": 3.0876845142665067e-06, "loss": 1.5541, "step": 10020 }, { "epoch": 0.7669523955303842, "grad_norm": 0.5951644873512447, "learning_rate": 3.0870049316824134e-06, "loss": 1.5368, "step": 10021 }, { "epoch": 0.7670289300474514, "grad_norm": 0.5235928183825175, "learning_rate": 3.0863255280814455e-06, "loss": 1.2921, "step": 10022 }, { "epoch": 0.7671054645645186, "grad_norm": 0.6803300318126406, "learning_rate": 3.0856463035053463e-06, "loss": 1.4251, "step": 10023 }, { "epoch": 0.7671819990815858, "grad_norm": 1.4432352595151374, "learning_rate": 3.0849672579958514e-06, "loss": 1.5862, "step": 10024 }, { "epoch": 0.767258533598653, "grad_norm": 0.6589866465095949, "learning_rate": 3.084288391594682e-06, "loss": 1.5295, "step": 10025 }, { "epoch": 0.7673350681157202, "grad_norm": 0.6360920903352897, "learning_rate": 3.0836097043435503e-06, "loss": 1.4676, "step": 10026 }, { "epoch": 0.7674116026327874, "grad_norm": 0.6172040194193616, "learning_rate": 3.082931196284157e-06, "loss": 1.4672, "step": 10027 }, { "epoch": 0.7674881371498545, "grad_norm": 0.6196597085550911, "learning_rate": 3.0822528674581897e-06, "loss": 1.4965, "step": 10028 }, { "epoch": 0.7675646716669218, "grad_norm": 0.6494818246559435, "learning_rate": 3.0815747179073298e-06, "loss": 1.5458, "step": 10029 }, { "epoch": 0.767641206183989, "grad_norm": 0.6111191459136267, "learning_rate": 3.0808967476732444e-06, "loss": 1.6805, "step": 10030 }, { "epoch": 0.7677177407010561, "grad_norm": 0.5747611651845027, "learning_rate": 3.080218956797588e-06, "loss": 1.4138, "step": 10031 }, { "epoch": 0.7677942752181234, "grad_norm": 1.5623065328369132, "learning_rate": 3.079541345322008e-06, "loss": 1.4675, "step": 10032 }, { "epoch": 0.7678708097351906, "grad_norm": 0.5582076602666511, "learning_rate": 3.078863913288137e-06, "loss": 1.1786, "step": 10033 }, { "epoch": 0.7679473442522577, "grad_norm": 0.6247136762896662, "learning_rate": 3.078186660737601e-06, "loss": 1.4513, "step": 10034 }, { "epoch": 0.768023878769325, "grad_norm": 0.6327113786038068, "learning_rate": 3.077509587712012e-06, "loss": 1.4373, "step": 10035 }, { "epoch": 0.7681004132863921, "grad_norm": 0.5833762087176404, "learning_rate": 3.076832694252969e-06, "loss": 1.2727, "step": 10036 }, { "epoch": 0.7681769478034594, "grad_norm": 0.5777167671395721, "learning_rate": 3.076155980402065e-06, "loss": 1.5588, "step": 10037 }, { "epoch": 0.7682534823205266, "grad_norm": 0.5406144090920978, "learning_rate": 3.0754794462008775e-06, "loss": 1.2777, "step": 10038 }, { "epoch": 0.7683300168375937, "grad_norm": 1.9288898562177574, "learning_rate": 3.0748030916909744e-06, "loss": 1.6528, "step": 10039 }, { "epoch": 0.768406551354661, "grad_norm": 0.6127989321901127, "learning_rate": 3.0741269169139152e-06, "loss": 1.2978, "step": 10040 }, { "epoch": 0.7684830858717282, "grad_norm": 0.8785777878619796, "learning_rate": 3.0734509219112453e-06, "loss": 1.2988, "step": 10041 }, { "epoch": 0.7685596203887953, "grad_norm": 0.6614129750037042, "learning_rate": 3.072775106724499e-06, "loss": 1.6403, "step": 10042 }, { "epoch": 0.7686361549058626, "grad_norm": 0.6406848410301444, "learning_rate": 3.072099471395202e-06, "loss": 1.3584, "step": 10043 }, { "epoch": 0.7687126894229297, "grad_norm": 0.6254295977472406, "learning_rate": 3.071424015964865e-06, "loss": 1.6433, "step": 10044 }, { "epoch": 0.7687892239399969, "grad_norm": 0.5877250951884085, "learning_rate": 3.0707487404749914e-06, "loss": 1.2746, "step": 10045 }, { "epoch": 0.7688657584570642, "grad_norm": 0.6177072169016135, "learning_rate": 3.0700736449670725e-06, "loss": 1.4905, "step": 10046 }, { "epoch": 0.7689422929741313, "grad_norm": 0.62812513935488, "learning_rate": 3.069398729482588e-06, "loss": 1.5761, "step": 10047 }, { "epoch": 0.7690188274911985, "grad_norm": 0.5886833006498748, "learning_rate": 3.068723994063005e-06, "loss": 1.3159, "step": 10048 }, { "epoch": 0.7690953620082658, "grad_norm": 0.6151664333973832, "learning_rate": 3.068049438749784e-06, "loss": 1.5364, "step": 10049 }, { "epoch": 0.7691718965253329, "grad_norm": 0.8367896223054478, "learning_rate": 3.067375063584368e-06, "loss": 1.5875, "step": 10050 }, { "epoch": 0.7692484310424001, "grad_norm": 0.5955031582240481, "learning_rate": 3.0667008686081967e-06, "loss": 1.3321, "step": 10051 }, { "epoch": 0.7693249655594673, "grad_norm": 0.5712558532204067, "learning_rate": 3.066026853862693e-06, "loss": 1.3249, "step": 10052 }, { "epoch": 0.7694015000765345, "grad_norm": 0.6262872715465676, "learning_rate": 3.06535301938927e-06, "loss": 1.5624, "step": 10053 }, { "epoch": 0.7694780345936018, "grad_norm": 0.6241118933162461, "learning_rate": 3.0646793652293304e-06, "loss": 1.5604, "step": 10054 }, { "epoch": 0.7695545691106689, "grad_norm": 0.49463448111095826, "learning_rate": 3.0640058914242654e-06, "loss": 1.1379, "step": 10055 }, { "epoch": 0.7696311036277361, "grad_norm": 0.526090884742867, "learning_rate": 3.0633325980154556e-06, "loss": 1.1725, "step": 10056 }, { "epoch": 0.7697076381448033, "grad_norm": 0.5824000951866978, "learning_rate": 3.06265948504427e-06, "loss": 1.4843, "step": 10057 }, { "epoch": 0.7697841726618705, "grad_norm": 0.9315016798891698, "learning_rate": 3.0619865525520664e-06, "loss": 1.4941, "step": 10058 }, { "epoch": 0.7698607071789377, "grad_norm": 0.6112956445112314, "learning_rate": 3.0613138005801925e-06, "loss": 1.4953, "step": 10059 }, { "epoch": 0.7699372416960049, "grad_norm": 0.608325341929928, "learning_rate": 3.0606412291699838e-06, "loss": 1.4835, "step": 10060 }, { "epoch": 0.7700137762130721, "grad_norm": 0.7350632664886458, "learning_rate": 3.0599688383627636e-06, "loss": 1.6338, "step": 10061 }, { "epoch": 0.7700903107301393, "grad_norm": 0.6927164917700563, "learning_rate": 3.059296628199848e-06, "loss": 1.7319, "step": 10062 }, { "epoch": 0.7701668452472065, "grad_norm": 0.6372313941856924, "learning_rate": 3.058624598722538e-06, "loss": 1.4953, "step": 10063 }, { "epoch": 0.7702433797642737, "grad_norm": 0.5199019255720527, "learning_rate": 3.0579527499721272e-06, "loss": 0.9424, "step": 10064 }, { "epoch": 0.7703199142813408, "grad_norm": 0.5626062938205166, "learning_rate": 3.0572810819898933e-06, "loss": 1.3614, "step": 10065 }, { "epoch": 0.7703964487984081, "grad_norm": 0.7183619933172847, "learning_rate": 3.0566095948171072e-06, "loss": 1.6813, "step": 10066 }, { "epoch": 0.7704729833154753, "grad_norm": 0.6091846372387749, "learning_rate": 3.055938288495026e-06, "loss": 1.512, "step": 10067 }, { "epoch": 0.7705495178325424, "grad_norm": 0.6401747700573631, "learning_rate": 3.055267163064899e-06, "loss": 1.4619, "step": 10068 }, { "epoch": 0.7706260523496097, "grad_norm": 0.571691804968653, "learning_rate": 3.0545962185679597e-06, "loss": 1.2785, "step": 10069 }, { "epoch": 0.7707025868666769, "grad_norm": 0.6207026700299132, "learning_rate": 3.0539254550454334e-06, "loss": 1.5582, "step": 10070 }, { "epoch": 0.770779121383744, "grad_norm": 0.6012771146045139, "learning_rate": 3.0532548725385345e-06, "loss": 1.2912, "step": 10071 }, { "epoch": 0.7708556559008113, "grad_norm": 0.5580435572574217, "learning_rate": 3.0525844710884644e-06, "loss": 1.1874, "step": 10072 }, { "epoch": 0.7709321904178784, "grad_norm": 0.5716365619615883, "learning_rate": 3.051914250736416e-06, "loss": 1.3317, "step": 10073 }, { "epoch": 0.7710087249349457, "grad_norm": 0.6658264609264678, "learning_rate": 3.0512442115235698e-06, "loss": 1.5532, "step": 10074 }, { "epoch": 0.7710852594520129, "grad_norm": 0.6068190944549043, "learning_rate": 3.0505743534910937e-06, "loss": 1.5413, "step": 10075 }, { "epoch": 0.77116179396908, "grad_norm": 0.5625145032658111, "learning_rate": 3.049904676680148e-06, "loss": 1.184, "step": 10076 }, { "epoch": 0.7712383284861473, "grad_norm": 0.5534364233611905, "learning_rate": 3.0492351811318754e-06, "loss": 1.2857, "step": 10077 }, { "epoch": 0.7713148630032145, "grad_norm": 0.687341505086915, "learning_rate": 3.0485658668874156e-06, "loss": 1.5717, "step": 10078 }, { "epoch": 0.7713913975202816, "grad_norm": 0.5555199806971293, "learning_rate": 3.047896733987893e-06, "loss": 1.1085, "step": 10079 }, { "epoch": 0.7714679320373489, "grad_norm": 0.680700141997188, "learning_rate": 3.047227782474419e-06, "loss": 1.4302, "step": 10080 }, { "epoch": 0.771544466554416, "grad_norm": 0.5467831801190075, "learning_rate": 3.046559012388097e-06, "loss": 1.202, "step": 10081 }, { "epoch": 0.7716210010714832, "grad_norm": 0.5900262027583024, "learning_rate": 3.0458904237700195e-06, "loss": 1.3639, "step": 10082 }, { "epoch": 0.7716975355885505, "grad_norm": 0.6641926233579064, "learning_rate": 3.045222016661264e-06, "loss": 1.6715, "step": 10083 }, { "epoch": 0.7717740701056176, "grad_norm": 0.6145758358298304, "learning_rate": 3.044553791102902e-06, "loss": 1.5182, "step": 10084 }, { "epoch": 0.7718506046226848, "grad_norm": 0.5590542016337077, "learning_rate": 3.04388574713599e-06, "loss": 1.2098, "step": 10085 }, { "epoch": 0.7719271391397521, "grad_norm": 0.4696347584492871, "learning_rate": 3.043217884801575e-06, "loss": 1.0776, "step": 10086 }, { "epoch": 0.7720036736568192, "grad_norm": 0.5665555258232038, "learning_rate": 3.0425502041406923e-06, "loss": 1.3649, "step": 10087 }, { "epoch": 0.7720802081738865, "grad_norm": 0.5172994647947281, "learning_rate": 3.0418827051943656e-06, "loss": 1.1638, "step": 10088 }, { "epoch": 0.7721567426909536, "grad_norm": 0.5475163236741467, "learning_rate": 3.0412153880036098e-06, "loss": 1.2398, "step": 10089 }, { "epoch": 0.7722332772080208, "grad_norm": 0.6798651924751883, "learning_rate": 3.040548252609424e-06, "loss": 1.7425, "step": 10090 }, { "epoch": 0.7723098117250881, "grad_norm": 0.5956659562010835, "learning_rate": 3.039881299052802e-06, "loss": 1.3963, "step": 10091 }, { "epoch": 0.7723863462421552, "grad_norm": 0.8380532990064185, "learning_rate": 3.039214527374722e-06, "loss": 1.5798, "step": 10092 }, { "epoch": 0.7724628807592224, "grad_norm": 0.7784645765201508, "learning_rate": 3.0385479376161516e-06, "loss": 1.3569, "step": 10093 }, { "epoch": 0.7725394152762896, "grad_norm": 0.6027519340220453, "learning_rate": 3.0378815298180487e-06, "loss": 1.5179, "step": 10094 }, { "epoch": 0.7726159497933568, "grad_norm": 0.8149900737761634, "learning_rate": 3.0372153040213604e-06, "loss": 1.2448, "step": 10095 }, { "epoch": 0.772692484310424, "grad_norm": 0.6603940355915271, "learning_rate": 3.0365492602670204e-06, "loss": 1.5177, "step": 10096 }, { "epoch": 0.7727690188274912, "grad_norm": 0.6182896655952748, "learning_rate": 3.0358833985959535e-06, "loss": 1.8198, "step": 10097 }, { "epoch": 0.7728455533445584, "grad_norm": 0.5540888414121629, "learning_rate": 3.0352177190490718e-06, "loss": 1.1355, "step": 10098 }, { "epoch": 0.7729220878616256, "grad_norm": 0.6451231547209667, "learning_rate": 3.034552221667274e-06, "loss": 1.5659, "step": 10099 }, { "epoch": 0.7729986223786928, "grad_norm": 0.7061814828317717, "learning_rate": 3.0338869064914557e-06, "loss": 1.4168, "step": 10100 }, { "epoch": 0.77307515689576, "grad_norm": 0.6904699633307063, "learning_rate": 3.0332217735624914e-06, "loss": 1.6131, "step": 10101 }, { "epoch": 0.7731516914128271, "grad_norm": 0.605625569199996, "learning_rate": 3.032556822921251e-06, "loss": 1.3435, "step": 10102 }, { "epoch": 0.7732282259298944, "grad_norm": 1.5604260526726725, "learning_rate": 3.0318920546085897e-06, "loss": 1.2833, "step": 10103 }, { "epoch": 0.7733047604469616, "grad_norm": 0.6374714004339167, "learning_rate": 3.031227468665354e-06, "loss": 1.4894, "step": 10104 }, { "epoch": 0.7733812949640287, "grad_norm": 0.7212022254592354, "learning_rate": 3.030563065132377e-06, "loss": 1.6543, "step": 10105 }, { "epoch": 0.773457829481096, "grad_norm": 0.5388865394755897, "learning_rate": 3.0298988440504828e-06, "loss": 1.2591, "step": 10106 }, { "epoch": 0.7735343639981632, "grad_norm": 0.7627846952675866, "learning_rate": 3.0292348054604824e-06, "loss": 1.2294, "step": 10107 }, { "epoch": 0.7736108985152303, "grad_norm": 0.5863952809360518, "learning_rate": 3.0285709494031767e-06, "loss": 1.4009, "step": 10108 }, { "epoch": 0.7736874330322976, "grad_norm": 0.6254406809101889, "learning_rate": 3.0279072759193544e-06, "loss": 1.5723, "step": 10109 }, { "epoch": 0.7737639675493647, "grad_norm": 0.5293754129568455, "learning_rate": 3.0272437850497936e-06, "loss": 1.1806, "step": 10110 }, { "epoch": 0.773840502066432, "grad_norm": 0.7967306384495514, "learning_rate": 3.0265804768352624e-06, "loss": 1.6033, "step": 10111 }, { "epoch": 0.7739170365834992, "grad_norm": 0.5864986765091277, "learning_rate": 3.025917351316515e-06, "loss": 1.3893, "step": 10112 }, { "epoch": 0.7739935711005663, "grad_norm": 0.574664767637126, "learning_rate": 3.0252544085342972e-06, "loss": 1.2715, "step": 10113 }, { "epoch": 0.7740701056176336, "grad_norm": 0.6425762789220758, "learning_rate": 3.02459164852934e-06, "loss": 1.4062, "step": 10114 }, { "epoch": 0.7741466401347008, "grad_norm": 0.9236396311828228, "learning_rate": 3.023929071342368e-06, "loss": 1.6012, "step": 10115 }, { "epoch": 0.7742231746517679, "grad_norm": 0.5859368641998283, "learning_rate": 3.0232666770140888e-06, "loss": 1.3117, "step": 10116 }, { "epoch": 0.7742997091688352, "grad_norm": 0.5875472990891456, "learning_rate": 3.022604465585205e-06, "loss": 1.2877, "step": 10117 }, { "epoch": 0.7743762436859023, "grad_norm": 0.6615980333963862, "learning_rate": 3.0219424370964033e-06, "loss": 1.4678, "step": 10118 }, { "epoch": 0.7744527782029695, "grad_norm": 0.6468040223161078, "learning_rate": 3.0212805915883616e-06, "loss": 1.6077, "step": 10119 }, { "epoch": 0.7745293127200368, "grad_norm": 0.8178552271468964, "learning_rate": 3.020618929101744e-06, "loss": 1.3504, "step": 10120 }, { "epoch": 0.7746058472371039, "grad_norm": 0.628606741551576, "learning_rate": 3.0199574496772054e-06, "loss": 1.5318, "step": 10121 }, { "epoch": 0.7746823817541711, "grad_norm": 1.0998924852450787, "learning_rate": 3.0192961533553913e-06, "loss": 1.6525, "step": 10122 }, { "epoch": 0.7747589162712383, "grad_norm": 0.6418154258115166, "learning_rate": 3.0186350401769315e-06, "loss": 1.4648, "step": 10123 }, { "epoch": 0.7748354507883055, "grad_norm": 0.5134470192311864, "learning_rate": 3.017974110182448e-06, "loss": 1.0112, "step": 10124 }, { "epoch": 0.7749119853053728, "grad_norm": 0.5629125601807334, "learning_rate": 3.0173133634125497e-06, "loss": 1.4041, "step": 10125 }, { "epoch": 0.7749885198224399, "grad_norm": 0.49791608347053834, "learning_rate": 3.016652799907833e-06, "loss": 1.2134, "step": 10126 }, { "epoch": 0.7750650543395071, "grad_norm": 1.0218828833099691, "learning_rate": 3.015992419708888e-06, "loss": 1.5223, "step": 10127 }, { "epoch": 0.7751415888565744, "grad_norm": 0.5667756113886493, "learning_rate": 3.0153322228562897e-06, "loss": 1.387, "step": 10128 }, { "epoch": 0.7752181233736415, "grad_norm": 0.9431127645346965, "learning_rate": 3.014672209390602e-06, "loss": 1.1188, "step": 10129 }, { "epoch": 0.7752946578907087, "grad_norm": 0.8979948327902029, "learning_rate": 3.014012379352378e-06, "loss": 1.6434, "step": 10130 }, { "epoch": 0.7753711924077759, "grad_norm": 0.5554919664964348, "learning_rate": 3.01335273278216e-06, "loss": 1.2699, "step": 10131 }, { "epoch": 0.7754477269248431, "grad_norm": 0.5628421291467343, "learning_rate": 3.012693269720477e-06, "loss": 1.2393, "step": 10132 }, { "epoch": 0.7755242614419103, "grad_norm": 0.6040065576891315, "learning_rate": 3.0120339902078517e-06, "loss": 1.1427, "step": 10133 }, { "epoch": 0.7756007959589775, "grad_norm": 0.6194024188891285, "learning_rate": 3.0113748942847897e-06, "loss": 1.5366, "step": 10134 }, { "epoch": 0.7756773304760447, "grad_norm": 0.595734682890595, "learning_rate": 3.0107159819917893e-06, "loss": 1.6625, "step": 10135 }, { "epoch": 0.7757538649931119, "grad_norm": 0.6303936974314605, "learning_rate": 3.010057253369334e-06, "loss": 1.5957, "step": 10136 }, { "epoch": 0.7758303995101791, "grad_norm": 0.6100171718845199, "learning_rate": 3.0093987084579e-06, "loss": 1.2443, "step": 10137 }, { "epoch": 0.7759069340272463, "grad_norm": 0.5181876154199223, "learning_rate": 3.0087403472979502e-06, "loss": 1.1041, "step": 10138 }, { "epoch": 0.7759834685443134, "grad_norm": 0.5505399847026149, "learning_rate": 3.0080821699299347e-06, "loss": 1.1799, "step": 10139 }, { "epoch": 0.7760600030613807, "grad_norm": 0.5931219467968439, "learning_rate": 3.0074241763942953e-06, "loss": 1.4492, "step": 10140 }, { "epoch": 0.7761365375784479, "grad_norm": 0.6256204612330081, "learning_rate": 3.0067663667314607e-06, "loss": 1.5239, "step": 10141 }, { "epoch": 0.776213072095515, "grad_norm": 0.6070994424123013, "learning_rate": 3.006108740981848e-06, "loss": 1.6499, "step": 10142 }, { "epoch": 0.7762896066125823, "grad_norm": 0.7662158100190254, "learning_rate": 3.005451299185863e-06, "loss": 1.4832, "step": 10143 }, { "epoch": 0.7763661411296495, "grad_norm": 0.6218456160393669, "learning_rate": 3.004794041383904e-06, "loss": 1.6131, "step": 10144 }, { "epoch": 0.7764426756467167, "grad_norm": 0.5852666089511559, "learning_rate": 3.004136967616352e-06, "loss": 1.3366, "step": 10145 }, { "epoch": 0.7765192101637839, "grad_norm": 0.5842862477702975, "learning_rate": 3.0034800779235816e-06, "loss": 1.4033, "step": 10146 }, { "epoch": 0.776595744680851, "grad_norm": 0.5296723397327648, "learning_rate": 3.002823372345953e-06, "loss": 1.0569, "step": 10147 }, { "epoch": 0.7766722791979183, "grad_norm": 0.5805461396834735, "learning_rate": 3.002166850923815e-06, "loss": 1.3325, "step": 10148 }, { "epoch": 0.7767488137149855, "grad_norm": 0.5428640796973961, "learning_rate": 3.0015105136975087e-06, "loss": 1.2358, "step": 10149 }, { "epoch": 0.7768253482320526, "grad_norm": 0.6407941074674001, "learning_rate": 3.0008543607073606e-06, "loss": 1.6138, "step": 10150 }, { "epoch": 0.7769018827491199, "grad_norm": 1.2612742129290502, "learning_rate": 3.000198391993685e-06, "loss": 1.4428, "step": 10151 }, { "epoch": 0.7769784172661871, "grad_norm": 0.6951089492751742, "learning_rate": 2.9995426075967893e-06, "loss": 1.4402, "step": 10152 }, { "epoch": 0.7770549517832542, "grad_norm": 0.6277037841195465, "learning_rate": 2.998887007556965e-06, "loss": 1.3725, "step": 10153 }, { "epoch": 0.7771314863003215, "grad_norm": 0.5844739841072443, "learning_rate": 2.9982315919144935e-06, "loss": 1.209, "step": 10154 }, { "epoch": 0.7772080208173886, "grad_norm": 0.6653707156587984, "learning_rate": 2.9975763607096477e-06, "loss": 1.31, "step": 10155 }, { "epoch": 0.7772845553344558, "grad_norm": 0.6394571786604406, "learning_rate": 2.9969213139826855e-06, "loss": 1.3526, "step": 10156 }, { "epoch": 0.7773610898515231, "grad_norm": 0.7273436962359304, "learning_rate": 2.9962664517738548e-06, "loss": 1.537, "step": 10157 }, { "epoch": 0.7774376243685902, "grad_norm": 0.5257942337888992, "learning_rate": 2.9956117741233935e-06, "loss": 1.0506, "step": 10158 }, { "epoch": 0.7775141588856574, "grad_norm": 0.5940040377289766, "learning_rate": 2.9949572810715265e-06, "loss": 1.3315, "step": 10159 }, { "epoch": 0.7775906934027246, "grad_norm": 0.5424526469721177, "learning_rate": 2.994302972658467e-06, "loss": 1.1197, "step": 10160 }, { "epoch": 0.7776672279197918, "grad_norm": 0.5809192945268986, "learning_rate": 2.9936488489244185e-06, "loss": 1.2033, "step": 10161 }, { "epoch": 0.777743762436859, "grad_norm": 0.5816881948128702, "learning_rate": 2.992994909909572e-06, "loss": 1.3648, "step": 10162 }, { "epoch": 0.7778202969539262, "grad_norm": 0.7577052195618458, "learning_rate": 2.9923411556541072e-06, "loss": 1.8422, "step": 10163 }, { "epoch": 0.7778968314709934, "grad_norm": 0.6156390565431478, "learning_rate": 2.991687586198193e-06, "loss": 1.5298, "step": 10164 }, { "epoch": 0.7779733659880607, "grad_norm": 0.6229334489466128, "learning_rate": 2.991034201581986e-06, "loss": 1.3546, "step": 10165 }, { "epoch": 0.7780499005051278, "grad_norm": 0.5840669079597918, "learning_rate": 2.9903810018456334e-06, "loss": 1.4364, "step": 10166 }, { "epoch": 0.778126435022195, "grad_norm": 0.5422155471108152, "learning_rate": 2.9897279870292687e-06, "loss": 1.2036, "step": 10167 }, { "epoch": 0.7782029695392622, "grad_norm": 0.5716749879932117, "learning_rate": 2.9890751571730157e-06, "loss": 1.4259, "step": 10168 }, { "epoch": 0.7782795040563294, "grad_norm": 0.5808076033677917, "learning_rate": 2.9884225123169864e-06, "loss": 1.48, "step": 10169 }, { "epoch": 0.7783560385733966, "grad_norm": 0.6661356301161585, "learning_rate": 2.98777005250128e-06, "loss": 1.2792, "step": 10170 }, { "epoch": 0.7784325730904638, "grad_norm": 0.6314910606293189, "learning_rate": 2.9871177777659877e-06, "loss": 1.7322, "step": 10171 }, { "epoch": 0.778509107607531, "grad_norm": 0.7455758230622146, "learning_rate": 2.986465688151185e-06, "loss": 1.5508, "step": 10172 }, { "epoch": 0.7785856421245982, "grad_norm": 1.2028318266983964, "learning_rate": 2.985813783696939e-06, "loss": 1.4783, "step": 10173 }, { "epoch": 0.7786621766416654, "grad_norm": 0.6750375498150833, "learning_rate": 2.9851620644433055e-06, "loss": 1.4485, "step": 10174 }, { "epoch": 0.7787387111587326, "grad_norm": 0.61933153409059, "learning_rate": 2.9845105304303277e-06, "loss": 1.42, "step": 10175 }, { "epoch": 0.7788152456757997, "grad_norm": 0.5367253020679597, "learning_rate": 2.9838591816980356e-06, "loss": 1.1989, "step": 10176 }, { "epoch": 0.778891780192867, "grad_norm": 0.7060361964424718, "learning_rate": 2.9832080182864536e-06, "loss": 1.4529, "step": 10177 }, { "epoch": 0.7789683147099342, "grad_norm": 0.6484676756161324, "learning_rate": 2.982557040235589e-06, "loss": 1.746, "step": 10178 }, { "epoch": 0.7790448492270013, "grad_norm": 0.559444871997214, "learning_rate": 2.981906247585441e-06, "loss": 1.352, "step": 10179 }, { "epoch": 0.7791213837440686, "grad_norm": 0.5603878343326071, "learning_rate": 2.981255640375995e-06, "loss": 1.1127, "step": 10180 }, { "epoch": 0.7791979182611358, "grad_norm": 0.5870901757092193, "learning_rate": 2.9806052186472267e-06, "loss": 1.2566, "step": 10181 }, { "epoch": 0.779274452778203, "grad_norm": 0.5644392747434712, "learning_rate": 2.9799549824391004e-06, "loss": 1.4172, "step": 10182 }, { "epoch": 0.7793509872952702, "grad_norm": 0.5736997729965102, "learning_rate": 2.9793049317915677e-06, "loss": 1.245, "step": 10183 }, { "epoch": 0.7794275218123373, "grad_norm": 0.6079644561248906, "learning_rate": 2.9786550667445706e-06, "loss": 1.5583, "step": 10184 }, { "epoch": 0.7795040563294046, "grad_norm": 0.5231570001716502, "learning_rate": 2.9780053873380384e-06, "loss": 0.9659, "step": 10185 }, { "epoch": 0.7795805908464718, "grad_norm": 0.623249288643772, "learning_rate": 2.9773558936118886e-06, "loss": 1.3987, "step": 10186 }, { "epoch": 0.7796571253635389, "grad_norm": 0.537593565250769, "learning_rate": 2.976706585606028e-06, "loss": 1.2544, "step": 10187 }, { "epoch": 0.7797336598806062, "grad_norm": 0.6342043289932283, "learning_rate": 2.9760574633603534e-06, "loss": 1.5686, "step": 10188 }, { "epoch": 0.7798101943976733, "grad_norm": 0.48356240211298385, "learning_rate": 2.9754085269147487e-06, "loss": 1.0244, "step": 10189 }, { "epoch": 0.7798867289147405, "grad_norm": 0.559915586304116, "learning_rate": 2.974759776309086e-06, "loss": 1.3954, "step": 10190 }, { "epoch": 0.7799632634318078, "grad_norm": 0.562782306934637, "learning_rate": 2.974111211583226e-06, "loss": 1.3398, "step": 10191 }, { "epoch": 0.7800397979488749, "grad_norm": 0.5722776290291399, "learning_rate": 2.9734628327770175e-06, "loss": 1.3126, "step": 10192 }, { "epoch": 0.7801163324659421, "grad_norm": 0.522405320219002, "learning_rate": 2.9728146399303014e-06, "loss": 1.2596, "step": 10193 }, { "epoch": 0.7801928669830094, "grad_norm": 0.601328553122837, "learning_rate": 2.9721666330829043e-06, "loss": 1.4428, "step": 10194 }, { "epoch": 0.7802694015000765, "grad_norm": 0.574359056948717, "learning_rate": 2.9715188122746397e-06, "loss": 1.3536, "step": 10195 }, { "epoch": 0.7803459360171437, "grad_norm": 0.6212603950229162, "learning_rate": 2.970871177545313e-06, "loss": 1.5393, "step": 10196 }, { "epoch": 0.7804224705342109, "grad_norm": 0.4272070217545985, "learning_rate": 2.970223728934717e-06, "loss": 0.7206, "step": 10197 }, { "epoch": 0.7804990050512781, "grad_norm": 0.6512327388857981, "learning_rate": 2.9695764664826308e-06, "loss": 1.6106, "step": 10198 }, { "epoch": 0.7805755395683454, "grad_norm": 0.509726006255748, "learning_rate": 2.968929390228827e-06, "loss": 1.118, "step": 10199 }, { "epoch": 0.7806520740854125, "grad_norm": 0.5471805492709368, "learning_rate": 2.9682825002130623e-06, "loss": 1.2664, "step": 10200 }, { "epoch": 0.7807286086024797, "grad_norm": 0.5945742886880312, "learning_rate": 2.9676357964750847e-06, "loss": 1.4146, "step": 10201 }, { "epoch": 0.780805143119547, "grad_norm": 0.5770495568801963, "learning_rate": 2.966989279054628e-06, "loss": 1.3731, "step": 10202 }, { "epoch": 0.7808816776366141, "grad_norm": 0.6010300038430926, "learning_rate": 2.9663429479914173e-06, "loss": 1.2868, "step": 10203 }, { "epoch": 0.7809582121536813, "grad_norm": 0.496678316648092, "learning_rate": 2.9656968033251643e-06, "loss": 1.0277, "step": 10204 }, { "epoch": 0.7810347466707485, "grad_norm": 0.8656523364068037, "learning_rate": 2.9650508450955712e-06, "loss": 1.6645, "step": 10205 }, { "epoch": 0.7811112811878157, "grad_norm": 0.6857138660155768, "learning_rate": 2.9644050733423267e-06, "loss": 1.4574, "step": 10206 }, { "epoch": 0.7811878157048829, "grad_norm": 3.647557338869193, "learning_rate": 2.963759488105109e-06, "loss": 1.4698, "step": 10207 }, { "epoch": 0.7812643502219501, "grad_norm": 1.1119137697112003, "learning_rate": 2.9631140894235853e-06, "loss": 1.5499, "step": 10208 }, { "epoch": 0.7813408847390173, "grad_norm": 0.532410016291953, "learning_rate": 2.96246887733741e-06, "loss": 1.1405, "step": 10209 }, { "epoch": 0.7814174192560845, "grad_norm": 0.5346553171640156, "learning_rate": 2.9618238518862273e-06, "loss": 1.2137, "step": 10210 }, { "epoch": 0.7814939537731517, "grad_norm": 0.5634667755230234, "learning_rate": 2.9611790131096706e-06, "loss": 1.3676, "step": 10211 }, { "epoch": 0.7815704882902189, "grad_norm": 0.5299589803173562, "learning_rate": 2.960534361047359e-06, "loss": 1.1807, "step": 10212 }, { "epoch": 0.781647022807286, "grad_norm": 0.6030417184925227, "learning_rate": 2.959889895738902e-06, "loss": 1.5039, "step": 10213 }, { "epoch": 0.7817235573243533, "grad_norm": 0.5961297581063111, "learning_rate": 2.959245617223898e-06, "loss": 1.254, "step": 10214 }, { "epoch": 0.7818000918414205, "grad_norm": 0.5868555501412551, "learning_rate": 2.958601525541934e-06, "loss": 1.3358, "step": 10215 }, { "epoch": 0.7818766263584876, "grad_norm": 0.713782403586312, "learning_rate": 2.9579576207325843e-06, "loss": 1.4187, "step": 10216 }, { "epoch": 0.7819531608755549, "grad_norm": 0.8066983291360081, "learning_rate": 2.9573139028354124e-06, "loss": 1.6439, "step": 10217 }, { "epoch": 0.7820296953926221, "grad_norm": 0.5517429061383173, "learning_rate": 2.95667037188997e-06, "loss": 1.2, "step": 10218 }, { "epoch": 0.7821062299096893, "grad_norm": 0.9091655419763013, "learning_rate": 2.9560270279357982e-06, "loss": 1.3742, "step": 10219 }, { "epoch": 0.7821827644267565, "grad_norm": 0.5661896379665826, "learning_rate": 2.9553838710124244e-06, "loss": 1.3937, "step": 10220 }, { "epoch": 0.7822592989438236, "grad_norm": 0.6021807475173366, "learning_rate": 2.9547409011593687e-06, "loss": 1.1992, "step": 10221 }, { "epoch": 0.7823358334608909, "grad_norm": 0.574022095057274, "learning_rate": 2.9540981184161345e-06, "loss": 1.4445, "step": 10222 }, { "epoch": 0.7824123679779581, "grad_norm": 0.494249881174051, "learning_rate": 2.9534555228222183e-06, "loss": 0.9637, "step": 10223 }, { "epoch": 0.7824889024950252, "grad_norm": 0.7125088857111023, "learning_rate": 2.9528131144171013e-06, "loss": 1.4997, "step": 10224 }, { "epoch": 0.7825654370120925, "grad_norm": 0.5868824023569881, "learning_rate": 2.952170893240257e-06, "loss": 1.4295, "step": 10225 }, { "epoch": 0.7826419715291596, "grad_norm": 0.6783993746343261, "learning_rate": 2.9515288593311444e-06, "loss": 1.479, "step": 10226 }, { "epoch": 0.7827185060462268, "grad_norm": 0.7094178317429376, "learning_rate": 2.950887012729211e-06, "loss": 1.7055, "step": 10227 }, { "epoch": 0.7827950405632941, "grad_norm": 0.5220889238853827, "learning_rate": 2.9502453534738953e-06, "loss": 1.2451, "step": 10228 }, { "epoch": 0.7828715750803612, "grad_norm": 0.5217600115806326, "learning_rate": 2.949603881604621e-06, "loss": 1.254, "step": 10229 }, { "epoch": 0.7829481095974284, "grad_norm": 0.5424949611740885, "learning_rate": 2.9489625971608043e-06, "loss": 1.3258, "step": 10230 }, { "epoch": 0.7830246441144957, "grad_norm": 0.8565000626919074, "learning_rate": 2.948321500181845e-06, "loss": 1.5082, "step": 10231 }, { "epoch": 0.7831011786315628, "grad_norm": 0.5113477512027663, "learning_rate": 2.9476805907071373e-06, "loss": 1.1155, "step": 10232 }, { "epoch": 0.78317771314863, "grad_norm": 0.5496463833917817, "learning_rate": 2.9470398687760576e-06, "loss": 1.3173, "step": 10233 }, { "epoch": 0.7832542476656972, "grad_norm": 0.5865956570853141, "learning_rate": 2.9463993344279755e-06, "loss": 1.3319, "step": 10234 }, { "epoch": 0.7833307821827644, "grad_norm": 0.6598155697102702, "learning_rate": 2.945758987702247e-06, "loss": 1.7315, "step": 10235 }, { "epoch": 0.7834073166998317, "grad_norm": 0.5489655341291458, "learning_rate": 2.9451188286382153e-06, "loss": 1.2798, "step": 10236 }, { "epoch": 0.7834838512168988, "grad_norm": 0.6257673343703135, "learning_rate": 2.9444788572752165e-06, "loss": 1.387, "step": 10237 }, { "epoch": 0.783560385733966, "grad_norm": 0.5782599490268745, "learning_rate": 2.9438390736525714e-06, "loss": 1.23, "step": 10238 }, { "epoch": 0.7836369202510333, "grad_norm": 0.5421644422269292, "learning_rate": 2.9431994778095883e-06, "loss": 1.2102, "step": 10239 }, { "epoch": 0.7837134547681004, "grad_norm": 0.481348670234651, "learning_rate": 2.9425600697855687e-06, "loss": 0.8873, "step": 10240 }, { "epoch": 0.7837899892851676, "grad_norm": 0.5231821697396299, "learning_rate": 2.941920849619798e-06, "loss": 1.1601, "step": 10241 }, { "epoch": 0.7838665238022348, "grad_norm": 0.5792228746027785, "learning_rate": 2.941281817351551e-06, "loss": 1.3633, "step": 10242 }, { "epoch": 0.783943058319302, "grad_norm": 0.6127148508277026, "learning_rate": 2.9406429730200948e-06, "loss": 1.5207, "step": 10243 }, { "epoch": 0.7840195928363692, "grad_norm": 0.5301494536926459, "learning_rate": 2.9400043166646797e-06, "loss": 1.167, "step": 10244 }, { "epoch": 0.7840961273534364, "grad_norm": 0.5012632736504561, "learning_rate": 2.939365848324547e-06, "loss": 1.0147, "step": 10245 }, { "epoch": 0.7841726618705036, "grad_norm": 0.5091158969491384, "learning_rate": 2.938727568038926e-06, "loss": 1.1398, "step": 10246 }, { "epoch": 0.7842491963875708, "grad_norm": 0.6638989385099989, "learning_rate": 2.9380894758470346e-06, "loss": 1.4104, "step": 10247 }, { "epoch": 0.784325730904638, "grad_norm": 0.6332303759450693, "learning_rate": 2.937451571788079e-06, "loss": 1.5951, "step": 10248 }, { "epoch": 0.7844022654217052, "grad_norm": 0.5407595712146145, "learning_rate": 2.936813855901255e-06, "loss": 1.1921, "step": 10249 }, { "epoch": 0.7844787999387723, "grad_norm": 0.5713847405518703, "learning_rate": 2.9361763282257436e-06, "loss": 1.3211, "step": 10250 }, { "epoch": 0.7845553344558396, "grad_norm": 0.5848068334396108, "learning_rate": 2.9355389888007185e-06, "loss": 1.335, "step": 10251 }, { "epoch": 0.7846318689729068, "grad_norm": 0.6458765956000228, "learning_rate": 2.934901837665339e-06, "loss": 1.5841, "step": 10252 }, { "epoch": 0.784708403489974, "grad_norm": 0.5134490901290752, "learning_rate": 2.934264874858752e-06, "loss": 1.0429, "step": 10253 }, { "epoch": 0.7847849380070412, "grad_norm": 1.2012553677564135, "learning_rate": 2.9336281004200975e-06, "loss": 1.4827, "step": 10254 }, { "epoch": 0.7848614725241084, "grad_norm": 1.0843922613010593, "learning_rate": 2.9329915143884985e-06, "loss": 1.5897, "step": 10255 }, { "epoch": 0.7849380070411756, "grad_norm": 0.69912398360167, "learning_rate": 2.9323551168030697e-06, "loss": 1.5085, "step": 10256 }, { "epoch": 0.7850145415582428, "grad_norm": 0.6738608237499979, "learning_rate": 2.9317189077029136e-06, "loss": 1.7107, "step": 10257 }, { "epoch": 0.7850910760753099, "grad_norm": 0.5205157542141836, "learning_rate": 2.9310828871271185e-06, "loss": 1.0885, "step": 10258 }, { "epoch": 0.7851676105923772, "grad_norm": 0.5980240784195489, "learning_rate": 2.9304470551147667e-06, "loss": 1.579, "step": 10259 }, { "epoch": 0.7852441451094444, "grad_norm": 0.6003919197654876, "learning_rate": 2.9298114117049236e-06, "loss": 1.131, "step": 10260 }, { "epoch": 0.7853206796265115, "grad_norm": 0.6010005231154103, "learning_rate": 2.929175956936645e-06, "loss": 1.3874, "step": 10261 }, { "epoch": 0.7853972141435788, "grad_norm": 0.5936134069537625, "learning_rate": 2.9285406908489765e-06, "loss": 1.4512, "step": 10262 }, { "epoch": 0.7854737486606459, "grad_norm": 0.7440585152106806, "learning_rate": 2.9279056134809503e-06, "loss": 1.5959, "step": 10263 }, { "epoch": 0.7855502831777131, "grad_norm": 0.8946664671223855, "learning_rate": 2.9272707248715846e-06, "loss": 1.4973, "step": 10264 }, { "epoch": 0.7856268176947804, "grad_norm": 0.6628246786786072, "learning_rate": 2.9266360250598935e-06, "loss": 1.4589, "step": 10265 }, { "epoch": 0.7857033522118475, "grad_norm": 0.5407203156878413, "learning_rate": 2.9260015140848725e-06, "loss": 1.3354, "step": 10266 }, { "epoch": 0.7857798867289147, "grad_norm": 0.5448082826873537, "learning_rate": 2.9253671919855077e-06, "loss": 1.2376, "step": 10267 }, { "epoch": 0.785856421245982, "grad_norm": 0.4837100968296632, "learning_rate": 2.924733058800774e-06, "loss": 1.0107, "step": 10268 }, { "epoch": 0.7859329557630491, "grad_norm": 0.537531957694371, "learning_rate": 2.924099114569634e-06, "loss": 1.4008, "step": 10269 }, { "epoch": 0.7860094902801164, "grad_norm": 0.601132972508668, "learning_rate": 2.9234653593310402e-06, "loss": 1.4742, "step": 10270 }, { "epoch": 0.7860860247971835, "grad_norm": 0.7400433080277915, "learning_rate": 2.9228317931239313e-06, "loss": 1.3582, "step": 10271 }, { "epoch": 0.7861625593142507, "grad_norm": 0.8508350910511974, "learning_rate": 2.9221984159872364e-06, "loss": 1.3841, "step": 10272 }, { "epoch": 0.786239093831318, "grad_norm": 0.5799360138109355, "learning_rate": 2.921565227959872e-06, "loss": 1.3043, "step": 10273 }, { "epoch": 0.7863156283483851, "grad_norm": 1.668807695577914, "learning_rate": 2.9209322290807428e-06, "loss": 1.701, "step": 10274 }, { "epoch": 0.7863921628654523, "grad_norm": 0.516039546605834, "learning_rate": 2.92029941938874e-06, "loss": 1.0602, "step": 10275 }, { "epoch": 0.7864686973825196, "grad_norm": 0.6438927412815034, "learning_rate": 2.919666798922749e-06, "loss": 1.4935, "step": 10276 }, { "epoch": 0.7865452318995867, "grad_norm": 0.8686928233137974, "learning_rate": 2.9190343677216392e-06, "loss": 1.2886, "step": 10277 }, { "epoch": 0.7866217664166539, "grad_norm": 0.5456143480379266, "learning_rate": 2.9184021258242667e-06, "loss": 1.3667, "step": 10278 }, { "epoch": 0.7866983009337211, "grad_norm": 0.9174473267435536, "learning_rate": 2.9177700732694814e-06, "loss": 1.3727, "step": 10279 }, { "epoch": 0.7867748354507883, "grad_norm": 0.6196046355575299, "learning_rate": 2.9171382100961144e-06, "loss": 1.4001, "step": 10280 }, { "epoch": 0.7868513699678555, "grad_norm": 0.5483499314907424, "learning_rate": 2.9165065363429932e-06, "loss": 1.2457, "step": 10281 }, { "epoch": 0.7869279044849227, "grad_norm": 0.7870795509190702, "learning_rate": 2.915875052048929e-06, "loss": 1.4503, "step": 10282 }, { "epoch": 0.7870044390019899, "grad_norm": 0.6252409373896116, "learning_rate": 2.915243757252721e-06, "loss": 1.2298, "step": 10283 }, { "epoch": 0.7870809735190571, "grad_norm": 0.7372355693559548, "learning_rate": 2.9146126519931583e-06, "loss": 1.6817, "step": 10284 }, { "epoch": 0.7871575080361243, "grad_norm": 0.7603860061187399, "learning_rate": 2.913981736309016e-06, "loss": 1.6065, "step": 10285 }, { "epoch": 0.7872340425531915, "grad_norm": 0.6128465103891453, "learning_rate": 2.913351010239064e-06, "loss": 1.5001, "step": 10286 }, { "epoch": 0.7873105770702586, "grad_norm": 0.7458326476777914, "learning_rate": 2.9127204738220515e-06, "loss": 1.4738, "step": 10287 }, { "epoch": 0.7873871115873259, "grad_norm": 0.5673805777940698, "learning_rate": 2.9120901270967226e-06, "loss": 1.2775, "step": 10288 }, { "epoch": 0.7874636461043931, "grad_norm": 0.6037454476987347, "learning_rate": 2.911459970101808e-06, "loss": 1.3857, "step": 10289 }, { "epoch": 0.7875401806214603, "grad_norm": 0.670913373443118, "learning_rate": 2.9108300028760255e-06, "loss": 1.5362, "step": 10290 }, { "epoch": 0.7876167151385275, "grad_norm": 0.6035628281319105, "learning_rate": 2.910200225458083e-06, "loss": 1.4429, "step": 10291 }, { "epoch": 0.7876932496555946, "grad_norm": 0.6199908472859839, "learning_rate": 2.909570637886675e-06, "loss": 1.5387, "step": 10292 }, { "epoch": 0.7877697841726619, "grad_norm": 0.6094917138726399, "learning_rate": 2.908941240200486e-06, "loss": 1.5918, "step": 10293 }, { "epoch": 0.7878463186897291, "grad_norm": 0.6712186055475502, "learning_rate": 2.9083120324381864e-06, "loss": 1.5531, "step": 10294 }, { "epoch": 0.7879228532067962, "grad_norm": 0.6402119845232694, "learning_rate": 2.9076830146384394e-06, "loss": 1.4411, "step": 10295 }, { "epoch": 0.7879993877238635, "grad_norm": 0.5471362299839346, "learning_rate": 2.9070541868398905e-06, "loss": 1.1633, "step": 10296 }, { "epoch": 0.7880759222409307, "grad_norm": 0.5593098794314236, "learning_rate": 2.9064255490811793e-06, "loss": 1.3062, "step": 10297 }, { "epoch": 0.7881524567579978, "grad_norm": 0.5330050861961299, "learning_rate": 2.9057971014009314e-06, "loss": 1.2642, "step": 10298 }, { "epoch": 0.7882289912750651, "grad_norm": 0.7216659231593265, "learning_rate": 2.905168843837758e-06, "loss": 1.3801, "step": 10299 }, { "epoch": 0.7883055257921322, "grad_norm": 0.5871483330333629, "learning_rate": 2.9045407764302635e-06, "loss": 1.3096, "step": 10300 }, { "epoch": 0.7883820603091994, "grad_norm": 0.7951843004773124, "learning_rate": 2.9039128992170375e-06, "loss": 1.4855, "step": 10301 }, { "epoch": 0.7884585948262667, "grad_norm": 0.5630630146197679, "learning_rate": 2.903285212236657e-06, "loss": 1.3013, "step": 10302 }, { "epoch": 0.7885351293433338, "grad_norm": 0.9168925607272204, "learning_rate": 2.9026577155276924e-06, "loss": 1.4057, "step": 10303 }, { "epoch": 0.788611663860401, "grad_norm": 0.7050218370268339, "learning_rate": 2.9020304091286965e-06, "loss": 1.7204, "step": 10304 }, { "epoch": 0.7886881983774683, "grad_norm": 0.7288021651058397, "learning_rate": 2.901403293078213e-06, "loss": 1.6686, "step": 10305 }, { "epoch": 0.7887647328945354, "grad_norm": 0.5877015379859422, "learning_rate": 2.900776367414775e-06, "loss": 1.4275, "step": 10306 }, { "epoch": 0.7888412674116027, "grad_norm": 0.602475492839941, "learning_rate": 2.9001496321769003e-06, "loss": 1.4565, "step": 10307 }, { "epoch": 0.7889178019286698, "grad_norm": 0.48023907223489143, "learning_rate": 2.8995230874031e-06, "loss": 1.0644, "step": 10308 }, { "epoch": 0.788994336445737, "grad_norm": 0.8045501401077547, "learning_rate": 2.89889673313187e-06, "loss": 1.5325, "step": 10309 }, { "epoch": 0.7890708709628043, "grad_norm": 0.6924876103540202, "learning_rate": 2.8982705694016953e-06, "loss": 1.5419, "step": 10310 }, { "epoch": 0.7891474054798714, "grad_norm": 0.5502458990405796, "learning_rate": 2.897644596251049e-06, "loss": 1.3373, "step": 10311 }, { "epoch": 0.7892239399969386, "grad_norm": 0.5492257129309441, "learning_rate": 2.897018813718394e-06, "loss": 1.19, "step": 10312 }, { "epoch": 0.7893004745140059, "grad_norm": 0.553052246576808, "learning_rate": 2.896393221842179e-06, "loss": 1.2674, "step": 10313 }, { "epoch": 0.789377009031073, "grad_norm": 0.47053903883393966, "learning_rate": 2.8957678206608407e-06, "loss": 1.0401, "step": 10314 }, { "epoch": 0.7894535435481402, "grad_norm": 0.5678120906279412, "learning_rate": 2.8951426102128088e-06, "loss": 1.3056, "step": 10315 }, { "epoch": 0.7895300780652074, "grad_norm": 0.6576695241007067, "learning_rate": 2.8945175905364963e-06, "loss": 1.3229, "step": 10316 }, { "epoch": 0.7896066125822746, "grad_norm": 0.5165283359152593, "learning_rate": 2.8938927616703074e-06, "loss": 1.2317, "step": 10317 }, { "epoch": 0.7896831470993418, "grad_norm": 0.5797942548051245, "learning_rate": 2.8932681236526304e-06, "loss": 1.5239, "step": 10318 }, { "epoch": 0.789759681616409, "grad_norm": 1.2412163772467226, "learning_rate": 2.892643676521849e-06, "loss": 1.1187, "step": 10319 }, { "epoch": 0.7898362161334762, "grad_norm": 0.9289486207793713, "learning_rate": 2.8920194203163297e-06, "loss": 1.6159, "step": 10320 }, { "epoch": 0.7899127506505434, "grad_norm": 0.7454507428105679, "learning_rate": 2.8913953550744285e-06, "loss": 1.5909, "step": 10321 }, { "epoch": 0.7899892851676106, "grad_norm": 0.6137816020089071, "learning_rate": 2.890771480834489e-06, "loss": 1.3489, "step": 10322 }, { "epoch": 0.7900658196846778, "grad_norm": 0.5476673402310767, "learning_rate": 2.8901477976348446e-06, "loss": 1.2147, "step": 10323 }, { "epoch": 0.790142354201745, "grad_norm": 0.6592689223894577, "learning_rate": 2.889524305513815e-06, "loss": 1.47, "step": 10324 }, { "epoch": 0.7902188887188122, "grad_norm": 0.6008648125590972, "learning_rate": 2.888901004509712e-06, "loss": 1.5466, "step": 10325 }, { "epoch": 0.7902954232358794, "grad_norm": 0.5547102113030984, "learning_rate": 2.8882778946608314e-06, "loss": 1.2179, "step": 10326 }, { "epoch": 0.7903719577529466, "grad_norm": 0.9296864675115398, "learning_rate": 2.8876549760054596e-06, "loss": 1.2433, "step": 10327 }, { "epoch": 0.7904484922700138, "grad_norm": 1.0067566124948257, "learning_rate": 2.8870322485818703e-06, "loss": 1.4147, "step": 10328 }, { "epoch": 0.7905250267870809, "grad_norm": 0.561996892144843, "learning_rate": 2.8864097124283243e-06, "loss": 1.1504, "step": 10329 }, { "epoch": 0.7906015613041482, "grad_norm": 0.6081308471536176, "learning_rate": 2.8857873675830738e-06, "loss": 1.3193, "step": 10330 }, { "epoch": 0.7906780958212154, "grad_norm": 0.5701195813922602, "learning_rate": 2.885165214084358e-06, "loss": 1.2684, "step": 10331 }, { "epoch": 0.7907546303382825, "grad_norm": 0.563592476593637, "learning_rate": 2.8845432519704033e-06, "loss": 1.3996, "step": 10332 }, { "epoch": 0.7908311648553498, "grad_norm": 0.5851589797625519, "learning_rate": 2.883921481279424e-06, "loss": 1.3746, "step": 10333 }, { "epoch": 0.790907699372417, "grad_norm": 0.7440434725035617, "learning_rate": 2.8832999020496245e-06, "loss": 1.6133, "step": 10334 }, { "epoch": 0.7909842338894841, "grad_norm": 0.5483663852331035, "learning_rate": 2.882678514319196e-06, "loss": 1.1877, "step": 10335 }, { "epoch": 0.7910607684065514, "grad_norm": 0.8544045394264628, "learning_rate": 2.8820573181263187e-06, "loss": 1.1983, "step": 10336 }, { "epoch": 0.7911373029236185, "grad_norm": 0.6056717194966332, "learning_rate": 2.8814363135091604e-06, "loss": 1.2137, "step": 10337 }, { "epoch": 0.7912138374406857, "grad_norm": 0.5653826833308712, "learning_rate": 2.880815500505878e-06, "loss": 1.3439, "step": 10338 }, { "epoch": 0.791290371957753, "grad_norm": 0.5533913169428893, "learning_rate": 2.8801948791546163e-06, "loss": 1.2651, "step": 10339 }, { "epoch": 0.7913669064748201, "grad_norm": 0.8157519550108385, "learning_rate": 2.8795744494935063e-06, "loss": 1.1909, "step": 10340 }, { "epoch": 0.7914434409918873, "grad_norm": 0.5870551224899566, "learning_rate": 2.8789542115606705e-06, "loss": 1.179, "step": 10341 }, { "epoch": 0.7915199755089546, "grad_norm": 3.0934155667693988, "learning_rate": 2.8783341653942187e-06, "loss": 1.801, "step": 10342 }, { "epoch": 0.7915965100260217, "grad_norm": 0.54510467349443, "learning_rate": 2.8777143110322486e-06, "loss": 1.1874, "step": 10343 }, { "epoch": 0.791673044543089, "grad_norm": 0.5913685020745243, "learning_rate": 2.877094648512845e-06, "loss": 1.4478, "step": 10344 }, { "epoch": 0.7917495790601561, "grad_norm": 0.6901197311279588, "learning_rate": 2.8764751778740808e-06, "loss": 1.4881, "step": 10345 }, { "epoch": 0.7918261135772233, "grad_norm": 0.5901923351223545, "learning_rate": 2.8758558991540186e-06, "loss": 1.3337, "step": 10346 }, { "epoch": 0.7919026480942906, "grad_norm": 0.5379000648593086, "learning_rate": 2.87523681239071e-06, "loss": 1.2284, "step": 10347 }, { "epoch": 0.7919791826113577, "grad_norm": 0.6867093279903195, "learning_rate": 2.8746179176221945e-06, "loss": 1.5423, "step": 10348 }, { "epoch": 0.7920557171284249, "grad_norm": 0.6970080924905315, "learning_rate": 2.873999214886496e-06, "loss": 1.841, "step": 10349 }, { "epoch": 0.7921322516454922, "grad_norm": 0.5992295001159688, "learning_rate": 2.873380704221631e-06, "loss": 1.4111, "step": 10350 }, { "epoch": 0.7922087861625593, "grad_norm": 0.535049544817717, "learning_rate": 2.8727623856656007e-06, "loss": 1.1625, "step": 10351 }, { "epoch": 0.7922853206796265, "grad_norm": 0.7327136461627958, "learning_rate": 2.8721442592563996e-06, "loss": 1.4789, "step": 10352 }, { "epoch": 0.7923618551966937, "grad_norm": 0.5146444147309892, "learning_rate": 2.8715263250320056e-06, "loss": 1.0472, "step": 10353 }, { "epoch": 0.7924383897137609, "grad_norm": 0.7200787696563441, "learning_rate": 2.8709085830303875e-06, "loss": 1.305, "step": 10354 }, { "epoch": 0.7925149242308281, "grad_norm": 0.5909603429989303, "learning_rate": 2.8702910332894996e-06, "loss": 1.475, "step": 10355 }, { "epoch": 0.7925914587478953, "grad_norm": 0.7303574869850511, "learning_rate": 2.8696736758472864e-06, "loss": 1.72, "step": 10356 }, { "epoch": 0.7926679932649625, "grad_norm": 0.5713072799467982, "learning_rate": 2.869056510741679e-06, "loss": 1.3076, "step": 10357 }, { "epoch": 0.7927445277820298, "grad_norm": 0.5970467226652179, "learning_rate": 2.868439538010601e-06, "loss": 1.3858, "step": 10358 }, { "epoch": 0.7928210622990969, "grad_norm": 0.6912285582716481, "learning_rate": 2.867822757691959e-06, "loss": 1.3925, "step": 10359 }, { "epoch": 0.7928975968161641, "grad_norm": 0.5217850545180109, "learning_rate": 2.8672061698236507e-06, "loss": 1.2037, "step": 10360 }, { "epoch": 0.7929741313332312, "grad_norm": 0.9800067426323699, "learning_rate": 2.8665897744435617e-06, "loss": 1.5604, "step": 10361 }, { "epoch": 0.7930506658502985, "grad_norm": 0.5744091710315861, "learning_rate": 2.865973571589562e-06, "loss": 1.2587, "step": 10362 }, { "epoch": 0.7931272003673657, "grad_norm": 0.5908566908480523, "learning_rate": 2.8653575612995153e-06, "loss": 1.5171, "step": 10363 }, { "epoch": 0.7932037348844329, "grad_norm": 0.8615532367653758, "learning_rate": 2.8647417436112717e-06, "loss": 1.322, "step": 10364 }, { "epoch": 0.7932802694015001, "grad_norm": 0.6590949773509768, "learning_rate": 2.864126118562668e-06, "loss": 1.4955, "step": 10365 }, { "epoch": 0.7933568039185672, "grad_norm": 0.6454277128480381, "learning_rate": 2.8635106861915296e-06, "loss": 1.5602, "step": 10366 }, { "epoch": 0.7934333384356345, "grad_norm": 0.5238662144105377, "learning_rate": 2.8628954465356708e-06, "loss": 1.238, "step": 10367 }, { "epoch": 0.7935098729527017, "grad_norm": 0.5977385601835918, "learning_rate": 2.862280399632893e-06, "loss": 1.346, "step": 10368 }, { "epoch": 0.7935864074697688, "grad_norm": 0.5953654760511077, "learning_rate": 2.8616655455209887e-06, "loss": 1.3929, "step": 10369 }, { "epoch": 0.7936629419868361, "grad_norm": 0.618753544330333, "learning_rate": 2.8610508842377348e-06, "loss": 1.1861, "step": 10370 }, { "epoch": 0.7937394765039033, "grad_norm": 0.6636365380744476, "learning_rate": 2.8604364158208984e-06, "loss": 1.5054, "step": 10371 }, { "epoch": 0.7938160110209704, "grad_norm": 0.6864973796679844, "learning_rate": 2.8598221403082344e-06, "loss": 1.5634, "step": 10372 }, { "epoch": 0.7938925455380377, "grad_norm": 0.5242587857397768, "learning_rate": 2.8592080577374836e-06, "loss": 1.132, "step": 10373 }, { "epoch": 0.7939690800551048, "grad_norm": 0.5764522748303317, "learning_rate": 2.8585941681463813e-06, "loss": 1.2678, "step": 10374 }, { "epoch": 0.794045614572172, "grad_norm": 0.9750593964218823, "learning_rate": 2.8579804715726434e-06, "loss": 1.889, "step": 10375 }, { "epoch": 0.7941221490892393, "grad_norm": 0.5065561220325543, "learning_rate": 2.857366968053978e-06, "loss": 1.0654, "step": 10376 }, { "epoch": 0.7941986836063064, "grad_norm": 0.5921209574678825, "learning_rate": 2.8567536576280815e-06, "loss": 1.3802, "step": 10377 }, { "epoch": 0.7942752181233736, "grad_norm": 0.8206741774179586, "learning_rate": 2.856140540332636e-06, "loss": 1.5397, "step": 10378 }, { "epoch": 0.7943517526404409, "grad_norm": 0.5428837949076442, "learning_rate": 2.8555276162053134e-06, "loss": 1.234, "step": 10379 }, { "epoch": 0.794428287157508, "grad_norm": 0.9743191946208072, "learning_rate": 2.8549148852837747e-06, "loss": 1.5307, "step": 10380 }, { "epoch": 0.7945048216745753, "grad_norm": 1.142042195688125, "learning_rate": 2.8543023476056684e-06, "loss": 1.5799, "step": 10381 }, { "epoch": 0.7945813561916424, "grad_norm": 0.634407520741652, "learning_rate": 2.8536900032086303e-06, "loss": 1.6042, "step": 10382 }, { "epoch": 0.7946578907087096, "grad_norm": 0.6662408152488217, "learning_rate": 2.8530778521302825e-06, "loss": 1.5644, "step": 10383 }, { "epoch": 0.7947344252257769, "grad_norm": 0.5718679774809023, "learning_rate": 2.8524658944082393e-06, "loss": 1.4098, "step": 10384 }, { "epoch": 0.794810959742844, "grad_norm": 0.5801666860511148, "learning_rate": 2.851854130080102e-06, "loss": 1.2306, "step": 10385 }, { "epoch": 0.7948874942599112, "grad_norm": 0.6344143859886597, "learning_rate": 2.851242559183457e-06, "loss": 1.4771, "step": 10386 }, { "epoch": 0.7949640287769785, "grad_norm": 0.6161419244251523, "learning_rate": 2.8506311817558822e-06, "loss": 1.3244, "step": 10387 }, { "epoch": 0.7950405632940456, "grad_norm": 0.5735201587995772, "learning_rate": 2.850019997834943e-06, "loss": 1.199, "step": 10388 }, { "epoch": 0.7951170978111128, "grad_norm": 0.6432645020304397, "learning_rate": 2.8494090074581914e-06, "loss": 1.5953, "step": 10389 }, { "epoch": 0.79519363232818, "grad_norm": 0.5800279709137839, "learning_rate": 2.8487982106631673e-06, "loss": 1.4311, "step": 10390 }, { "epoch": 0.7952701668452472, "grad_norm": 0.6444928493016899, "learning_rate": 2.8481876074874037e-06, "loss": 1.3684, "step": 10391 }, { "epoch": 0.7953467013623144, "grad_norm": 1.853353075191783, "learning_rate": 2.8475771979684143e-06, "loss": 1.332, "step": 10392 }, { "epoch": 0.7954232358793816, "grad_norm": 0.5807820899045788, "learning_rate": 2.8469669821437062e-06, "loss": 1.3691, "step": 10393 }, { "epoch": 0.7954997703964488, "grad_norm": 0.5601452052251329, "learning_rate": 2.8463569600507734e-06, "loss": 1.2644, "step": 10394 }, { "epoch": 0.7955763049135159, "grad_norm": 0.7540021787317936, "learning_rate": 2.8457471317270945e-06, "loss": 1.1261, "step": 10395 }, { "epoch": 0.7956528394305832, "grad_norm": 0.6112882176746038, "learning_rate": 2.845137497210143e-06, "loss": 1.4048, "step": 10396 }, { "epoch": 0.7957293739476504, "grad_norm": 0.576395641205161, "learning_rate": 2.8445280565373744e-06, "loss": 1.3535, "step": 10397 }, { "epoch": 0.7958059084647175, "grad_norm": 0.6553969769781204, "learning_rate": 2.8439188097462355e-06, "loss": 1.3508, "step": 10398 }, { "epoch": 0.7958824429817848, "grad_norm": 1.2818835594866773, "learning_rate": 2.84330975687416e-06, "loss": 1.7163, "step": 10399 }, { "epoch": 0.795958977498852, "grad_norm": 1.1109065410155994, "learning_rate": 2.842700897958569e-06, "loss": 1.7033, "step": 10400 }, { "epoch": 0.7960355120159192, "grad_norm": 0.5628165013586597, "learning_rate": 2.8420922330368726e-06, "loss": 1.2337, "step": 10401 }, { "epoch": 0.7961120465329864, "grad_norm": 0.6452078892115535, "learning_rate": 2.8414837621464706e-06, "loss": 1.4305, "step": 10402 }, { "epoch": 0.7961885810500535, "grad_norm": 1.021798217812575, "learning_rate": 2.8408754853247487e-06, "loss": 1.6767, "step": 10403 }, { "epoch": 0.7962651155671208, "grad_norm": 0.5536866025954595, "learning_rate": 2.8402674026090813e-06, "loss": 1.4258, "step": 10404 }, { "epoch": 0.796341650084188, "grad_norm": 0.5349925073564943, "learning_rate": 2.83965951403683e-06, "loss": 1.2824, "step": 10405 }, { "epoch": 0.7964181846012551, "grad_norm": 0.630353732833249, "learning_rate": 2.839051819645346e-06, "loss": 1.5995, "step": 10406 }, { "epoch": 0.7964947191183224, "grad_norm": 0.6826576019790728, "learning_rate": 2.838444319471967e-06, "loss": 1.5775, "step": 10407 }, { "epoch": 0.7965712536353896, "grad_norm": 0.8600233065595302, "learning_rate": 2.837837013554021e-06, "loss": 1.2827, "step": 10408 }, { "epoch": 0.7966477881524567, "grad_norm": 0.5794671129961512, "learning_rate": 2.8372299019288217e-06, "loss": 1.3218, "step": 10409 }, { "epoch": 0.796724322669524, "grad_norm": 0.5929582584690546, "learning_rate": 2.8366229846336726e-06, "loss": 1.2816, "step": 10410 }, { "epoch": 0.7968008571865911, "grad_norm": 0.4548910483604133, "learning_rate": 2.8360162617058627e-06, "loss": 0.865, "step": 10411 }, { "epoch": 0.7968773917036583, "grad_norm": 0.6351162715551391, "learning_rate": 2.8354097331826724e-06, "loss": 1.5098, "step": 10412 }, { "epoch": 0.7969539262207256, "grad_norm": 0.527483228784211, "learning_rate": 2.834803399101369e-06, "loss": 1.2432, "step": 10413 }, { "epoch": 0.7970304607377927, "grad_norm": 0.6287279452087933, "learning_rate": 2.834197259499207e-06, "loss": 1.529, "step": 10414 }, { "epoch": 0.79710699525486, "grad_norm": 0.5004631950880297, "learning_rate": 2.83359131441343e-06, "loss": 1.1316, "step": 10415 }, { "epoch": 0.7971835297719272, "grad_norm": 0.4962148228928312, "learning_rate": 2.8329855638812675e-06, "loss": 1.0134, "step": 10416 }, { "epoch": 0.7972600642889943, "grad_norm": 0.5773728687516841, "learning_rate": 2.8323800079399387e-06, "loss": 1.2874, "step": 10417 }, { "epoch": 0.7973365988060616, "grad_norm": 0.5568582640601237, "learning_rate": 2.8317746466266526e-06, "loss": 1.298, "step": 10418 }, { "epoch": 0.7974131333231287, "grad_norm": 0.7294839769568553, "learning_rate": 2.8311694799786035e-06, "loss": 1.3854, "step": 10419 }, { "epoch": 0.7974896678401959, "grad_norm": 0.5458075495979446, "learning_rate": 2.8305645080329746e-06, "loss": 1.324, "step": 10420 }, { "epoch": 0.7975662023572632, "grad_norm": 0.9461135112632933, "learning_rate": 2.8299597308269365e-06, "loss": 1.4812, "step": 10421 }, { "epoch": 0.7976427368743303, "grad_norm": 2.0222642305244976, "learning_rate": 2.82935514839765e-06, "loss": 1.6986, "step": 10422 }, { "epoch": 0.7977192713913975, "grad_norm": 0.6104398789400034, "learning_rate": 2.8287507607822595e-06, "loss": 1.4685, "step": 10423 }, { "epoch": 0.7977958059084648, "grad_norm": 0.7314681364358766, "learning_rate": 2.828146568017904e-06, "loss": 1.4479, "step": 10424 }, { "epoch": 0.7978723404255319, "grad_norm": 0.5285057038534869, "learning_rate": 2.827542570141707e-06, "loss": 1.196, "step": 10425 }, { "epoch": 0.7979488749425991, "grad_norm": 0.5636102401274625, "learning_rate": 2.8269387671907765e-06, "loss": 1.3707, "step": 10426 }, { "epoch": 0.7980254094596663, "grad_norm": 0.5888751411061764, "learning_rate": 2.826335159202215e-06, "loss": 1.3403, "step": 10427 }, { "epoch": 0.7981019439767335, "grad_norm": 0.6257736199036745, "learning_rate": 2.825731746213107e-06, "loss": 1.4093, "step": 10428 }, { "epoch": 0.7981784784938007, "grad_norm": 0.9622697383530072, "learning_rate": 2.8251285282605312e-06, "loss": 1.6804, "step": 10429 }, { "epoch": 0.7982550130108679, "grad_norm": 0.7054893547173818, "learning_rate": 2.824525505381549e-06, "loss": 1.4229, "step": 10430 }, { "epoch": 0.7983315475279351, "grad_norm": 0.7973003448335425, "learning_rate": 2.8239226776132135e-06, "loss": 1.3917, "step": 10431 }, { "epoch": 0.7984080820450022, "grad_norm": 0.7586230884406585, "learning_rate": 2.8233200449925625e-06, "loss": 1.6518, "step": 10432 }, { "epoch": 0.7984846165620695, "grad_norm": 0.585782757091924, "learning_rate": 2.8227176075566236e-06, "loss": 1.2913, "step": 10433 }, { "epoch": 0.7985611510791367, "grad_norm": 0.5705132473545678, "learning_rate": 2.8221153653424145e-06, "loss": 1.3894, "step": 10434 }, { "epoch": 0.7986376855962039, "grad_norm": 0.5752924606380654, "learning_rate": 2.821513318386937e-06, "loss": 1.3749, "step": 10435 }, { "epoch": 0.7987142201132711, "grad_norm": 0.6224813420132449, "learning_rate": 2.820911466727183e-06, "loss": 1.2891, "step": 10436 }, { "epoch": 0.7987907546303383, "grad_norm": 0.5985312712773138, "learning_rate": 2.8203098104001326e-06, "loss": 1.365, "step": 10437 }, { "epoch": 0.7988672891474055, "grad_norm": 0.5526917703763654, "learning_rate": 2.8197083494427526e-06, "loss": 1.2862, "step": 10438 }, { "epoch": 0.7989438236644727, "grad_norm": 0.5098308303910307, "learning_rate": 2.8191070838919975e-06, "loss": 1.1707, "step": 10439 }, { "epoch": 0.7990203581815398, "grad_norm": 0.5790759441045368, "learning_rate": 2.818506013784814e-06, "loss": 1.48, "step": 10440 }, { "epoch": 0.7990968926986071, "grad_norm": 1.0670286593313663, "learning_rate": 2.817905139158131e-06, "loss": 1.241, "step": 10441 }, { "epoch": 0.7991734272156743, "grad_norm": 0.5095332226759498, "learning_rate": 2.81730446004887e-06, "loss": 1.0871, "step": 10442 }, { "epoch": 0.7992499617327414, "grad_norm": 0.6392408629682805, "learning_rate": 2.8167039764939368e-06, "loss": 1.1696, "step": 10443 }, { "epoch": 0.7993264962498087, "grad_norm": 0.627687469636782, "learning_rate": 2.816103688530227e-06, "loss": 1.3288, "step": 10444 }, { "epoch": 0.7994030307668759, "grad_norm": 0.5747341889055423, "learning_rate": 2.815503596194625e-06, "loss": 1.3174, "step": 10445 }, { "epoch": 0.799479565283943, "grad_norm": 0.6150245213004645, "learning_rate": 2.814903699524002e-06, "loss": 1.5459, "step": 10446 }, { "epoch": 0.7995560998010103, "grad_norm": 0.8695737657097405, "learning_rate": 2.814303998555218e-06, "loss": 1.6236, "step": 10447 }, { "epoch": 0.7996326343180774, "grad_norm": 0.6622484887367307, "learning_rate": 2.813704493325119e-06, "loss": 1.4873, "step": 10448 }, { "epoch": 0.7997091688351446, "grad_norm": 0.652723749578061, "learning_rate": 2.8131051838705407e-06, "loss": 1.666, "step": 10449 }, { "epoch": 0.7997857033522119, "grad_norm": 0.5410531731584756, "learning_rate": 2.812506070228308e-06, "loss": 1.2712, "step": 10450 }, { "epoch": 0.799862237869279, "grad_norm": 0.5943946177195718, "learning_rate": 2.8119071524352303e-06, "loss": 1.3223, "step": 10451 }, { "epoch": 0.7999387723863463, "grad_norm": 0.601926443760846, "learning_rate": 2.811308430528108e-06, "loss": 1.7437, "step": 10452 }, { "epoch": 0.8000153069034135, "grad_norm": 0.6092336330022998, "learning_rate": 2.8107099045437286e-06, "loss": 1.5194, "step": 10453 }, { "epoch": 0.8000918414204806, "grad_norm": 0.5949043638759874, "learning_rate": 2.810111574518867e-06, "loss": 1.5073, "step": 10454 }, { "epoch": 0.8001683759375479, "grad_norm": 0.6331422650567152, "learning_rate": 2.8095134404902834e-06, "loss": 1.5345, "step": 10455 }, { "epoch": 0.800244910454615, "grad_norm": 0.5732384094537653, "learning_rate": 2.8089155024947346e-06, "loss": 1.0884, "step": 10456 }, { "epoch": 0.8003214449716822, "grad_norm": 0.5413091748733144, "learning_rate": 2.808317760568956e-06, "loss": 1.3207, "step": 10457 }, { "epoch": 0.8003979794887495, "grad_norm": 0.6177208873250236, "learning_rate": 2.8077202147496763e-06, "loss": 1.5274, "step": 10458 }, { "epoch": 0.8004745140058166, "grad_norm": 0.597897713362493, "learning_rate": 2.8071228650736093e-06, "loss": 1.3168, "step": 10459 }, { "epoch": 0.8005510485228838, "grad_norm": 0.6116552766818902, "learning_rate": 2.806525711577458e-06, "loss": 1.3949, "step": 10460 }, { "epoch": 0.800627583039951, "grad_norm": 0.8958198177956322, "learning_rate": 2.805928754297913e-06, "loss": 1.5545, "step": 10461 }, { "epoch": 0.8007041175570182, "grad_norm": 0.6484190052164001, "learning_rate": 2.805331993271654e-06, "loss": 1.2585, "step": 10462 }, { "epoch": 0.8007806520740854, "grad_norm": 0.5317425014913667, "learning_rate": 2.804735428535349e-06, "loss": 1.2016, "step": 10463 }, { "epoch": 0.8008571865911526, "grad_norm": 0.5942511198504311, "learning_rate": 2.8041390601256503e-06, "loss": 1.3855, "step": 10464 }, { "epoch": 0.8009337211082198, "grad_norm": 0.7205916171373146, "learning_rate": 2.8035428880792015e-06, "loss": 1.268, "step": 10465 }, { "epoch": 0.801010255625287, "grad_norm": 0.5806324481241161, "learning_rate": 2.8029469124326325e-06, "loss": 1.3657, "step": 10466 }, { "epoch": 0.8010867901423542, "grad_norm": 0.7351883584471218, "learning_rate": 2.802351133222564e-06, "loss": 1.247, "step": 10467 }, { "epoch": 0.8011633246594214, "grad_norm": 0.5640143973658542, "learning_rate": 2.8017555504856005e-06, "loss": 1.3537, "step": 10468 }, { "epoch": 0.8012398591764885, "grad_norm": 0.7143358403167973, "learning_rate": 2.8011601642583365e-06, "loss": 1.3288, "step": 10469 }, { "epoch": 0.8013163936935558, "grad_norm": 0.7130328801903275, "learning_rate": 2.8005649745773556e-06, "loss": 1.3868, "step": 10470 }, { "epoch": 0.801392928210623, "grad_norm": 0.5540367564900415, "learning_rate": 2.799969981479227e-06, "loss": 1.0998, "step": 10471 }, { "epoch": 0.8014694627276902, "grad_norm": 0.5260757134632097, "learning_rate": 2.799375185000509e-06, "loss": 1.2057, "step": 10472 }, { "epoch": 0.8015459972447574, "grad_norm": 0.7542268286007737, "learning_rate": 2.7987805851777473e-06, "loss": 1.1234, "step": 10473 }, { "epoch": 0.8016225317618246, "grad_norm": 0.5333115564984303, "learning_rate": 2.7981861820474764e-06, "loss": 1.1678, "step": 10474 }, { "epoch": 0.8016990662788918, "grad_norm": 0.5064601205083193, "learning_rate": 2.797591975646218e-06, "loss": 1.0862, "step": 10475 }, { "epoch": 0.801775600795959, "grad_norm": 0.6318313371774408, "learning_rate": 2.7969979660104817e-06, "loss": 1.3669, "step": 10476 }, { "epoch": 0.8018521353130261, "grad_norm": 0.9976860739704381, "learning_rate": 2.7964041531767654e-06, "loss": 1.4602, "step": 10477 }, { "epoch": 0.8019286698300934, "grad_norm": 0.6592941409111301, "learning_rate": 2.795810537181555e-06, "loss": 1.5008, "step": 10478 }, { "epoch": 0.8020052043471606, "grad_norm": 0.659715430360873, "learning_rate": 2.795217118061325e-06, "loss": 1.3142, "step": 10479 }, { "epoch": 0.8020817388642277, "grad_norm": 0.5088678158942863, "learning_rate": 2.7946238958525347e-06, "loss": 1.1901, "step": 10480 }, { "epoch": 0.802158273381295, "grad_norm": 0.5786564310863884, "learning_rate": 2.7940308705916345e-06, "loss": 1.3912, "step": 10481 }, { "epoch": 0.8022348078983622, "grad_norm": 0.7938375808787058, "learning_rate": 2.793438042315062e-06, "loss": 1.4662, "step": 10482 }, { "epoch": 0.8023113424154293, "grad_norm": 0.633671330496378, "learning_rate": 2.7928454110592413e-06, "loss": 1.3843, "step": 10483 }, { "epoch": 0.8023878769324966, "grad_norm": 0.607438876727414, "learning_rate": 2.7922529768605868e-06, "loss": 1.4697, "step": 10484 }, { "epoch": 0.8024644114495637, "grad_norm": 0.5518466445820979, "learning_rate": 2.7916607397554986e-06, "loss": 1.209, "step": 10485 }, { "epoch": 0.802540945966631, "grad_norm": 0.5939203607400236, "learning_rate": 2.791068699780366e-06, "loss": 1.402, "step": 10486 }, { "epoch": 0.8026174804836982, "grad_norm": 0.6017184376704227, "learning_rate": 2.7904768569715647e-06, "loss": 1.4301, "step": 10487 }, { "epoch": 0.8026940150007653, "grad_norm": 0.8376528060285882, "learning_rate": 2.7898852113654596e-06, "loss": 1.6955, "step": 10488 }, { "epoch": 0.8027705495178326, "grad_norm": 0.5911832999267908, "learning_rate": 2.7892937629984042e-06, "loss": 1.4239, "step": 10489 }, { "epoch": 0.8028470840348998, "grad_norm": 0.7564482587230504, "learning_rate": 2.7887025119067377e-06, "loss": 1.6332, "step": 10490 }, { "epoch": 0.8029236185519669, "grad_norm": 0.615971584790189, "learning_rate": 2.7881114581267897e-06, "loss": 1.3624, "step": 10491 }, { "epoch": 0.8030001530690342, "grad_norm": 0.6211131151602748, "learning_rate": 2.7875206016948754e-06, "loss": 1.3905, "step": 10492 }, { "epoch": 0.8030766875861013, "grad_norm": 0.58948894723342, "learning_rate": 2.7869299426472983e-06, "loss": 1.3832, "step": 10493 }, { "epoch": 0.8031532221031685, "grad_norm": 0.646914227177337, "learning_rate": 2.7863394810203515e-06, "loss": 1.6491, "step": 10494 }, { "epoch": 0.8032297566202358, "grad_norm": 0.5321815862302007, "learning_rate": 2.785749216850313e-06, "loss": 1.1948, "step": 10495 }, { "epoch": 0.8033062911373029, "grad_norm": 0.5525153025605821, "learning_rate": 2.7851591501734527e-06, "loss": 1.2563, "step": 10496 }, { "epoch": 0.8033828256543701, "grad_norm": 0.6548105183066228, "learning_rate": 2.784569281026024e-06, "loss": 1.6931, "step": 10497 }, { "epoch": 0.8034593601714373, "grad_norm": 0.6415505187230653, "learning_rate": 2.7839796094442705e-06, "loss": 1.4978, "step": 10498 }, { "epoch": 0.8035358946885045, "grad_norm": 0.5070190547206278, "learning_rate": 2.7833901354644233e-06, "loss": 1.0979, "step": 10499 }, { "epoch": 0.8036124292055717, "grad_norm": 0.5907443083083395, "learning_rate": 2.782800859122703e-06, "loss": 1.4451, "step": 10500 }, { "epoch": 0.8036889637226389, "grad_norm": 0.5164564831207311, "learning_rate": 2.782211780455316e-06, "loss": 1.1549, "step": 10501 }, { "epoch": 0.8037654982397061, "grad_norm": 0.5693473498915099, "learning_rate": 2.7816228994984564e-06, "loss": 1.3572, "step": 10502 }, { "epoch": 0.8038420327567734, "grad_norm": 0.6831344685455222, "learning_rate": 2.7810342162883065e-06, "loss": 1.2982, "step": 10503 }, { "epoch": 0.8039185672738405, "grad_norm": 0.5512894474139736, "learning_rate": 2.7804457308610367e-06, "loss": 1.1119, "step": 10504 }, { "epoch": 0.8039951017909077, "grad_norm": 0.5051680049146886, "learning_rate": 2.7798574432528056e-06, "loss": 1.1196, "step": 10505 }, { "epoch": 0.8040716363079748, "grad_norm": 0.588213388962704, "learning_rate": 2.7792693534997604e-06, "loss": 1.4071, "step": 10506 }, { "epoch": 0.8041481708250421, "grad_norm": 0.5704682084544462, "learning_rate": 2.778681461638035e-06, "loss": 1.2738, "step": 10507 }, { "epoch": 0.8042247053421093, "grad_norm": 0.5983776416602495, "learning_rate": 2.77809376770375e-06, "loss": 1.3171, "step": 10508 }, { "epoch": 0.8043012398591765, "grad_norm": 0.6437567131918707, "learning_rate": 2.777506271733016e-06, "loss": 1.5737, "step": 10509 }, { "epoch": 0.8043777743762437, "grad_norm": 0.7680421336696285, "learning_rate": 2.776918973761928e-06, "loss": 1.5828, "step": 10510 }, { "epoch": 0.8044543088933109, "grad_norm": 0.5783632387664794, "learning_rate": 2.7763318738265753e-06, "loss": 1.4335, "step": 10511 }, { "epoch": 0.8045308434103781, "grad_norm": 0.5822099790577445, "learning_rate": 2.775744971963028e-06, "loss": 1.2255, "step": 10512 }, { "epoch": 0.8046073779274453, "grad_norm": 0.4776625823701729, "learning_rate": 2.77515826820735e-06, "loss": 0.9722, "step": 10513 }, { "epoch": 0.8046839124445124, "grad_norm": 0.6384118477769569, "learning_rate": 2.7745717625955878e-06, "loss": 1.6215, "step": 10514 }, { "epoch": 0.8047604469615797, "grad_norm": 0.7842152546274399, "learning_rate": 2.773985455163778e-06, "loss": 1.4069, "step": 10515 }, { "epoch": 0.8048369814786469, "grad_norm": 0.7070549279943673, "learning_rate": 2.7733993459479464e-06, "loss": 1.3485, "step": 10516 }, { "epoch": 0.804913515995714, "grad_norm": 1.5354562805706122, "learning_rate": 2.7728134349841047e-06, "loss": 1.3695, "step": 10517 }, { "epoch": 0.8049900505127813, "grad_norm": 0.835936133963205, "learning_rate": 2.772227722308253e-06, "loss": 1.3647, "step": 10518 }, { "epoch": 0.8050665850298485, "grad_norm": 0.469876992909017, "learning_rate": 2.771642207956379e-06, "loss": 0.9309, "step": 10519 }, { "epoch": 0.8051431195469156, "grad_norm": 0.730396463846632, "learning_rate": 2.771056891964459e-06, "loss": 1.5178, "step": 10520 }, { "epoch": 0.8052196540639829, "grad_norm": 0.658746523861485, "learning_rate": 2.770471774368455e-06, "loss": 1.4559, "step": 10521 }, { "epoch": 0.80529618858105, "grad_norm": 0.6230546669280933, "learning_rate": 2.769886855204321e-06, "loss": 1.5099, "step": 10522 }, { "epoch": 0.8053727230981172, "grad_norm": 0.5961821145281285, "learning_rate": 2.769302134507995e-06, "loss": 1.4079, "step": 10523 }, { "epoch": 0.8054492576151845, "grad_norm": 0.6159461596420254, "learning_rate": 2.768717612315403e-06, "loss": 1.4654, "step": 10524 }, { "epoch": 0.8055257921322516, "grad_norm": 0.5776438498887294, "learning_rate": 2.7681332886624614e-06, "loss": 1.2945, "step": 10525 }, { "epoch": 0.8056023266493189, "grad_norm": 0.6506650675899204, "learning_rate": 2.7675491635850714e-06, "loss": 1.3821, "step": 10526 }, { "epoch": 0.8056788611663861, "grad_norm": 0.5862254493483218, "learning_rate": 2.766965237119123e-06, "loss": 1.3223, "step": 10527 }, { "epoch": 0.8057553956834532, "grad_norm": 0.5582528446213415, "learning_rate": 2.7663815093004964e-06, "loss": 1.2889, "step": 10528 }, { "epoch": 0.8058319302005205, "grad_norm": 0.549879022043925, "learning_rate": 2.765797980165057e-06, "loss": 1.2343, "step": 10529 }, { "epoch": 0.8059084647175876, "grad_norm": 0.5642557209721449, "learning_rate": 2.7652146497486576e-06, "loss": 1.264, "step": 10530 }, { "epoch": 0.8059849992346548, "grad_norm": 0.5653514103017097, "learning_rate": 2.764631518087141e-06, "loss": 1.3474, "step": 10531 }, { "epoch": 0.8060615337517221, "grad_norm": 0.6421561724482392, "learning_rate": 2.7640485852163336e-06, "loss": 1.4432, "step": 10532 }, { "epoch": 0.8061380682687892, "grad_norm": 0.5750208767686733, "learning_rate": 2.7634658511720567e-06, "loss": 1.3559, "step": 10533 }, { "epoch": 0.8062146027858564, "grad_norm": 0.5756971803111087, "learning_rate": 2.7628833159901125e-06, "loss": 1.3411, "step": 10534 }, { "epoch": 0.8062911373029236, "grad_norm": 0.6159807067387686, "learning_rate": 2.7623009797062955e-06, "loss": 1.5742, "step": 10535 }, { "epoch": 0.8063676718199908, "grad_norm": 0.7067013069578368, "learning_rate": 2.7617188423563846e-06, "loss": 1.511, "step": 10536 }, { "epoch": 0.806444206337058, "grad_norm": 0.6861551130854169, "learning_rate": 2.7611369039761494e-06, "loss": 1.6901, "step": 10537 }, { "epoch": 0.8065207408541252, "grad_norm": 0.5289863471364585, "learning_rate": 2.760555164601344e-06, "loss": 1.1444, "step": 10538 }, { "epoch": 0.8065972753711924, "grad_norm": 0.5885848629922863, "learning_rate": 2.7599736242677144e-06, "loss": 1.336, "step": 10539 }, { "epoch": 0.8066738098882597, "grad_norm": 0.5702224506272641, "learning_rate": 2.759392283010991e-06, "loss": 1.3162, "step": 10540 }, { "epoch": 0.8067503444053268, "grad_norm": 0.689240273206332, "learning_rate": 2.758811140866893e-06, "loss": 1.6701, "step": 10541 }, { "epoch": 0.806826878922394, "grad_norm": 0.6933619476087952, "learning_rate": 2.7582301978711275e-06, "loss": 1.4611, "step": 10542 }, { "epoch": 0.8069034134394611, "grad_norm": 0.7220632333899799, "learning_rate": 2.757649454059389e-06, "loss": 1.4011, "step": 10543 }, { "epoch": 0.8069799479565284, "grad_norm": 0.6825442089913034, "learning_rate": 2.757068909467362e-06, "loss": 1.7097, "step": 10544 }, { "epoch": 0.8070564824735956, "grad_norm": 0.5262008781953096, "learning_rate": 2.756488564130716e-06, "loss": 1.1557, "step": 10545 }, { "epoch": 0.8071330169906628, "grad_norm": 0.5500724319484275, "learning_rate": 2.755908418085108e-06, "loss": 1.0918, "step": 10546 }, { "epoch": 0.80720955150773, "grad_norm": 0.5289940781856848, "learning_rate": 2.755328471366185e-06, "loss": 1.1719, "step": 10547 }, { "epoch": 0.8072860860247972, "grad_norm": 0.6216290726111884, "learning_rate": 2.754748724009581e-06, "loss": 1.4446, "step": 10548 }, { "epoch": 0.8073626205418644, "grad_norm": 0.6766433948251727, "learning_rate": 2.7541691760509154e-06, "loss": 1.4305, "step": 10549 }, { "epoch": 0.8074391550589316, "grad_norm": 0.5579288735350719, "learning_rate": 2.7535898275257998e-06, "loss": 1.2099, "step": 10550 }, { "epoch": 0.8075156895759987, "grad_norm": 1.903150137997662, "learning_rate": 2.7530106784698307e-06, "loss": 1.4173, "step": 10551 }, { "epoch": 0.807592224093066, "grad_norm": 0.7900144259661935, "learning_rate": 2.7524317289185916e-06, "loss": 1.3481, "step": 10552 }, { "epoch": 0.8076687586101332, "grad_norm": 0.5568126133368847, "learning_rate": 2.7518529789076553e-06, "loss": 1.1025, "step": 10553 }, { "epoch": 0.8077452931272003, "grad_norm": 0.5229750899862633, "learning_rate": 2.751274428472581e-06, "loss": 1.0704, "step": 10554 }, { "epoch": 0.8078218276442676, "grad_norm": 0.9692516196783062, "learning_rate": 2.75069607764892e-06, "loss": 1.4688, "step": 10555 }, { "epoch": 0.8078983621613348, "grad_norm": 0.5955468606839175, "learning_rate": 2.7501179264722045e-06, "loss": 1.252, "step": 10556 }, { "epoch": 0.8079748966784019, "grad_norm": 0.6400216761893586, "learning_rate": 2.749539974977959e-06, "loss": 1.4339, "step": 10557 }, { "epoch": 0.8080514311954692, "grad_norm": 0.693543855032561, "learning_rate": 2.7489622232016944e-06, "loss": 1.7024, "step": 10558 }, { "epoch": 0.8081279657125363, "grad_norm": 0.5672050759326762, "learning_rate": 2.7483846711789097e-06, "loss": 1.3288, "step": 10559 }, { "epoch": 0.8082045002296036, "grad_norm": 0.6867787899900294, "learning_rate": 2.7478073189450904e-06, "loss": 1.1138, "step": 10560 }, { "epoch": 0.8082810347466708, "grad_norm": 0.5675192999074233, "learning_rate": 2.7472301665357124e-06, "loss": 1.3188, "step": 10561 }, { "epoch": 0.8083575692637379, "grad_norm": 0.6335260549673651, "learning_rate": 2.746653213986237e-06, "loss": 1.2973, "step": 10562 }, { "epoch": 0.8084341037808052, "grad_norm": 0.545596238912522, "learning_rate": 2.7460764613321143e-06, "loss": 1.1298, "step": 10563 }, { "epoch": 0.8085106382978723, "grad_norm": 0.649204033359443, "learning_rate": 2.7454999086087803e-06, "loss": 1.6424, "step": 10564 }, { "epoch": 0.8085871728149395, "grad_norm": 1.00144000215444, "learning_rate": 2.74492355585166e-06, "loss": 1.5645, "step": 10565 }, { "epoch": 0.8086637073320068, "grad_norm": 0.6051784166564946, "learning_rate": 2.7443474030961694e-06, "loss": 1.4379, "step": 10566 }, { "epoch": 0.8087402418490739, "grad_norm": 0.5649333642453135, "learning_rate": 2.743771450377706e-06, "loss": 1.3706, "step": 10567 }, { "epoch": 0.8088167763661411, "grad_norm": 0.5641481170215863, "learning_rate": 2.743195697731659e-06, "loss": 1.3473, "step": 10568 }, { "epoch": 0.8088933108832084, "grad_norm": 0.6607151595258767, "learning_rate": 2.7426201451934042e-06, "loss": 1.5868, "step": 10569 }, { "epoch": 0.8089698454002755, "grad_norm": 0.47764023955908647, "learning_rate": 2.7420447927983057e-06, "loss": 0.9616, "step": 10570 }, { "epoch": 0.8090463799173427, "grad_norm": 0.4884873238225705, "learning_rate": 2.7414696405817133e-06, "loss": 1.0346, "step": 10571 }, { "epoch": 0.8091229144344099, "grad_norm": 0.5678710705495835, "learning_rate": 2.7408946885789683e-06, "loss": 1.3928, "step": 10572 }, { "epoch": 0.8091994489514771, "grad_norm": 0.8286535318156892, "learning_rate": 2.740319936825397e-06, "loss": 1.3678, "step": 10573 }, { "epoch": 0.8092759834685443, "grad_norm": 0.7916582160958962, "learning_rate": 2.739745385356313e-06, "loss": 1.6434, "step": 10574 }, { "epoch": 0.8093525179856115, "grad_norm": 0.4589408641246339, "learning_rate": 2.7391710342070184e-06, "loss": 0.9441, "step": 10575 }, { "epoch": 0.8094290525026787, "grad_norm": 0.6136539921785655, "learning_rate": 2.738596883412803e-06, "loss": 1.427, "step": 10576 }, { "epoch": 0.809505587019746, "grad_norm": 0.6027438052644801, "learning_rate": 2.738022933008946e-06, "loss": 1.3591, "step": 10577 }, { "epoch": 0.8095821215368131, "grad_norm": 0.5081498227784372, "learning_rate": 2.737449183030711e-06, "loss": 1.196, "step": 10578 }, { "epoch": 0.8096586560538803, "grad_norm": 0.6286372080882949, "learning_rate": 2.7368756335133515e-06, "loss": 1.4387, "step": 10579 }, { "epoch": 0.8097351905709474, "grad_norm": 0.6331406350392801, "learning_rate": 2.736302284492108e-06, "loss": 1.52, "step": 10580 }, { "epoch": 0.8098117250880147, "grad_norm": 0.5844811309376199, "learning_rate": 2.7357291360022075e-06, "loss": 1.2575, "step": 10581 }, { "epoch": 0.8098882596050819, "grad_norm": 0.7976590411513726, "learning_rate": 2.735156188078869e-06, "loss": 1.7065, "step": 10582 }, { "epoch": 0.8099647941221491, "grad_norm": 0.6499305678355737, "learning_rate": 2.7345834407572937e-06, "loss": 1.5009, "step": 10583 }, { "epoch": 0.8100413286392163, "grad_norm": 0.5403116150432589, "learning_rate": 2.734010894072674e-06, "loss": 1.3683, "step": 10584 }, { "epoch": 0.8101178631562835, "grad_norm": 0.5838575902384113, "learning_rate": 2.7334385480601876e-06, "loss": 1.3662, "step": 10585 }, { "epoch": 0.8101943976733507, "grad_norm": 0.6674318647142022, "learning_rate": 2.7328664027550033e-06, "loss": 1.4248, "step": 10586 }, { "epoch": 0.8102709321904179, "grad_norm": 0.5668110706873649, "learning_rate": 2.7322944581922728e-06, "loss": 1.3301, "step": 10587 }, { "epoch": 0.810347466707485, "grad_norm": 0.4788986094834068, "learning_rate": 2.731722714407139e-06, "loss": 1.0959, "step": 10588 }, { "epoch": 0.8104240012245523, "grad_norm": 0.671420005130898, "learning_rate": 2.731151171434733e-06, "loss": 1.5252, "step": 10589 }, { "epoch": 0.8105005357416195, "grad_norm": 0.6322463127233332, "learning_rate": 2.730579829310171e-06, "loss": 1.4356, "step": 10590 }, { "epoch": 0.8105770702586866, "grad_norm": 0.5845378830213357, "learning_rate": 2.730008688068557e-06, "loss": 1.3692, "step": 10591 }, { "epoch": 0.8106536047757539, "grad_norm": 0.5850668300141738, "learning_rate": 2.7294377477449853e-06, "loss": 1.4173, "step": 10592 }, { "epoch": 0.8107301392928211, "grad_norm": 0.5001212809908684, "learning_rate": 2.728867008374535e-06, "loss": 1.0231, "step": 10593 }, { "epoch": 0.8108066738098882, "grad_norm": 0.6698436210637247, "learning_rate": 2.7282964699922754e-06, "loss": 1.5407, "step": 10594 }, { "epoch": 0.8108832083269555, "grad_norm": 0.6106498938874696, "learning_rate": 2.7277261326332606e-06, "loss": 1.4308, "step": 10595 }, { "epoch": 0.8109597428440226, "grad_norm": 0.5396808361992513, "learning_rate": 2.727155996332535e-06, "loss": 1.1852, "step": 10596 }, { "epoch": 0.8110362773610899, "grad_norm": 0.5645534221001501, "learning_rate": 2.7265860611251284e-06, "loss": 1.1597, "step": 10597 }, { "epoch": 0.8111128118781571, "grad_norm": 0.6069909971014815, "learning_rate": 2.7260163270460594e-06, "loss": 1.4806, "step": 10598 }, { "epoch": 0.8111893463952242, "grad_norm": 0.6316673152685434, "learning_rate": 2.7254467941303354e-06, "loss": 1.5045, "step": 10599 }, { "epoch": 0.8112658809122915, "grad_norm": 0.6887337729516856, "learning_rate": 2.7248774624129495e-06, "loss": 1.4567, "step": 10600 }, { "epoch": 0.8113424154293586, "grad_norm": 0.5261063434073697, "learning_rate": 2.724308331928882e-06, "loss": 1.2371, "step": 10601 }, { "epoch": 0.8114189499464258, "grad_norm": 0.6730466014290596, "learning_rate": 2.723739402713104e-06, "loss": 1.5357, "step": 10602 }, { "epoch": 0.8114954844634931, "grad_norm": 1.2150129883151786, "learning_rate": 2.7231706748005704e-06, "loss": 1.8994, "step": 10603 }, { "epoch": 0.8115720189805602, "grad_norm": 0.6078081867237136, "learning_rate": 2.722602148226227e-06, "loss": 1.4638, "step": 10604 }, { "epoch": 0.8116485534976274, "grad_norm": 0.5794223060897608, "learning_rate": 2.7220338230250056e-06, "loss": 1.5434, "step": 10605 }, { "epoch": 0.8117250880146947, "grad_norm": 0.5829981124780959, "learning_rate": 2.721465699231825e-06, "loss": 1.4113, "step": 10606 }, { "epoch": 0.8118016225317618, "grad_norm": 0.5424001321511982, "learning_rate": 2.7208977768815932e-06, "loss": 1.1468, "step": 10607 }, { "epoch": 0.811878157048829, "grad_norm": 0.5031459429329512, "learning_rate": 2.7203300560092036e-06, "loss": 1.0268, "step": 10608 }, { "epoch": 0.8119546915658962, "grad_norm": 1.015305726156042, "learning_rate": 2.719762536649541e-06, "loss": 1.4287, "step": 10609 }, { "epoch": 0.8120312260829634, "grad_norm": 0.6250274357408878, "learning_rate": 2.7191952188374737e-06, "loss": 1.5482, "step": 10610 }, { "epoch": 0.8121077606000306, "grad_norm": 0.5706546259035608, "learning_rate": 2.71862810260786e-06, "loss": 1.4896, "step": 10611 }, { "epoch": 0.8121842951170978, "grad_norm": 0.6131963590353142, "learning_rate": 2.7180611879955443e-06, "loss": 1.396, "step": 10612 }, { "epoch": 0.812260829634165, "grad_norm": 0.5665651796110663, "learning_rate": 2.7174944750353617e-06, "loss": 1.1936, "step": 10613 }, { "epoch": 0.8123373641512323, "grad_norm": 0.7756427994670838, "learning_rate": 2.7169279637621294e-06, "loss": 1.738, "step": 10614 }, { "epoch": 0.8124138986682994, "grad_norm": 0.5531457213244536, "learning_rate": 2.7163616542106588e-06, "loss": 1.2134, "step": 10615 }, { "epoch": 0.8124904331853666, "grad_norm": 0.6700760780549313, "learning_rate": 2.715795546415744e-06, "loss": 1.5099, "step": 10616 }, { "epoch": 0.8125669677024338, "grad_norm": 0.5782059104748943, "learning_rate": 2.715229640412169e-06, "loss": 1.42, "step": 10617 }, { "epoch": 0.812643502219501, "grad_norm": 0.6081198199088595, "learning_rate": 2.714663936234705e-06, "loss": 1.4082, "step": 10618 }, { "epoch": 0.8127200367365682, "grad_norm": 0.5998888339051444, "learning_rate": 2.7140984339181097e-06, "loss": 1.4422, "step": 10619 }, { "epoch": 0.8127965712536354, "grad_norm": 0.7311690073016814, "learning_rate": 2.713533133497128e-06, "loss": 1.6286, "step": 10620 }, { "epoch": 0.8128731057707026, "grad_norm": 0.5824168087338827, "learning_rate": 2.7129680350064975e-06, "loss": 1.3199, "step": 10621 }, { "epoch": 0.8129496402877698, "grad_norm": 0.5925176454698499, "learning_rate": 2.7124031384809362e-06, "loss": 1.2546, "step": 10622 }, { "epoch": 0.813026174804837, "grad_norm": 0.6885229239324078, "learning_rate": 2.711838443955155e-06, "loss": 1.9368, "step": 10623 }, { "epoch": 0.8131027093219042, "grad_norm": 0.6370202022035841, "learning_rate": 2.7112739514638486e-06, "loss": 1.545, "step": 10624 }, { "epoch": 0.8131792438389713, "grad_norm": 0.626643206903564, "learning_rate": 2.7107096610417015e-06, "loss": 1.3387, "step": 10625 }, { "epoch": 0.8132557783560386, "grad_norm": 0.629308864872548, "learning_rate": 2.7101455727233867e-06, "loss": 1.3893, "step": 10626 }, { "epoch": 0.8133323128731058, "grad_norm": 0.6345788482816876, "learning_rate": 2.709581686543563e-06, "loss": 1.5282, "step": 10627 }, { "epoch": 0.8134088473901729, "grad_norm": 0.6387730126238891, "learning_rate": 2.7090180025368767e-06, "loss": 1.5594, "step": 10628 }, { "epoch": 0.8134853819072402, "grad_norm": 0.5363266850473964, "learning_rate": 2.7084545207379624e-06, "loss": 1.1956, "step": 10629 }, { "epoch": 0.8135619164243073, "grad_norm": 0.872964625591804, "learning_rate": 2.7078912411814423e-06, "loss": 1.3777, "step": 10630 }, { "epoch": 0.8136384509413745, "grad_norm": 0.6007722026199099, "learning_rate": 2.7073281639019263e-06, "loss": 1.4713, "step": 10631 }, { "epoch": 0.8137149854584418, "grad_norm": 0.7471051554630808, "learning_rate": 2.706765288934011e-06, "loss": 1.4149, "step": 10632 }, { "epoch": 0.8137915199755089, "grad_norm": 0.6269897517581223, "learning_rate": 2.7062026163122813e-06, "loss": 1.3694, "step": 10633 }, { "epoch": 0.8138680544925762, "grad_norm": 0.7239055934953141, "learning_rate": 2.705640146071309e-06, "loss": 1.6727, "step": 10634 }, { "epoch": 0.8139445890096434, "grad_norm": 0.5820469811938213, "learning_rate": 2.7050778782456545e-06, "loss": 1.2972, "step": 10635 }, { "epoch": 0.8140211235267105, "grad_norm": 0.5036597875271965, "learning_rate": 2.704515812869864e-06, "loss": 1.0405, "step": 10636 }, { "epoch": 0.8140976580437778, "grad_norm": 0.6473172918235929, "learning_rate": 2.703953949978475e-06, "loss": 1.3003, "step": 10637 }, { "epoch": 0.8141741925608449, "grad_norm": 0.5219489559840012, "learning_rate": 2.703392289606008e-06, "loss": 1.01, "step": 10638 }, { "epoch": 0.8142507270779121, "grad_norm": 0.5467605408481635, "learning_rate": 2.702830831786974e-06, "loss": 1.2803, "step": 10639 }, { "epoch": 0.8143272615949794, "grad_norm": 0.5479480466193154, "learning_rate": 2.702269576555871e-06, "loss": 1.3142, "step": 10640 }, { "epoch": 0.8144037961120465, "grad_norm": 0.5335073921021314, "learning_rate": 2.7017085239471823e-06, "loss": 1.1632, "step": 10641 }, { "epoch": 0.8144803306291137, "grad_norm": 0.6806436775739566, "learning_rate": 2.7011476739953818e-06, "loss": 1.4669, "step": 10642 }, { "epoch": 0.814556865146181, "grad_norm": 0.559320520557653, "learning_rate": 2.700587026734931e-06, "loss": 1.4248, "step": 10643 }, { "epoch": 0.8146333996632481, "grad_norm": 0.6092161031642493, "learning_rate": 2.7000265822002763e-06, "loss": 1.4246, "step": 10644 }, { "epoch": 0.8147099341803153, "grad_norm": 0.641618301071624, "learning_rate": 2.6994663404258535e-06, "loss": 1.5017, "step": 10645 }, { "epoch": 0.8147864686973825, "grad_norm": 0.5842723648851222, "learning_rate": 2.6989063014460852e-06, "loss": 1.3793, "step": 10646 }, { "epoch": 0.8148630032144497, "grad_norm": 0.8602341878129356, "learning_rate": 2.698346465295381e-06, "loss": 1.3362, "step": 10647 }, { "epoch": 0.814939537731517, "grad_norm": 0.7516190983645575, "learning_rate": 2.697786832008142e-06, "loss": 1.5794, "step": 10648 }, { "epoch": 0.8150160722485841, "grad_norm": 0.5308699481080409, "learning_rate": 2.697227401618751e-06, "loss": 1.1615, "step": 10649 }, { "epoch": 0.8150926067656513, "grad_norm": 0.6217647620162308, "learning_rate": 2.696668174161582e-06, "loss": 1.515, "step": 10650 }, { "epoch": 0.8151691412827186, "grad_norm": 0.6040442918605926, "learning_rate": 2.6961091496709957e-06, "loss": 1.3281, "step": 10651 }, { "epoch": 0.8152456757997857, "grad_norm": 0.5850358635520225, "learning_rate": 2.69555032818134e-06, "loss": 1.3963, "step": 10652 }, { "epoch": 0.8153222103168529, "grad_norm": 0.6728491209813942, "learning_rate": 2.6949917097269503e-06, "loss": 1.5968, "step": 10653 }, { "epoch": 0.81539874483392, "grad_norm": 0.7006264835736663, "learning_rate": 2.69443329434215e-06, "loss": 1.6893, "step": 10654 }, { "epoch": 0.8154752793509873, "grad_norm": 0.5815682064089598, "learning_rate": 2.6938750820612496e-06, "loss": 1.3653, "step": 10655 }, { "epoch": 0.8155518138680545, "grad_norm": 0.56543827560337, "learning_rate": 2.693317072918548e-06, "loss": 1.394, "step": 10656 }, { "epoch": 0.8156283483851217, "grad_norm": 0.5749689777505985, "learning_rate": 2.6927592669483306e-06, "loss": 1.3199, "step": 10657 }, { "epoch": 0.8157048829021889, "grad_norm": 0.6134572551130238, "learning_rate": 2.6922016641848695e-06, "loss": 1.3847, "step": 10658 }, { "epoch": 0.8157814174192561, "grad_norm": 0.5863657185724567, "learning_rate": 2.6916442646624275e-06, "loss": 1.3052, "step": 10659 }, { "epoch": 0.8158579519363233, "grad_norm": 0.5356982781626212, "learning_rate": 2.6910870684152524e-06, "loss": 1.1164, "step": 10660 }, { "epoch": 0.8159344864533905, "grad_norm": 0.6229626615142965, "learning_rate": 2.6905300754775795e-06, "loss": 1.2706, "step": 10661 }, { "epoch": 0.8160110209704576, "grad_norm": 0.6483330145826096, "learning_rate": 2.689973285883631e-06, "loss": 1.0325, "step": 10662 }, { "epoch": 0.8160875554875249, "grad_norm": 0.5841707168820996, "learning_rate": 2.68941669966762e-06, "loss": 1.4656, "step": 10663 }, { "epoch": 0.8161640900045921, "grad_norm": 0.603457715354369, "learning_rate": 2.6888603168637426e-06, "loss": 1.5049, "step": 10664 }, { "epoch": 0.8162406245216592, "grad_norm": 0.5959392409275381, "learning_rate": 2.6883041375061867e-06, "loss": 1.1514, "step": 10665 }, { "epoch": 0.8163171590387265, "grad_norm": 0.5359487524068006, "learning_rate": 2.687748161629125e-06, "loss": 1.3537, "step": 10666 }, { "epoch": 0.8163936935557936, "grad_norm": 0.6142995650648189, "learning_rate": 2.687192389266718e-06, "loss": 1.2933, "step": 10667 }, { "epoch": 0.8164702280728608, "grad_norm": 0.7491831416339843, "learning_rate": 2.6866368204531135e-06, "loss": 1.441, "step": 10668 }, { "epoch": 0.8165467625899281, "grad_norm": 0.57248442859978, "learning_rate": 2.6860814552224467e-06, "loss": 1.2245, "step": 10669 }, { "epoch": 0.8166232971069952, "grad_norm": 0.5704107459511439, "learning_rate": 2.685526293608844e-06, "loss": 1.2756, "step": 10670 }, { "epoch": 0.8166998316240625, "grad_norm": 0.5396846211647964, "learning_rate": 2.6849713356464136e-06, "loss": 1.1682, "step": 10671 }, { "epoch": 0.8167763661411297, "grad_norm": 0.5225961733048979, "learning_rate": 2.6844165813692543e-06, "loss": 1.2455, "step": 10672 }, { "epoch": 0.8168529006581968, "grad_norm": 0.5304262287367444, "learning_rate": 2.683862030811452e-06, "loss": 1.0358, "step": 10673 }, { "epoch": 0.8169294351752641, "grad_norm": 0.581772440628154, "learning_rate": 2.683307684007079e-06, "loss": 1.2557, "step": 10674 }, { "epoch": 0.8170059696923312, "grad_norm": 0.5309443957227523, "learning_rate": 2.6827535409901984e-06, "loss": 1.1099, "step": 10675 }, { "epoch": 0.8170825042093984, "grad_norm": 0.5575578731139741, "learning_rate": 2.6821996017948566e-06, "loss": 1.2502, "step": 10676 }, { "epoch": 0.8171590387264657, "grad_norm": 0.669211500467636, "learning_rate": 2.6816458664550892e-06, "loss": 1.6132, "step": 10677 }, { "epoch": 0.8172355732435328, "grad_norm": 0.5310191542931277, "learning_rate": 2.68109233500492e-06, "loss": 1.2591, "step": 10678 }, { "epoch": 0.8173121077606, "grad_norm": 0.7562904591725749, "learning_rate": 2.6805390074783595e-06, "loss": 1.5989, "step": 10679 }, { "epoch": 0.8173886422776673, "grad_norm": 0.6913055015865346, "learning_rate": 2.679985883909404e-06, "loss": 1.4721, "step": 10680 }, { "epoch": 0.8174651767947344, "grad_norm": 0.5359113520461859, "learning_rate": 2.6794329643320427e-06, "loss": 1.3137, "step": 10681 }, { "epoch": 0.8175417113118016, "grad_norm": 0.6696116706059949, "learning_rate": 2.6788802487802463e-06, "loss": 1.3031, "step": 10682 }, { "epoch": 0.8176182458288688, "grad_norm": 0.6253163060365017, "learning_rate": 2.678327737287975e-06, "loss": 1.4844, "step": 10683 }, { "epoch": 0.817694780345936, "grad_norm": 0.5701991648602799, "learning_rate": 2.6777754298891776e-06, "loss": 1.2833, "step": 10684 }, { "epoch": 0.8177713148630033, "grad_norm": 0.5486948635536665, "learning_rate": 2.6772233266177903e-06, "loss": 1.2498, "step": 10685 }, { "epoch": 0.8178478493800704, "grad_norm": 0.9524782097303182, "learning_rate": 2.676671427507733e-06, "loss": 1.3258, "step": 10686 }, { "epoch": 0.8179243838971376, "grad_norm": 0.5092161851493647, "learning_rate": 2.67611973259292e-06, "loss": 1.1233, "step": 10687 }, { "epoch": 0.8180009184142049, "grad_norm": 0.8127898784208141, "learning_rate": 2.675568241907247e-06, "loss": 1.6348, "step": 10688 }, { "epoch": 0.818077452931272, "grad_norm": 0.5980227138331711, "learning_rate": 2.675016955484599e-06, "loss": 1.2466, "step": 10689 }, { "epoch": 0.8181539874483392, "grad_norm": 0.5639869685708172, "learning_rate": 2.674465873358849e-06, "loss": 1.1802, "step": 10690 }, { "epoch": 0.8182305219654064, "grad_norm": 0.5538448045402214, "learning_rate": 2.673914995563856e-06, "loss": 1.2172, "step": 10691 }, { "epoch": 0.8183070564824736, "grad_norm": 0.7158235177768838, "learning_rate": 2.6733643221334704e-06, "loss": 1.5299, "step": 10692 }, { "epoch": 0.8183835909995408, "grad_norm": 0.6827636707395344, "learning_rate": 2.672813853101525e-06, "loss": 1.6024, "step": 10693 }, { "epoch": 0.818460125516608, "grad_norm": 0.5555988199494091, "learning_rate": 2.672263588501843e-06, "loss": 1.1868, "step": 10694 }, { "epoch": 0.8185366600336752, "grad_norm": 0.5656366808699608, "learning_rate": 2.6717135283682346e-06, "loss": 1.2651, "step": 10695 }, { "epoch": 0.8186131945507424, "grad_norm": 1.0825290322291163, "learning_rate": 2.6711636727344964e-06, "loss": 1.6975, "step": 10696 }, { "epoch": 0.8186897290678096, "grad_norm": 0.6183028374534824, "learning_rate": 2.6706140216344135e-06, "loss": 1.4544, "step": 10697 }, { "epoch": 0.8187662635848768, "grad_norm": 0.6724150992778187, "learning_rate": 2.670064575101758e-06, "loss": 1.5619, "step": 10698 }, { "epoch": 0.8188427981019439, "grad_norm": 0.5928476704973941, "learning_rate": 2.6695153331702893e-06, "loss": 1.4353, "step": 10699 }, { "epoch": 0.8189193326190112, "grad_norm": 0.5231797115331747, "learning_rate": 2.6689662958737556e-06, "loss": 1.0984, "step": 10700 }, { "epoch": 0.8189958671360784, "grad_norm": 0.5795683104046764, "learning_rate": 2.66841746324589e-06, "loss": 1.374, "step": 10701 }, { "epoch": 0.8190724016531455, "grad_norm": 0.8159800543448437, "learning_rate": 2.667868835320414e-06, "loss": 1.8403, "step": 10702 }, { "epoch": 0.8191489361702128, "grad_norm": 0.5538883085823516, "learning_rate": 2.667320412131039e-06, "loss": 1.4171, "step": 10703 }, { "epoch": 0.8192254706872799, "grad_norm": 0.6204864605378817, "learning_rate": 2.666772193711461e-06, "loss": 1.4355, "step": 10704 }, { "epoch": 0.8193020052043472, "grad_norm": 0.6322096722528213, "learning_rate": 2.6662241800953632e-06, "loss": 1.4192, "step": 10705 }, { "epoch": 0.8193785397214144, "grad_norm": 0.7044321670051751, "learning_rate": 2.6656763713164186e-06, "loss": 1.5555, "step": 10706 }, { "epoch": 0.8194550742384815, "grad_norm": 0.5859612169795406, "learning_rate": 2.6651287674082847e-06, "loss": 1.3846, "step": 10707 }, { "epoch": 0.8195316087555488, "grad_norm": 0.5683359792890776, "learning_rate": 2.6645813684046086e-06, "loss": 1.408, "step": 10708 }, { "epoch": 0.819608143272616, "grad_norm": 0.49325878103495013, "learning_rate": 2.664034174339025e-06, "loss": 1.0527, "step": 10709 }, { "epoch": 0.8196846777896831, "grad_norm": 0.5722418760070767, "learning_rate": 2.6634871852451538e-06, "loss": 1.2531, "step": 10710 }, { "epoch": 0.8197612123067504, "grad_norm": 0.6688503783811828, "learning_rate": 2.6629404011566044e-06, "loss": 1.693, "step": 10711 }, { "epoch": 0.8198377468238175, "grad_norm": 0.6072119376499935, "learning_rate": 2.662393822106973e-06, "loss": 1.3805, "step": 10712 }, { "epoch": 0.8199142813408847, "grad_norm": 0.6430817943776609, "learning_rate": 2.661847448129842e-06, "loss": 1.5245, "step": 10713 }, { "epoch": 0.819990815857952, "grad_norm": 0.8590095095222757, "learning_rate": 2.6613012792587833e-06, "loss": 1.5154, "step": 10714 }, { "epoch": 0.8200673503750191, "grad_norm": 0.7135023486148878, "learning_rate": 2.6607553155273545e-06, "loss": 1.8419, "step": 10715 }, { "epoch": 0.8201438848920863, "grad_norm": 0.5035308384982844, "learning_rate": 2.660209556969102e-06, "loss": 1.146, "step": 10716 }, { "epoch": 0.8202204194091536, "grad_norm": 0.6613384969352712, "learning_rate": 2.659664003617559e-06, "loss": 1.5264, "step": 10717 }, { "epoch": 0.8202969539262207, "grad_norm": 0.6237599724530688, "learning_rate": 2.659118655506245e-06, "loss": 1.3256, "step": 10718 }, { "epoch": 0.820373488443288, "grad_norm": 0.807710486366975, "learning_rate": 2.658573512668667e-06, "loss": 1.3857, "step": 10719 }, { "epoch": 0.8204500229603551, "grad_norm": 0.5818902362704117, "learning_rate": 2.658028575138323e-06, "loss": 1.2693, "step": 10720 }, { "epoch": 0.8205265574774223, "grad_norm": 0.5750010051891656, "learning_rate": 2.657483842948693e-06, "loss": 1.2596, "step": 10721 }, { "epoch": 0.8206030919944896, "grad_norm": 0.6204204338628844, "learning_rate": 2.656939316133248e-06, "loss": 1.4724, "step": 10722 }, { "epoch": 0.8206796265115567, "grad_norm": 0.7649432293340082, "learning_rate": 2.656394994725445e-06, "loss": 1.6684, "step": 10723 }, { "epoch": 0.8207561610286239, "grad_norm": 0.6128978345034565, "learning_rate": 2.655850878758729e-06, "loss": 1.3989, "step": 10724 }, { "epoch": 0.8208326955456912, "grad_norm": 0.5344794280439766, "learning_rate": 2.655306968266532e-06, "loss": 1.2167, "step": 10725 }, { "epoch": 0.8209092300627583, "grad_norm": 0.5278617228383857, "learning_rate": 2.6547632632822744e-06, "loss": 1.2823, "step": 10726 }, { "epoch": 0.8209857645798255, "grad_norm": 0.5845441628919507, "learning_rate": 2.654219763839362e-06, "loss": 1.3747, "step": 10727 }, { "epoch": 0.8210622990968927, "grad_norm": 0.6670844242822151, "learning_rate": 2.653676469971189e-06, "loss": 1.676, "step": 10728 }, { "epoch": 0.8211388336139599, "grad_norm": 0.559635361582352, "learning_rate": 2.653133381711137e-06, "loss": 1.3138, "step": 10729 }, { "epoch": 0.8212153681310271, "grad_norm": 0.6802952689972338, "learning_rate": 2.6525904990925754e-06, "loss": 1.7951, "step": 10730 }, { "epoch": 0.8212919026480943, "grad_norm": 0.6200893705532035, "learning_rate": 2.652047822148861e-06, "loss": 1.4668, "step": 10731 }, { "epoch": 0.8213684371651615, "grad_norm": 0.5458706677520966, "learning_rate": 2.6515053509133366e-06, "loss": 1.3794, "step": 10732 }, { "epoch": 0.8214449716822286, "grad_norm": 0.643911711321856, "learning_rate": 2.6509630854193338e-06, "loss": 1.434, "step": 10733 }, { "epoch": 0.8215215061992959, "grad_norm": 1.2952863953407263, "learning_rate": 2.65042102570017e-06, "loss": 1.6805, "step": 10734 }, { "epoch": 0.8215980407163631, "grad_norm": 0.772340147187538, "learning_rate": 2.6498791717891516e-06, "loss": 1.446, "step": 10735 }, { "epoch": 0.8216745752334302, "grad_norm": 0.5930016151941875, "learning_rate": 2.6493375237195723e-06, "loss": 1.5194, "step": 10736 }, { "epoch": 0.8217511097504975, "grad_norm": 0.539589856353166, "learning_rate": 2.648796081524712e-06, "loss": 1.3604, "step": 10737 }, { "epoch": 0.8218276442675647, "grad_norm": 0.7033442693800267, "learning_rate": 2.6482548452378393e-06, "loss": 1.0656, "step": 10738 }, { "epoch": 0.8219041787846318, "grad_norm": 0.6084522507208795, "learning_rate": 2.647713814892208e-06, "loss": 1.52, "step": 10739 }, { "epoch": 0.8219807133016991, "grad_norm": 0.5236401646894385, "learning_rate": 2.647172990521062e-06, "loss": 1.2362, "step": 10740 }, { "epoch": 0.8220572478187662, "grad_norm": 0.7312907438041757, "learning_rate": 2.64663237215763e-06, "loss": 1.5445, "step": 10741 }, { "epoch": 0.8221337823358335, "grad_norm": 0.6993313342924674, "learning_rate": 2.64609195983513e-06, "loss": 1.7613, "step": 10742 }, { "epoch": 0.8222103168529007, "grad_norm": 0.57840778664039, "learning_rate": 2.6455517535867663e-06, "loss": 1.3409, "step": 10743 }, { "epoch": 0.8222868513699678, "grad_norm": 0.6926819868262076, "learning_rate": 2.645011753445731e-06, "loss": 1.664, "step": 10744 }, { "epoch": 0.8223633858870351, "grad_norm": 0.5653163592597146, "learning_rate": 2.644471959445203e-06, "loss": 1.1968, "step": 10745 }, { "epoch": 0.8224399204041023, "grad_norm": 0.6380998309962096, "learning_rate": 2.643932371618347e-06, "loss": 1.6916, "step": 10746 }, { "epoch": 0.8225164549211694, "grad_norm": 0.5136292487470441, "learning_rate": 2.643392989998321e-06, "loss": 1.0922, "step": 10747 }, { "epoch": 0.8225929894382367, "grad_norm": 0.6804887576149272, "learning_rate": 2.642853814618263e-06, "loss": 1.388, "step": 10748 }, { "epoch": 0.8226695239553038, "grad_norm": 0.9489007623552774, "learning_rate": 2.642314845511303e-06, "loss": 1.4287, "step": 10749 }, { "epoch": 0.822746058472371, "grad_norm": 0.6096306374396706, "learning_rate": 2.6417760827105566e-06, "loss": 1.6884, "step": 10750 }, { "epoch": 0.8228225929894383, "grad_norm": 0.6370724753123383, "learning_rate": 2.6412375262491257e-06, "loss": 1.6172, "step": 10751 }, { "epoch": 0.8228991275065054, "grad_norm": 0.5856226840759292, "learning_rate": 2.6406991761601036e-06, "loss": 1.3019, "step": 10752 }, { "epoch": 0.8229756620235726, "grad_norm": 0.5949961697342941, "learning_rate": 2.640161032476566e-06, "loss": 1.4228, "step": 10753 }, { "epoch": 0.8230521965406399, "grad_norm": 0.5746833261395898, "learning_rate": 2.639623095231578e-06, "loss": 1.4758, "step": 10754 }, { "epoch": 0.823128731057707, "grad_norm": 0.5708710191650376, "learning_rate": 2.6390853644581937e-06, "loss": 1.2432, "step": 10755 }, { "epoch": 0.8232052655747742, "grad_norm": 0.6805013927741066, "learning_rate": 2.6385478401894513e-06, "loss": 1.3467, "step": 10756 }, { "epoch": 0.8232818000918414, "grad_norm": 0.6583800705190934, "learning_rate": 2.638010522458377e-06, "loss": 1.3659, "step": 10757 }, { "epoch": 0.8233583346089086, "grad_norm": 0.5731167336086911, "learning_rate": 2.6374734112979883e-06, "loss": 1.3885, "step": 10758 }, { "epoch": 0.8234348691259759, "grad_norm": 0.5521970151666873, "learning_rate": 2.6369365067412857e-06, "loss": 1.1883, "step": 10759 }, { "epoch": 0.823511403643043, "grad_norm": 0.9033830869052561, "learning_rate": 2.6363998088212565e-06, "loss": 1.8587, "step": 10760 }, { "epoch": 0.8235879381601102, "grad_norm": 0.638167089825247, "learning_rate": 2.6358633175708793e-06, "loss": 1.4793, "step": 10761 }, { "epoch": 0.8236644726771775, "grad_norm": 0.6006838531029484, "learning_rate": 2.635327033023116e-06, "loss": 1.6111, "step": 10762 }, { "epoch": 0.8237410071942446, "grad_norm": 0.608167643159901, "learning_rate": 2.6347909552109185e-06, "loss": 1.3685, "step": 10763 }, { "epoch": 0.8238175417113118, "grad_norm": 0.753899464575582, "learning_rate": 2.6342550841672255e-06, "loss": 1.4725, "step": 10764 }, { "epoch": 0.823894076228379, "grad_norm": 0.5067074530183914, "learning_rate": 2.6337194199249605e-06, "loss": 1.0319, "step": 10765 }, { "epoch": 0.8239706107454462, "grad_norm": 0.5474335064058226, "learning_rate": 2.6331839625170388e-06, "loss": 1.055, "step": 10766 }, { "epoch": 0.8240471452625134, "grad_norm": 0.5379668508795736, "learning_rate": 2.632648711976358e-06, "loss": 1.2057, "step": 10767 }, { "epoch": 0.8241236797795806, "grad_norm": 0.6139641196390199, "learning_rate": 2.6321136683358073e-06, "loss": 1.4297, "step": 10768 }, { "epoch": 0.8242002142966478, "grad_norm": 0.6651888142479007, "learning_rate": 2.6315788316282608e-06, "loss": 1.8023, "step": 10769 }, { "epoch": 0.8242767488137149, "grad_norm": 0.6371395687917678, "learning_rate": 2.6310442018865814e-06, "loss": 1.4859, "step": 10770 }, { "epoch": 0.8243532833307822, "grad_norm": 0.5743385811373937, "learning_rate": 2.630509779143616e-06, "loss": 1.1947, "step": 10771 }, { "epoch": 0.8244298178478494, "grad_norm": 0.5266172924329083, "learning_rate": 2.629975563432204e-06, "loss": 1.3308, "step": 10772 }, { "epoch": 0.8245063523649165, "grad_norm": 0.5663552603891429, "learning_rate": 2.6294415547851664e-06, "loss": 1.195, "step": 10773 }, { "epoch": 0.8245828868819838, "grad_norm": 0.514224158445032, "learning_rate": 2.6289077532353168e-06, "loss": 1.171, "step": 10774 }, { "epoch": 0.824659421399051, "grad_norm": 0.6648233337940823, "learning_rate": 2.6283741588154528e-06, "loss": 1.4565, "step": 10775 }, { "epoch": 0.8247359559161181, "grad_norm": 0.6326707992548043, "learning_rate": 2.6278407715583593e-06, "loss": 1.8427, "step": 10776 }, { "epoch": 0.8248124904331854, "grad_norm": 0.6256019747625691, "learning_rate": 2.6273075914968097e-06, "loss": 1.3888, "step": 10777 }, { "epoch": 0.8248890249502525, "grad_norm": 0.6658800218132483, "learning_rate": 2.626774618663564e-06, "loss": 1.5259, "step": 10778 }, { "epoch": 0.8249655594673198, "grad_norm": 0.5497477426524271, "learning_rate": 2.626241853091369e-06, "loss": 1.219, "step": 10779 }, { "epoch": 0.825042093984387, "grad_norm": 0.5674269183117391, "learning_rate": 2.6257092948129616e-06, "loss": 1.3412, "step": 10780 }, { "epoch": 0.8251186285014541, "grad_norm": 0.6687366802336953, "learning_rate": 2.625176943861061e-06, "loss": 1.7052, "step": 10781 }, { "epoch": 0.8251951630185214, "grad_norm": 0.7183103949148829, "learning_rate": 2.6246448002683787e-06, "loss": 1.3771, "step": 10782 }, { "epoch": 0.8252716975355886, "grad_norm": 0.6093208778676595, "learning_rate": 2.62411286406761e-06, "loss": 1.3147, "step": 10783 }, { "epoch": 0.8253482320526557, "grad_norm": 0.5865359643616895, "learning_rate": 2.623581135291439e-06, "loss": 1.3017, "step": 10784 }, { "epoch": 0.825424766569723, "grad_norm": 0.5937867403433306, "learning_rate": 2.623049613972537e-06, "loss": 1.3419, "step": 10785 }, { "epoch": 0.8255013010867901, "grad_norm": 0.6177233817806279, "learning_rate": 2.6225183001435617e-06, "loss": 1.4497, "step": 10786 }, { "epoch": 0.8255778356038573, "grad_norm": 0.6701215085238726, "learning_rate": 2.6219871938371583e-06, "loss": 1.6279, "step": 10787 }, { "epoch": 0.8256543701209246, "grad_norm": 0.5443771251703675, "learning_rate": 2.62145629508596e-06, "loss": 1.3533, "step": 10788 }, { "epoch": 0.8257309046379917, "grad_norm": 0.5478115707624939, "learning_rate": 2.620925603922588e-06, "loss": 1.3332, "step": 10789 }, { "epoch": 0.8258074391550589, "grad_norm": 0.6109063368641707, "learning_rate": 2.620395120379646e-06, "loss": 1.5108, "step": 10790 }, { "epoch": 0.8258839736721262, "grad_norm": 0.6578074759311929, "learning_rate": 2.619864844489733e-06, "loss": 1.622, "step": 10791 }, { "epoch": 0.8259605081891933, "grad_norm": 0.6199677890532762, "learning_rate": 2.6193347762854277e-06, "loss": 1.4924, "step": 10792 }, { "epoch": 0.8260370427062605, "grad_norm": 0.7868041664451246, "learning_rate": 2.6188049157993005e-06, "loss": 1.3188, "step": 10793 }, { "epoch": 0.8261135772233277, "grad_norm": 0.5072214263806681, "learning_rate": 2.6182752630639075e-06, "loss": 1.196, "step": 10794 }, { "epoch": 0.8261901117403949, "grad_norm": 0.5811140600544924, "learning_rate": 2.61774581811179e-06, "loss": 1.2902, "step": 10795 }, { "epoch": 0.8262666462574622, "grad_norm": 0.5783319497208135, "learning_rate": 2.617216580975482e-06, "loss": 1.5291, "step": 10796 }, { "epoch": 0.8263431807745293, "grad_norm": 0.5483669154991011, "learning_rate": 2.6166875516874995e-06, "loss": 1.1337, "step": 10797 }, { "epoch": 0.8264197152915965, "grad_norm": 1.0522894929756963, "learning_rate": 2.616158730280348e-06, "loss": 1.6861, "step": 10798 }, { "epoch": 0.8264962498086637, "grad_norm": 0.5945658985252887, "learning_rate": 2.6156301167865205e-06, "loss": 1.3636, "step": 10799 }, { "epoch": 0.8265727843257309, "grad_norm": 0.6027793974516317, "learning_rate": 2.6151017112384954e-06, "loss": 1.3128, "step": 10800 }, { "epoch": 0.8266493188427981, "grad_norm": 0.5716248335608467, "learning_rate": 2.6145735136687387e-06, "loss": 1.4138, "step": 10801 }, { "epoch": 0.8267258533598653, "grad_norm": 0.5130636244308019, "learning_rate": 2.614045524109708e-06, "loss": 1.3053, "step": 10802 }, { "epoch": 0.8268023878769325, "grad_norm": 0.5831750436792337, "learning_rate": 2.613517742593841e-06, "loss": 1.3014, "step": 10803 }, { "epoch": 0.8268789223939997, "grad_norm": 0.565792154708273, "learning_rate": 2.612990169153569e-06, "loss": 1.3592, "step": 10804 }, { "epoch": 0.8269554569110669, "grad_norm": 0.5856525262395734, "learning_rate": 2.612462803821305e-06, "loss": 1.4819, "step": 10805 }, { "epoch": 0.8270319914281341, "grad_norm": 0.6158769212742677, "learning_rate": 2.6119356466294533e-06, "loss": 1.4269, "step": 10806 }, { "epoch": 0.8271085259452012, "grad_norm": 0.6560456947567649, "learning_rate": 2.6114086976104035e-06, "loss": 1.6272, "step": 10807 }, { "epoch": 0.8271850604622685, "grad_norm": 0.6092960267662922, "learning_rate": 2.6108819567965337e-06, "loss": 1.4654, "step": 10808 }, { "epoch": 0.8272615949793357, "grad_norm": 0.603421038138574, "learning_rate": 2.6103554242202088e-06, "loss": 1.4829, "step": 10809 }, { "epoch": 0.8273381294964028, "grad_norm": 0.5864448502939409, "learning_rate": 2.6098290999137798e-06, "loss": 1.1385, "step": 10810 }, { "epoch": 0.8274146640134701, "grad_norm": 0.6211096635977099, "learning_rate": 2.609302983909585e-06, "loss": 1.4432, "step": 10811 }, { "epoch": 0.8274911985305373, "grad_norm": 0.6318275448412144, "learning_rate": 2.6087770762399502e-06, "loss": 1.4721, "step": 10812 }, { "epoch": 0.8275677330476044, "grad_norm": 0.6502625314938966, "learning_rate": 2.608251376937191e-06, "loss": 1.6485, "step": 10813 }, { "epoch": 0.8276442675646717, "grad_norm": 0.7722044788349174, "learning_rate": 2.6077258860336057e-06, "loss": 1.78, "step": 10814 }, { "epoch": 0.8277208020817388, "grad_norm": 1.0144548570542475, "learning_rate": 2.6072006035614837e-06, "loss": 1.5916, "step": 10815 }, { "epoch": 0.827797336598806, "grad_norm": 0.766638932830751, "learning_rate": 2.6066755295530986e-06, "loss": 1.4643, "step": 10816 }, { "epoch": 0.8278738711158733, "grad_norm": 0.6082163346946611, "learning_rate": 2.6061506640407125e-06, "loss": 1.518, "step": 10817 }, { "epoch": 0.8279504056329404, "grad_norm": 0.584745835979046, "learning_rate": 2.6056260070565764e-06, "loss": 1.3142, "step": 10818 }, { "epoch": 0.8280269401500077, "grad_norm": 0.5346760600426275, "learning_rate": 2.6051015586329252e-06, "loss": 1.0321, "step": 10819 }, { "epoch": 0.8281034746670749, "grad_norm": 0.5246749338600551, "learning_rate": 2.6045773188019836e-06, "loss": 1.0925, "step": 10820 }, { "epoch": 0.828180009184142, "grad_norm": 0.6843788589872994, "learning_rate": 2.604053287595961e-06, "loss": 1.3387, "step": 10821 }, { "epoch": 0.8282565437012093, "grad_norm": 0.9222695631471703, "learning_rate": 2.6035294650470567e-06, "loss": 1.6027, "step": 10822 }, { "epoch": 0.8283330782182764, "grad_norm": 0.5671044467297549, "learning_rate": 2.6030058511874552e-06, "loss": 1.1776, "step": 10823 }, { "epoch": 0.8284096127353436, "grad_norm": 0.6472375812761341, "learning_rate": 2.6024824460493296e-06, "loss": 1.4303, "step": 10824 }, { "epoch": 0.8284861472524109, "grad_norm": 0.6406250805002882, "learning_rate": 2.60195924966484e-06, "loss": 1.4534, "step": 10825 }, { "epoch": 0.828562681769478, "grad_norm": 0.6236423833871559, "learning_rate": 2.6014362620661315e-06, "loss": 1.4979, "step": 10826 }, { "epoch": 0.8286392162865452, "grad_norm": 0.6565356474620194, "learning_rate": 2.600913483285339e-06, "loss": 1.7263, "step": 10827 }, { "epoch": 0.8287157508036125, "grad_norm": 0.6129921521570983, "learning_rate": 2.6003909133545825e-06, "loss": 1.535, "step": 10828 }, { "epoch": 0.8287922853206796, "grad_norm": 0.58859893563254, "learning_rate": 2.5998685523059726e-06, "loss": 1.2931, "step": 10829 }, { "epoch": 0.8288688198377469, "grad_norm": 0.6023698457140084, "learning_rate": 2.599346400171603e-06, "loss": 1.4593, "step": 10830 }, { "epoch": 0.828945354354814, "grad_norm": 0.5526567881275577, "learning_rate": 2.598824456983556e-06, "loss": 1.229, "step": 10831 }, { "epoch": 0.8290218888718812, "grad_norm": 0.6304404026336896, "learning_rate": 2.5983027227739023e-06, "loss": 1.5575, "step": 10832 }, { "epoch": 0.8290984233889485, "grad_norm": 0.6312393099259955, "learning_rate": 2.5977811975746985e-06, "loss": 1.6094, "step": 10833 }, { "epoch": 0.8291749579060156, "grad_norm": 0.5844532582299822, "learning_rate": 2.5972598814179885e-06, "loss": 1.2533, "step": 10834 }, { "epoch": 0.8292514924230828, "grad_norm": 0.6991001557003178, "learning_rate": 2.596738774335804e-06, "loss": 1.5189, "step": 10835 }, { "epoch": 0.82932802694015, "grad_norm": 0.6424870394796987, "learning_rate": 2.596217876360163e-06, "loss": 1.3874, "step": 10836 }, { "epoch": 0.8294045614572172, "grad_norm": 0.5524440430175348, "learning_rate": 2.595697187523071e-06, "loss": 1.3298, "step": 10837 }, { "epoch": 0.8294810959742844, "grad_norm": 0.5657298960359738, "learning_rate": 2.5951767078565206e-06, "loss": 1.3665, "step": 10838 }, { "epoch": 0.8295576304913516, "grad_norm": 0.6899792399172205, "learning_rate": 2.594656437392491e-06, "loss": 1.9066, "step": 10839 }, { "epoch": 0.8296341650084188, "grad_norm": 0.909856700298607, "learning_rate": 2.59413637616295e-06, "loss": 1.6611, "step": 10840 }, { "epoch": 0.829710699525486, "grad_norm": 0.5846693339147033, "learning_rate": 2.593616524199853e-06, "loss": 1.4914, "step": 10841 }, { "epoch": 0.8297872340425532, "grad_norm": 0.7799559926864786, "learning_rate": 2.5930968815351387e-06, "loss": 1.8409, "step": 10842 }, { "epoch": 0.8298637685596204, "grad_norm": 0.6156201311238629, "learning_rate": 2.592577448200737e-06, "loss": 1.4921, "step": 10843 }, { "epoch": 0.8299403030766875, "grad_norm": 0.6196622437775372, "learning_rate": 2.592058224228564e-06, "loss": 1.4003, "step": 10844 }, { "epoch": 0.8300168375937548, "grad_norm": 0.5509836564949592, "learning_rate": 2.59153920965052e-06, "loss": 1.1153, "step": 10845 }, { "epoch": 0.830093372110822, "grad_norm": 0.7455844807821401, "learning_rate": 2.5910204044984964e-06, "loss": 1.7129, "step": 10846 }, { "epoch": 0.8301699066278891, "grad_norm": 0.5394926435131427, "learning_rate": 2.5905018088043715e-06, "loss": 1.2773, "step": 10847 }, { "epoch": 0.8302464411449564, "grad_norm": 0.680861342265784, "learning_rate": 2.589983422600007e-06, "loss": 1.6199, "step": 10848 }, { "epoch": 0.8303229756620236, "grad_norm": 0.8565218872628384, "learning_rate": 2.5894652459172553e-06, "loss": 1.638, "step": 10849 }, { "epoch": 0.8303995101790907, "grad_norm": 0.6836308659166787, "learning_rate": 2.5889472787879538e-06, "loss": 1.4189, "step": 10850 }, { "epoch": 0.830476044696158, "grad_norm": 0.6179924917031738, "learning_rate": 2.5884295212439293e-06, "loss": 1.3821, "step": 10851 }, { "epoch": 0.8305525792132251, "grad_norm": 0.6372428214003011, "learning_rate": 2.587911973316994e-06, "loss": 1.4688, "step": 10852 }, { "epoch": 0.8306291137302924, "grad_norm": 0.7189402871846672, "learning_rate": 2.5873946350389467e-06, "loss": 1.6373, "step": 10853 }, { "epoch": 0.8307056482473596, "grad_norm": 0.47879668438252215, "learning_rate": 2.5868775064415745e-06, "loss": 1.0731, "step": 10854 }, { "epoch": 0.8307821827644267, "grad_norm": 0.6910956584828382, "learning_rate": 2.586360587556652e-06, "loss": 1.7785, "step": 10855 }, { "epoch": 0.830858717281494, "grad_norm": 0.6203931557822608, "learning_rate": 2.58584387841594e-06, "loss": 1.3388, "step": 10856 }, { "epoch": 0.8309352517985612, "grad_norm": 0.585875064979404, "learning_rate": 2.585327379051186e-06, "loss": 1.3904, "step": 10857 }, { "epoch": 0.8310117863156283, "grad_norm": 0.4992457205644284, "learning_rate": 2.5848110894941265e-06, "loss": 0.9236, "step": 10858 }, { "epoch": 0.8310883208326956, "grad_norm": 0.8883579368328782, "learning_rate": 2.584295009776483e-06, "loss": 1.4502, "step": 10859 }, { "epoch": 0.8311648553497627, "grad_norm": 0.5994010412855353, "learning_rate": 2.583779139929965e-06, "loss": 1.3583, "step": 10860 }, { "epoch": 0.8312413898668299, "grad_norm": 0.6356746190186389, "learning_rate": 2.5832634799862683e-06, "loss": 1.4815, "step": 10861 }, { "epoch": 0.8313179243838972, "grad_norm": 0.532873228190914, "learning_rate": 2.582748029977078e-06, "loss": 1.1918, "step": 10862 }, { "epoch": 0.8313944589009643, "grad_norm": 0.6158494262521299, "learning_rate": 2.582232789934065e-06, "loss": 1.3213, "step": 10863 }, { "epoch": 0.8314709934180315, "grad_norm": 0.7416722189850332, "learning_rate": 2.581717759888886e-06, "loss": 1.3868, "step": 10864 }, { "epoch": 0.8315475279350988, "grad_norm": 0.6521345432407467, "learning_rate": 2.5812029398731865e-06, "loss": 1.4573, "step": 10865 }, { "epoch": 0.8316240624521659, "grad_norm": 0.7607401345568989, "learning_rate": 2.5806883299185992e-06, "loss": 1.1964, "step": 10866 }, { "epoch": 0.8317005969692332, "grad_norm": 0.6182206503166533, "learning_rate": 2.5801739300567414e-06, "loss": 1.5954, "step": 10867 }, { "epoch": 0.8317771314863003, "grad_norm": 0.6061225910292789, "learning_rate": 2.5796597403192214e-06, "loss": 1.4245, "step": 10868 }, { "epoch": 0.8318536660033675, "grad_norm": 0.5623725453514669, "learning_rate": 2.5791457607376323e-06, "loss": 1.2969, "step": 10869 }, { "epoch": 0.8319302005204348, "grad_norm": 0.7434802832442862, "learning_rate": 2.5786319913435543e-06, "loss": 1.2845, "step": 10870 }, { "epoch": 0.8320067350375019, "grad_norm": 0.7425512649989718, "learning_rate": 2.578118432168554e-06, "loss": 1.3218, "step": 10871 }, { "epoch": 0.8320832695545691, "grad_norm": 0.5837723165862347, "learning_rate": 2.5776050832441856e-06, "loss": 1.2996, "step": 10872 }, { "epoch": 0.8321598040716363, "grad_norm": 0.5665151232481761, "learning_rate": 2.5770919446019934e-06, "loss": 1.3895, "step": 10873 }, { "epoch": 0.8322363385887035, "grad_norm": 0.5957395963166735, "learning_rate": 2.5765790162735045e-06, "loss": 1.242, "step": 10874 }, { "epoch": 0.8323128731057707, "grad_norm": 0.6150973764002148, "learning_rate": 2.576066298290235e-06, "loss": 1.5405, "step": 10875 }, { "epoch": 0.8323894076228379, "grad_norm": 0.5429604715303638, "learning_rate": 2.5755537906836876e-06, "loss": 1.2, "step": 10876 }, { "epoch": 0.8324659421399051, "grad_norm": 0.5313384956230269, "learning_rate": 2.575041493485353e-06, "loss": 1.2986, "step": 10877 }, { "epoch": 0.8325424766569723, "grad_norm": 0.5581938786479943, "learning_rate": 2.5745294067267067e-06, "loss": 1.2457, "step": 10878 }, { "epoch": 0.8326190111740395, "grad_norm": 0.48610033060010943, "learning_rate": 2.5740175304392146e-06, "loss": 1.0435, "step": 10879 }, { "epoch": 0.8326955456911067, "grad_norm": 0.5323676790054293, "learning_rate": 2.573505864654327e-06, "loss": 1.1279, "step": 10880 }, { "epoch": 0.8327720802081738, "grad_norm": 0.5590530347774755, "learning_rate": 2.5729944094034826e-06, "loss": 1.1837, "step": 10881 }, { "epoch": 0.8328486147252411, "grad_norm": 2.7667186779832913, "learning_rate": 2.5724831647181063e-06, "loss": 1.3609, "step": 10882 }, { "epoch": 0.8329251492423083, "grad_norm": 0.5840351871496321, "learning_rate": 2.5719721306296098e-06, "loss": 1.2797, "step": 10883 }, { "epoch": 0.8330016837593754, "grad_norm": 0.7519477159139349, "learning_rate": 2.571461307169394e-06, "loss": 1.5322, "step": 10884 }, { "epoch": 0.8330782182764427, "grad_norm": 0.5464330687385061, "learning_rate": 2.570950694368845e-06, "loss": 1.291, "step": 10885 }, { "epoch": 0.8331547527935099, "grad_norm": 0.6276568959463157, "learning_rate": 2.5704402922593365e-06, "loss": 1.5513, "step": 10886 }, { "epoch": 0.833231287310577, "grad_norm": 0.7624753926142387, "learning_rate": 2.569930100872229e-06, "loss": 1.6727, "step": 10887 }, { "epoch": 0.8333078218276443, "grad_norm": 0.6689983877812611, "learning_rate": 2.5694201202388684e-06, "loss": 1.241, "step": 10888 }, { "epoch": 0.8333843563447114, "grad_norm": 0.64026292689976, "learning_rate": 2.568910350390593e-06, "loss": 1.4695, "step": 10889 }, { "epoch": 0.8334608908617787, "grad_norm": 0.507315065934358, "learning_rate": 2.5684007913587217e-06, "loss": 0.9408, "step": 10890 }, { "epoch": 0.8335374253788459, "grad_norm": 0.5039966459123207, "learning_rate": 2.567891443174565e-06, "loss": 1.1835, "step": 10891 }, { "epoch": 0.833613959895913, "grad_norm": 0.6060138500348399, "learning_rate": 2.5673823058694173e-06, "loss": 1.2873, "step": 10892 }, { "epoch": 0.8336904944129803, "grad_norm": 0.8866258041315205, "learning_rate": 2.566873379474564e-06, "loss": 1.6794, "step": 10893 }, { "epoch": 0.8337670289300475, "grad_norm": 0.4826932009525542, "learning_rate": 2.56636466402127e-06, "loss": 1.1534, "step": 10894 }, { "epoch": 0.8338435634471146, "grad_norm": 0.646432238320399, "learning_rate": 2.565856159540798e-06, "loss": 1.6875, "step": 10895 }, { "epoch": 0.8339200979641819, "grad_norm": 0.8053913233209171, "learning_rate": 2.5653478660643898e-06, "loss": 1.6055, "step": 10896 }, { "epoch": 0.833996632481249, "grad_norm": 0.5203017338962523, "learning_rate": 2.5648397836232765e-06, "loss": 1.1416, "step": 10897 }, { "epoch": 0.8340731669983162, "grad_norm": 0.6424764794622105, "learning_rate": 2.564331912248676e-06, "loss": 1.583, "step": 10898 }, { "epoch": 0.8341497015153835, "grad_norm": 0.5461227100769613, "learning_rate": 2.563824251971793e-06, "loss": 1.2667, "step": 10899 }, { "epoch": 0.8342262360324506, "grad_norm": 0.6890174628873987, "learning_rate": 2.5633168028238203e-06, "loss": 1.5253, "step": 10900 }, { "epoch": 0.8343027705495178, "grad_norm": 0.7103500327303175, "learning_rate": 2.5628095648359375e-06, "loss": 1.4246, "step": 10901 }, { "epoch": 0.834379305066585, "grad_norm": 0.530280736496186, "learning_rate": 2.5623025380393096e-06, "loss": 1.1799, "step": 10902 }, { "epoch": 0.8344558395836522, "grad_norm": 0.5794724495718926, "learning_rate": 2.5617957224650902e-06, "loss": 1.3813, "step": 10903 }, { "epoch": 0.8345323741007195, "grad_norm": 0.6546373320423501, "learning_rate": 2.561289118144421e-06, "loss": 1.5593, "step": 10904 }, { "epoch": 0.8346089086177866, "grad_norm": 0.59046286612036, "learning_rate": 2.560782725108426e-06, "loss": 1.3998, "step": 10905 }, { "epoch": 0.8346854431348538, "grad_norm": 0.5971429300521529, "learning_rate": 2.5602765433882236e-06, "loss": 1.4627, "step": 10906 }, { "epoch": 0.8347619776519211, "grad_norm": 0.5899633493073033, "learning_rate": 2.559770573014913e-06, "loss": 1.2608, "step": 10907 }, { "epoch": 0.8348385121689882, "grad_norm": 0.6498608931350572, "learning_rate": 2.5592648140195824e-06, "loss": 1.1769, "step": 10908 }, { "epoch": 0.8349150466860554, "grad_norm": 0.6094552604749665, "learning_rate": 2.5587592664333075e-06, "loss": 1.476, "step": 10909 }, { "epoch": 0.8349915812031226, "grad_norm": 0.7260702726235336, "learning_rate": 2.55825393028715e-06, "loss": 1.3085, "step": 10910 }, { "epoch": 0.8350681157201898, "grad_norm": 0.7133148460142585, "learning_rate": 2.557748805612161e-06, "loss": 1.6609, "step": 10911 }, { "epoch": 0.835144650237257, "grad_norm": 0.6178941381460675, "learning_rate": 2.5572438924393757e-06, "loss": 1.4803, "step": 10912 }, { "epoch": 0.8352211847543242, "grad_norm": 0.5091495597217395, "learning_rate": 2.556739190799817e-06, "loss": 1.0357, "step": 10913 }, { "epoch": 0.8352977192713914, "grad_norm": 0.6345788032037625, "learning_rate": 2.556234700724496e-06, "loss": 1.5943, "step": 10914 }, { "epoch": 0.8353742537884586, "grad_norm": 0.6672692907546666, "learning_rate": 2.5557304222444096e-06, "loss": 1.7224, "step": 10915 }, { "epoch": 0.8354507883055258, "grad_norm": 0.7218888226365296, "learning_rate": 2.555226355390542e-06, "loss": 1.2464, "step": 10916 }, { "epoch": 0.835527322822593, "grad_norm": 0.5508092946784034, "learning_rate": 2.5547225001938657e-06, "loss": 1.2403, "step": 10917 }, { "epoch": 0.8356038573396601, "grad_norm": 0.5815115506693953, "learning_rate": 2.554218856685339e-06, "loss": 1.1673, "step": 10918 }, { "epoch": 0.8356803918567274, "grad_norm": 0.58979245837935, "learning_rate": 2.553715424895906e-06, "loss": 1.4048, "step": 10919 }, { "epoch": 0.8357569263737946, "grad_norm": 0.6092140984583262, "learning_rate": 2.5532122048565e-06, "loss": 1.3555, "step": 10920 }, { "epoch": 0.8358334608908617, "grad_norm": 0.5709523094099868, "learning_rate": 2.55270919659804e-06, "loss": 1.3631, "step": 10921 }, { "epoch": 0.835909995407929, "grad_norm": 0.5679553447458332, "learning_rate": 2.5522064001514324e-06, "loss": 1.3079, "step": 10922 }, { "epoch": 0.8359865299249962, "grad_norm": 0.6269435886306415, "learning_rate": 2.5517038155475704e-06, "loss": 1.4548, "step": 10923 }, { "epoch": 0.8360630644420634, "grad_norm": 0.5142142172780357, "learning_rate": 2.551201442817335e-06, "loss": 1.2011, "step": 10924 }, { "epoch": 0.8361395989591306, "grad_norm": 0.592519957385858, "learning_rate": 2.550699281991592e-06, "loss": 1.3815, "step": 10925 }, { "epoch": 0.8362161334761977, "grad_norm": 0.5649981842833227, "learning_rate": 2.550197333101197e-06, "loss": 1.2986, "step": 10926 }, { "epoch": 0.836292667993265, "grad_norm": 0.6390463977976186, "learning_rate": 2.5496955961769897e-06, "loss": 1.5683, "step": 10927 }, { "epoch": 0.8363692025103322, "grad_norm": 0.5841785303413741, "learning_rate": 2.549194071249801e-06, "loss": 1.312, "step": 10928 }, { "epoch": 0.8364457370273993, "grad_norm": 0.6083559854366375, "learning_rate": 2.548692758350444e-06, "loss": 1.5091, "step": 10929 }, { "epoch": 0.8365222715444666, "grad_norm": 0.9183136031240752, "learning_rate": 2.548191657509721e-06, "loss": 1.4803, "step": 10930 }, { "epoch": 0.8365988060615338, "grad_norm": 0.5572312377679927, "learning_rate": 2.5476907687584216e-06, "loss": 1.3439, "step": 10931 }, { "epoch": 0.8366753405786009, "grad_norm": 0.5497340677496168, "learning_rate": 2.5471900921273218e-06, "loss": 1.3538, "step": 10932 }, { "epoch": 0.8367518750956682, "grad_norm": 0.5846195107498214, "learning_rate": 2.5466896276471847e-06, "loss": 1.5025, "step": 10933 }, { "epoch": 0.8368284096127353, "grad_norm": 0.9051215264454338, "learning_rate": 2.5461893753487603e-06, "loss": 1.6207, "step": 10934 }, { "epoch": 0.8369049441298025, "grad_norm": 0.6285368216017025, "learning_rate": 2.5456893352627865e-06, "loss": 1.5589, "step": 10935 }, { "epoch": 0.8369814786468698, "grad_norm": 0.6049434487532461, "learning_rate": 2.5451895074199856e-06, "loss": 1.2887, "step": 10936 }, { "epoch": 0.8370580131639369, "grad_norm": 0.7117173909565537, "learning_rate": 2.5446898918510693e-06, "loss": 1.8432, "step": 10937 }, { "epoch": 0.8371345476810041, "grad_norm": 0.5867021052660466, "learning_rate": 2.5441904885867354e-06, "loss": 1.3576, "step": 10938 }, { "epoch": 0.8372110821980713, "grad_norm": 0.5588901465985427, "learning_rate": 2.543691297657668e-06, "loss": 1.2756, "step": 10939 }, { "epoch": 0.8372876167151385, "grad_norm": 0.6521212101900992, "learning_rate": 2.5431923190945406e-06, "loss": 1.3177, "step": 10940 }, { "epoch": 0.8373641512322058, "grad_norm": 0.8528567439694311, "learning_rate": 2.5426935529280116e-06, "loss": 1.339, "step": 10941 }, { "epoch": 0.8374406857492729, "grad_norm": 0.6697470623099331, "learning_rate": 2.542194999188725e-06, "loss": 1.6446, "step": 10942 }, { "epoch": 0.8375172202663401, "grad_norm": 0.5615314931632359, "learning_rate": 2.5416966579073156e-06, "loss": 1.3064, "step": 10943 }, { "epoch": 0.8375937547834074, "grad_norm": 0.9477526922360836, "learning_rate": 2.5411985291144015e-06, "loss": 1.4685, "step": 10944 }, { "epoch": 0.8376702893004745, "grad_norm": 0.655527589290132, "learning_rate": 2.5407006128405893e-06, "loss": 1.6472, "step": 10945 }, { "epoch": 0.8377468238175417, "grad_norm": 0.5529846414478106, "learning_rate": 2.540202909116473e-06, "loss": 1.2734, "step": 10946 }, { "epoch": 0.8378233583346089, "grad_norm": 0.8923673120041545, "learning_rate": 2.5397054179726328e-06, "loss": 1.6341, "step": 10947 }, { "epoch": 0.8378998928516761, "grad_norm": 1.0254514576596727, "learning_rate": 2.5392081394396357e-06, "loss": 1.5651, "step": 10948 }, { "epoch": 0.8379764273687433, "grad_norm": 0.7051525398985371, "learning_rate": 2.538711073548036e-06, "loss": 1.6073, "step": 10949 }, { "epoch": 0.8380529618858105, "grad_norm": 0.6041219636958441, "learning_rate": 2.5382142203283756e-06, "loss": 1.5185, "step": 10950 }, { "epoch": 0.8381294964028777, "grad_norm": 0.6233756406991231, "learning_rate": 2.5377175798111825e-06, "loss": 1.5092, "step": 10951 }, { "epoch": 0.8382060309199449, "grad_norm": 0.6504521310257713, "learning_rate": 2.5372211520269718e-06, "loss": 1.5269, "step": 10952 }, { "epoch": 0.8382825654370121, "grad_norm": 0.7760840258345172, "learning_rate": 2.5367249370062447e-06, "loss": 1.7617, "step": 10953 }, { "epoch": 0.8383590999540793, "grad_norm": 0.9462139833445758, "learning_rate": 2.53622893477949e-06, "loss": 1.3626, "step": 10954 }, { "epoch": 0.8384356344711464, "grad_norm": 0.6983451371951125, "learning_rate": 2.5357331453771854e-06, "loss": 1.5135, "step": 10955 }, { "epoch": 0.8385121689882137, "grad_norm": 0.5275456833035325, "learning_rate": 2.5352375688297927e-06, "loss": 1.1173, "step": 10956 }, { "epoch": 0.8385887035052809, "grad_norm": 0.6313391848995152, "learning_rate": 2.5347422051677605e-06, "loss": 1.6862, "step": 10957 }, { "epoch": 0.838665238022348, "grad_norm": 0.6971708295555139, "learning_rate": 2.5342470544215277e-06, "loss": 1.5606, "step": 10958 }, { "epoch": 0.8387417725394153, "grad_norm": 0.5949655361134566, "learning_rate": 2.5337521166215152e-06, "loss": 1.4109, "step": 10959 }, { "epoch": 0.8388183070564825, "grad_norm": 0.5791212305025129, "learning_rate": 2.5332573917981347e-06, "loss": 1.3317, "step": 10960 }, { "epoch": 0.8388948415735497, "grad_norm": 0.5729003716645064, "learning_rate": 2.5327628799817843e-06, "loss": 1.1348, "step": 10961 }, { "epoch": 0.8389713760906169, "grad_norm": 0.6712726589339251, "learning_rate": 2.5322685812028476e-06, "loss": 1.7282, "step": 10962 }, { "epoch": 0.839047910607684, "grad_norm": 0.5328013611884134, "learning_rate": 2.531774495491696e-06, "loss": 1.2783, "step": 10963 }, { "epoch": 0.8391244451247513, "grad_norm": 0.5571784122045838, "learning_rate": 2.531280622878687e-06, "loss": 1.3416, "step": 10964 }, { "epoch": 0.8392009796418185, "grad_norm": 0.4951540632357696, "learning_rate": 2.530786963394166e-06, "loss": 1.1615, "step": 10965 }, { "epoch": 0.8392775141588856, "grad_norm": 0.6656104619807667, "learning_rate": 2.530293517068465e-06, "loss": 1.2495, "step": 10966 }, { "epoch": 0.8393540486759529, "grad_norm": 0.543599095364734, "learning_rate": 2.5298002839319037e-06, "loss": 1.2885, "step": 10967 }, { "epoch": 0.83943058319302, "grad_norm": 0.6586738811645869, "learning_rate": 2.529307264014786e-06, "loss": 1.6906, "step": 10968 }, { "epoch": 0.8395071177100872, "grad_norm": 0.6548132304017453, "learning_rate": 2.5288144573474056e-06, "loss": 1.6052, "step": 10969 }, { "epoch": 0.8395836522271545, "grad_norm": 0.6047354414933931, "learning_rate": 2.5283218639600413e-06, "loss": 1.3949, "step": 10970 }, { "epoch": 0.8396601867442216, "grad_norm": 0.6566672062863793, "learning_rate": 2.52782948388296e-06, "loss": 1.5511, "step": 10971 }, { "epoch": 0.8397367212612888, "grad_norm": 0.6908784927605384, "learning_rate": 2.527337317146415e-06, "loss": 1.4839, "step": 10972 }, { "epoch": 0.8398132557783561, "grad_norm": 0.6932931685500926, "learning_rate": 2.526845363780647e-06, "loss": 1.4273, "step": 10973 }, { "epoch": 0.8398897902954232, "grad_norm": 0.6267815486855753, "learning_rate": 2.5263536238158825e-06, "loss": 1.6045, "step": 10974 }, { "epoch": 0.8399663248124905, "grad_norm": 0.5959944592642943, "learning_rate": 2.5258620972823363e-06, "loss": 1.4538, "step": 10975 }, { "epoch": 0.8400428593295576, "grad_norm": 0.6559756033664168, "learning_rate": 2.5253707842102064e-06, "loss": 1.6072, "step": 10976 }, { "epoch": 0.8401193938466248, "grad_norm": 0.850161191488318, "learning_rate": 2.524879684629685e-06, "loss": 1.6065, "step": 10977 }, { "epoch": 0.8401959283636921, "grad_norm": 0.6779106365038736, "learning_rate": 2.524388798570943e-06, "loss": 1.592, "step": 10978 }, { "epoch": 0.8402724628807592, "grad_norm": 0.5552225469019633, "learning_rate": 2.5238981260641435e-06, "loss": 1.3831, "step": 10979 }, { "epoch": 0.8403489973978264, "grad_norm": 0.5169432273494732, "learning_rate": 2.523407667139436e-06, "loss": 1.1267, "step": 10980 }, { "epoch": 0.8404255319148937, "grad_norm": 0.5666087756288104, "learning_rate": 2.5229174218269536e-06, "loss": 1.2264, "step": 10981 }, { "epoch": 0.8405020664319608, "grad_norm": 0.876617774363156, "learning_rate": 2.5224273901568186e-06, "loss": 1.4979, "step": 10982 }, { "epoch": 0.840578600949028, "grad_norm": 0.5620392285403344, "learning_rate": 2.521937572159141e-06, "loss": 1.3232, "step": 10983 }, { "epoch": 0.8406551354660952, "grad_norm": 0.8542923358443395, "learning_rate": 2.521447967864017e-06, "loss": 1.5925, "step": 10984 }, { "epoch": 0.8407316699831624, "grad_norm": 0.8288436184634072, "learning_rate": 2.520958577301529e-06, "loss": 1.5934, "step": 10985 }, { "epoch": 0.8408082045002296, "grad_norm": 0.5558456703435415, "learning_rate": 2.5204694005017457e-06, "loss": 1.2083, "step": 10986 }, { "epoch": 0.8408847390172968, "grad_norm": 0.7345370204225964, "learning_rate": 2.5199804374947245e-06, "loss": 1.4395, "step": 10987 }, { "epoch": 0.840961273534364, "grad_norm": 0.6978642331186351, "learning_rate": 2.519491688310508e-06, "loss": 1.6378, "step": 10988 }, { "epoch": 0.8410378080514312, "grad_norm": 0.7415195644595659, "learning_rate": 2.519003152979127e-06, "loss": 1.1548, "step": 10989 }, { "epoch": 0.8411143425684984, "grad_norm": 0.6598006060500798, "learning_rate": 2.518514831530599e-06, "loss": 1.6853, "step": 10990 }, { "epoch": 0.8411908770855656, "grad_norm": 0.6828778189099847, "learning_rate": 2.5180267239949274e-06, "loss": 1.6749, "step": 10991 }, { "epoch": 0.8412674116026327, "grad_norm": 0.5994221122680671, "learning_rate": 2.517538830402103e-06, "loss": 1.2464, "step": 10992 }, { "epoch": 0.8413439461197, "grad_norm": 0.4744021467243158, "learning_rate": 2.517051150782102e-06, "loss": 1.0353, "step": 10993 }, { "epoch": 0.8414204806367672, "grad_norm": 0.5733685462652591, "learning_rate": 2.5165636851648924e-06, "loss": 1.2755, "step": 10994 }, { "epoch": 0.8414970151538343, "grad_norm": 0.635605224590152, "learning_rate": 2.516076433580422e-06, "loss": 1.544, "step": 10995 }, { "epoch": 0.8415735496709016, "grad_norm": 0.5177132330685589, "learning_rate": 2.5155893960586314e-06, "loss": 0.9763, "step": 10996 }, { "epoch": 0.8416500841879688, "grad_norm": 0.7137216457085338, "learning_rate": 2.515102572629444e-06, "loss": 1.5699, "step": 10997 }, { "epoch": 0.841726618705036, "grad_norm": 0.5894154947981083, "learning_rate": 2.514615963322773e-06, "loss": 1.4336, "step": 10998 }, { "epoch": 0.8418031532221032, "grad_norm": 0.550956492366986, "learning_rate": 2.5141295681685162e-06, "loss": 1.3067, "step": 10999 }, { "epoch": 0.8418796877391703, "grad_norm": 0.6674858054889408, "learning_rate": 2.51364338719656e-06, "loss": 1.2031, "step": 11000 }, { "epoch": 0.8419562222562376, "grad_norm": 0.5306213472111531, "learning_rate": 2.513157420436776e-06, "loss": 1.184, "step": 11001 }, { "epoch": 0.8420327567733048, "grad_norm": 0.5361943723479464, "learning_rate": 2.5126716679190235e-06, "loss": 1.2275, "step": 11002 }, { "epoch": 0.8421092912903719, "grad_norm": 0.7570185508057556, "learning_rate": 2.5121861296731488e-06, "loss": 1.8229, "step": 11003 }, { "epoch": 0.8421858258074392, "grad_norm": 0.5520940246130379, "learning_rate": 2.511700805728985e-06, "loss": 1.3658, "step": 11004 }, { "epoch": 0.8422623603245063, "grad_norm": 0.6076048561652277, "learning_rate": 2.5112156961163514e-06, "loss": 1.3127, "step": 11005 }, { "epoch": 0.8423388948415735, "grad_norm": 0.5521776895249416, "learning_rate": 2.5107308008650556e-06, "loss": 1.3637, "step": 11006 }, { "epoch": 0.8424154293586408, "grad_norm": 0.7933313586319172, "learning_rate": 2.51024612000489e-06, "loss": 1.9742, "step": 11007 }, { "epoch": 0.8424919638757079, "grad_norm": 0.5863438622461106, "learning_rate": 2.5097616535656357e-06, "loss": 1.3595, "step": 11008 }, { "epoch": 0.8425684983927751, "grad_norm": 0.6074531667261702, "learning_rate": 2.509277401577059e-06, "loss": 1.4092, "step": 11009 }, { "epoch": 0.8426450329098424, "grad_norm": 0.5861233559385873, "learning_rate": 2.5087933640689134e-06, "loss": 1.2344, "step": 11010 }, { "epoch": 0.8427215674269095, "grad_norm": 0.5809889234024911, "learning_rate": 2.508309541070941e-06, "loss": 1.2594, "step": 11011 }, { "epoch": 0.8427981019439768, "grad_norm": 0.5607975041420088, "learning_rate": 2.5078259326128686e-06, "loss": 1.2645, "step": 11012 }, { "epoch": 0.8428746364610439, "grad_norm": 0.5983597656389723, "learning_rate": 2.5073425387244097e-06, "loss": 1.441, "step": 11013 }, { "epoch": 0.8429511709781111, "grad_norm": 0.5678390632489206, "learning_rate": 2.506859359435267e-06, "loss": 1.2837, "step": 11014 }, { "epoch": 0.8430277054951784, "grad_norm": 0.5419445847094189, "learning_rate": 2.506376394775126e-06, "loss": 1.185, "step": 11015 }, { "epoch": 0.8431042400122455, "grad_norm": 0.7840544441879909, "learning_rate": 2.505893644773665e-06, "loss": 1.4319, "step": 11016 }, { "epoch": 0.8431807745293127, "grad_norm": 0.5002966612544815, "learning_rate": 2.505411109460544e-06, "loss": 1.1397, "step": 11017 }, { "epoch": 0.84325730904638, "grad_norm": 0.5509215001442611, "learning_rate": 2.5049287888654108e-06, "loss": 1.2153, "step": 11018 }, { "epoch": 0.8433338435634471, "grad_norm": 0.6948066630494976, "learning_rate": 2.504446683017901e-06, "loss": 1.6035, "step": 11019 }, { "epoch": 0.8434103780805143, "grad_norm": 1.062625187227026, "learning_rate": 2.5039647919476363e-06, "loss": 1.4291, "step": 11020 }, { "epoch": 0.8434869125975815, "grad_norm": 0.7022802026843729, "learning_rate": 2.5034831156842265e-06, "loss": 1.6767, "step": 11021 }, { "epoch": 0.8435634471146487, "grad_norm": 0.7712284555387553, "learning_rate": 2.5030016542572666e-06, "loss": 1.4302, "step": 11022 }, { "epoch": 0.8436399816317159, "grad_norm": 0.6013198975267735, "learning_rate": 2.502520407696339e-06, "loss": 1.4464, "step": 11023 }, { "epoch": 0.8437165161487831, "grad_norm": 0.5594211999856038, "learning_rate": 2.5020393760310137e-06, "loss": 1.3846, "step": 11024 }, { "epoch": 0.8437930506658503, "grad_norm": 0.6223380377292091, "learning_rate": 2.5015585592908453e-06, "loss": 1.6018, "step": 11025 }, { "epoch": 0.8438695851829175, "grad_norm": 0.6026573846049003, "learning_rate": 2.5010779575053768e-06, "loss": 1.2171, "step": 11026 }, { "epoch": 0.8439461196999847, "grad_norm": 0.7462401403744074, "learning_rate": 2.50059757070414e-06, "loss": 2.0092, "step": 11027 }, { "epoch": 0.8440226542170519, "grad_norm": 0.5862936926705887, "learning_rate": 2.5001173989166483e-06, "loss": 1.3917, "step": 11028 }, { "epoch": 0.844099188734119, "grad_norm": 0.632252960876726, "learning_rate": 2.499637442172407e-06, "loss": 1.3896, "step": 11029 }, { "epoch": 0.8441757232511863, "grad_norm": 0.6696748960217095, "learning_rate": 2.4991577005009052e-06, "loss": 1.4069, "step": 11030 }, { "epoch": 0.8442522577682535, "grad_norm": 0.6203868968714992, "learning_rate": 2.498678173931619e-06, "loss": 1.7131, "step": 11031 }, { "epoch": 0.8443287922853207, "grad_norm": 0.6540603861410142, "learning_rate": 2.4981988624940133e-06, "loss": 1.0804, "step": 11032 }, { "epoch": 0.8444053268023879, "grad_norm": 0.6334141776073638, "learning_rate": 2.4977197662175385e-06, "loss": 1.4966, "step": 11033 }, { "epoch": 0.8444818613194551, "grad_norm": 1.1160777969162594, "learning_rate": 2.4972408851316314e-06, "loss": 1.7542, "step": 11034 }, { "epoch": 0.8445583958365223, "grad_norm": 0.56792378581211, "learning_rate": 2.496762219265715e-06, "loss": 1.3674, "step": 11035 }, { "epoch": 0.8446349303535895, "grad_norm": 0.6471416196190866, "learning_rate": 2.4962837686491996e-06, "loss": 1.5846, "step": 11036 }, { "epoch": 0.8447114648706566, "grad_norm": 0.5957215033529285, "learning_rate": 2.495805533311484e-06, "loss": 1.3976, "step": 11037 }, { "epoch": 0.8447879993877239, "grad_norm": 1.1207898906111964, "learning_rate": 2.4953275132819525e-06, "loss": 1.8943, "step": 11038 }, { "epoch": 0.8448645339047911, "grad_norm": 0.6063202612441981, "learning_rate": 2.4948497085899755e-06, "loss": 1.4297, "step": 11039 }, { "epoch": 0.8449410684218582, "grad_norm": 0.6175201458588704, "learning_rate": 2.4943721192649112e-06, "loss": 1.6579, "step": 11040 }, { "epoch": 0.8450176029389255, "grad_norm": 0.5965481442964263, "learning_rate": 2.493894745336103e-06, "loss": 1.3396, "step": 11041 }, { "epoch": 0.8450941374559926, "grad_norm": 0.5428984577213363, "learning_rate": 2.4934175868328826e-06, "loss": 1.3059, "step": 11042 }, { "epoch": 0.8451706719730598, "grad_norm": 0.6078301325963088, "learning_rate": 2.492940643784569e-06, "loss": 1.4034, "step": 11043 }, { "epoch": 0.8452472064901271, "grad_norm": 0.5702751519251845, "learning_rate": 2.492463916220466e-06, "loss": 1.326, "step": 11044 }, { "epoch": 0.8453237410071942, "grad_norm": 0.6172815227204217, "learning_rate": 2.491987404169866e-06, "loss": 1.5666, "step": 11045 }, { "epoch": 0.8454002755242614, "grad_norm": 0.5834209707733081, "learning_rate": 2.4915111076620464e-06, "loss": 1.295, "step": 11046 }, { "epoch": 0.8454768100413287, "grad_norm": 0.6510744895341448, "learning_rate": 2.4910350267262725e-06, "loss": 1.6065, "step": 11047 }, { "epoch": 0.8455533445583958, "grad_norm": 0.6788087183442486, "learning_rate": 2.4905591613917967e-06, "loss": 1.3385, "step": 11048 }, { "epoch": 0.845629879075463, "grad_norm": 0.6512265417086966, "learning_rate": 2.4900835116878573e-06, "loss": 1.3229, "step": 11049 }, { "epoch": 0.8457064135925302, "grad_norm": 0.6922703710863242, "learning_rate": 2.48960807764368e-06, "loss": 1.6347, "step": 11050 }, { "epoch": 0.8457829481095974, "grad_norm": 0.7985141363070086, "learning_rate": 2.489132859288476e-06, "loss": 1.5568, "step": 11051 }, { "epoch": 0.8458594826266647, "grad_norm": 1.107531749098794, "learning_rate": 2.488657856651445e-06, "loss": 1.6004, "step": 11052 }, { "epoch": 0.8459360171437318, "grad_norm": 0.5641127523708964, "learning_rate": 2.4881830697617715e-06, "loss": 1.2073, "step": 11053 }, { "epoch": 0.846012551660799, "grad_norm": 0.8432190578451191, "learning_rate": 2.4877084986486286e-06, "loss": 1.7655, "step": 11054 }, { "epoch": 0.8460890861778663, "grad_norm": 0.5305104219969047, "learning_rate": 2.487234143341176e-06, "loss": 1.1287, "step": 11055 }, { "epoch": 0.8461656206949334, "grad_norm": 0.818529601990499, "learning_rate": 2.486760003868558e-06, "loss": 1.2595, "step": 11056 }, { "epoch": 0.8462421552120006, "grad_norm": 0.5741188822077337, "learning_rate": 2.4862860802599092e-06, "loss": 1.3588, "step": 11057 }, { "epoch": 0.8463186897290678, "grad_norm": 0.5582547269110774, "learning_rate": 2.4858123725443465e-06, "loss": 1.3089, "step": 11058 }, { "epoch": 0.846395224246135, "grad_norm": 0.5429541640404636, "learning_rate": 2.4853388807509775e-06, "loss": 1.0619, "step": 11059 }, { "epoch": 0.8464717587632022, "grad_norm": 0.587258993524751, "learning_rate": 2.484865604908895e-06, "loss": 1.3613, "step": 11060 }, { "epoch": 0.8465482932802694, "grad_norm": 0.6034553280572867, "learning_rate": 2.4843925450471775e-06, "loss": 1.4656, "step": 11061 }, { "epoch": 0.8466248277973366, "grad_norm": 0.5787652273361946, "learning_rate": 2.483919701194891e-06, "loss": 1.2633, "step": 11062 }, { "epoch": 0.8467013623144038, "grad_norm": 0.6530312387274708, "learning_rate": 2.48344707338109e-06, "loss": 1.5677, "step": 11063 }, { "epoch": 0.846777896831471, "grad_norm": 0.5582988475344193, "learning_rate": 2.4829746616348123e-06, "loss": 1.2163, "step": 11064 }, { "epoch": 0.8468544313485382, "grad_norm": 0.6115370653230284, "learning_rate": 2.4825024659850865e-06, "loss": 1.5603, "step": 11065 }, { "epoch": 0.8469309658656053, "grad_norm": 0.6315995701243262, "learning_rate": 2.482030486460924e-06, "loss": 1.5362, "step": 11066 }, { "epoch": 0.8470075003826726, "grad_norm": 0.5416592442751135, "learning_rate": 2.4815587230913254e-06, "loss": 1.2302, "step": 11067 }, { "epoch": 0.8470840348997398, "grad_norm": 0.5303774648368983, "learning_rate": 2.4810871759052766e-06, "loss": 1.1059, "step": 11068 }, { "epoch": 0.847160569416807, "grad_norm": 0.6400894376896151, "learning_rate": 2.480615844931751e-06, "loss": 1.2842, "step": 11069 }, { "epoch": 0.8472371039338742, "grad_norm": 0.6146335270102256, "learning_rate": 2.4801447301997096e-06, "loss": 1.5458, "step": 11070 }, { "epoch": 0.8473136384509413, "grad_norm": 0.7592355916062772, "learning_rate": 2.4796738317380975e-06, "loss": 1.3441, "step": 11071 }, { "epoch": 0.8473901729680086, "grad_norm": 0.5503628290176042, "learning_rate": 2.4792031495758496e-06, "loss": 1.3164, "step": 11072 }, { "epoch": 0.8474667074850758, "grad_norm": 0.6047565939040012, "learning_rate": 2.4787326837418854e-06, "loss": 1.3637, "step": 11073 }, { "epoch": 0.8475432420021429, "grad_norm": 0.6798486869517205, "learning_rate": 2.478262434265112e-06, "loss": 1.5402, "step": 11074 }, { "epoch": 0.8476197765192102, "grad_norm": 0.9465488847571444, "learning_rate": 2.4777924011744215e-06, "loss": 1.1893, "step": 11075 }, { "epoch": 0.8476963110362774, "grad_norm": 0.5981334092561307, "learning_rate": 2.477322584498696e-06, "loss": 1.1202, "step": 11076 }, { "epoch": 0.8477728455533445, "grad_norm": 0.6049336881681008, "learning_rate": 2.4768529842668015e-06, "loss": 1.4839, "step": 11077 }, { "epoch": 0.8478493800704118, "grad_norm": 0.7697195606429555, "learning_rate": 2.476383600507592e-06, "loss": 1.8361, "step": 11078 }, { "epoch": 0.8479259145874789, "grad_norm": 0.7065906330122369, "learning_rate": 2.4759144332499074e-06, "loss": 1.1347, "step": 11079 }, { "epoch": 0.8480024491045461, "grad_norm": 0.6469636114500157, "learning_rate": 2.4754454825225758e-06, "loss": 1.5786, "step": 11080 }, { "epoch": 0.8480789836216134, "grad_norm": 0.8458629520652186, "learning_rate": 2.4749767483544096e-06, "loss": 1.553, "step": 11081 }, { "epoch": 0.8481555181386805, "grad_norm": 0.5899290831428753, "learning_rate": 2.4745082307742103e-06, "loss": 1.379, "step": 11082 }, { "epoch": 0.8482320526557477, "grad_norm": 0.6383557978452313, "learning_rate": 2.474039929810764e-06, "loss": 1.6568, "step": 11083 }, { "epoch": 0.848308587172815, "grad_norm": 0.5235221955959697, "learning_rate": 2.4735718454928454e-06, "loss": 1.3128, "step": 11084 }, { "epoch": 0.8483851216898821, "grad_norm": 0.6001975261849907, "learning_rate": 2.473103977849214e-06, "loss": 1.4593, "step": 11085 }, { "epoch": 0.8484616562069494, "grad_norm": 0.5650729579260065, "learning_rate": 2.4726363269086174e-06, "loss": 1.3569, "step": 11086 }, { "epoch": 0.8485381907240165, "grad_norm": 0.6068603024841491, "learning_rate": 2.47216889269979e-06, "loss": 1.5041, "step": 11087 }, { "epoch": 0.8486147252410837, "grad_norm": 0.5720835183746904, "learning_rate": 2.4717016752514526e-06, "loss": 1.37, "step": 11088 }, { "epoch": 0.848691259758151, "grad_norm": 0.609839871631001, "learning_rate": 2.471234674592312e-06, "loss": 1.5092, "step": 11089 }, { "epoch": 0.8487677942752181, "grad_norm": 0.6172135050659989, "learning_rate": 2.4707678907510615e-06, "loss": 1.5121, "step": 11090 }, { "epoch": 0.8488443287922853, "grad_norm": 0.6089138812285247, "learning_rate": 2.4703013237563815e-06, "loss": 1.47, "step": 11091 }, { "epoch": 0.8489208633093526, "grad_norm": 0.6326606861667143, "learning_rate": 2.4698349736369406e-06, "loss": 1.5834, "step": 11092 }, { "epoch": 0.8489973978264197, "grad_norm": 0.5312299226906908, "learning_rate": 2.469368840421392e-06, "loss": 1.1789, "step": 11093 }, { "epoch": 0.8490739323434869, "grad_norm": 0.6008474214504805, "learning_rate": 2.468902924138377e-06, "loss": 1.4646, "step": 11094 }, { "epoch": 0.8491504668605541, "grad_norm": 0.8141052453012527, "learning_rate": 2.4684372248165223e-06, "loss": 2.0972, "step": 11095 }, { "epoch": 0.8492270013776213, "grad_norm": 0.5034903956941499, "learning_rate": 2.4679717424844416e-06, "loss": 0.9559, "step": 11096 }, { "epoch": 0.8493035358946885, "grad_norm": 0.6244175157063173, "learning_rate": 2.4675064771707345e-06, "loss": 1.415, "step": 11097 }, { "epoch": 0.8493800704117557, "grad_norm": 0.5626208774611499, "learning_rate": 2.467041428903991e-06, "loss": 1.1821, "step": 11098 }, { "epoch": 0.8494566049288229, "grad_norm": 0.6002466217970304, "learning_rate": 2.4665765977127837e-06, "loss": 1.1626, "step": 11099 }, { "epoch": 0.8495331394458902, "grad_norm": 0.5959171206937951, "learning_rate": 2.4661119836256727e-06, "loss": 1.3952, "step": 11100 }, { "epoch": 0.8496096739629573, "grad_norm": 0.6567593838094257, "learning_rate": 2.465647586671206e-06, "loss": 1.7526, "step": 11101 }, { "epoch": 0.8496862084800245, "grad_norm": 0.8922239834512372, "learning_rate": 2.4651834068779173e-06, "loss": 1.4062, "step": 11102 }, { "epoch": 0.8497627429970916, "grad_norm": 0.6369210282553781, "learning_rate": 2.464719444274327e-06, "loss": 1.6534, "step": 11103 }, { "epoch": 0.8498392775141589, "grad_norm": 0.52104751877872, "learning_rate": 2.4642556988889427e-06, "loss": 1.0201, "step": 11104 }, { "epoch": 0.8499158120312261, "grad_norm": 0.5602929304968869, "learning_rate": 2.4637921707502584e-06, "loss": 1.1494, "step": 11105 }, { "epoch": 0.8499923465482933, "grad_norm": 0.6032415420717382, "learning_rate": 2.463328859886754e-06, "loss": 1.394, "step": 11106 }, { "epoch": 0.8500688810653605, "grad_norm": 0.6171968924960941, "learning_rate": 2.4628657663268974e-06, "loss": 1.5916, "step": 11107 }, { "epoch": 0.8501454155824276, "grad_norm": 0.8920540936417122, "learning_rate": 2.462402890099141e-06, "loss": 1.6893, "step": 11108 }, { "epoch": 0.8502219500994949, "grad_norm": 0.5914659285714382, "learning_rate": 2.4619402312319274e-06, "loss": 1.4001, "step": 11109 }, { "epoch": 0.8502984846165621, "grad_norm": 0.6742978017596545, "learning_rate": 2.461477789753682e-06, "loss": 1.2159, "step": 11110 }, { "epoch": 0.8503750191336292, "grad_norm": 0.6615570380358304, "learning_rate": 2.4610155656928207e-06, "loss": 1.419, "step": 11111 }, { "epoch": 0.8504515536506965, "grad_norm": 0.5263272058451188, "learning_rate": 2.460553559077742e-06, "loss": 1.0156, "step": 11112 }, { "epoch": 0.8505280881677637, "grad_norm": 0.5797551830124849, "learning_rate": 2.460091769936833e-06, "loss": 1.3913, "step": 11113 }, { "epoch": 0.8506046226848308, "grad_norm": 0.5239522336186484, "learning_rate": 2.4596301982984684e-06, "loss": 1.1144, "step": 11114 }, { "epoch": 0.8506811572018981, "grad_norm": 0.6877905785315377, "learning_rate": 2.4591688441910078e-06, "loss": 1.5598, "step": 11115 }, { "epoch": 0.8507576917189652, "grad_norm": 0.6344579398705135, "learning_rate": 2.458707707642799e-06, "loss": 1.4137, "step": 11116 }, { "epoch": 0.8508342262360324, "grad_norm": 0.5127568816665304, "learning_rate": 2.4582467886821753e-06, "loss": 1.0217, "step": 11117 }, { "epoch": 0.8509107607530997, "grad_norm": 0.7023796190961731, "learning_rate": 2.4577860873374567e-06, "loss": 1.5407, "step": 11118 }, { "epoch": 0.8509872952701668, "grad_norm": 0.7340470140116464, "learning_rate": 2.4573256036369484e-06, "loss": 1.6805, "step": 11119 }, { "epoch": 0.851063829787234, "grad_norm": 0.5333191969628576, "learning_rate": 2.4568653376089473e-06, "loss": 1.0682, "step": 11120 }, { "epoch": 0.8511403643043013, "grad_norm": 0.5191080050974436, "learning_rate": 2.4564052892817315e-06, "loss": 1.1363, "step": 11121 }, { "epoch": 0.8512168988213684, "grad_norm": 0.6465206558652768, "learning_rate": 2.455945458683568e-06, "loss": 1.4134, "step": 11122 }, { "epoch": 0.8512934333384357, "grad_norm": 0.6845075475136547, "learning_rate": 2.45548584584271e-06, "loss": 1.404, "step": 11123 }, { "epoch": 0.8513699678555028, "grad_norm": 0.44763066965669657, "learning_rate": 2.4550264507873982e-06, "loss": 0.8198, "step": 11124 }, { "epoch": 0.85144650237257, "grad_norm": 0.6785244219378384, "learning_rate": 2.4545672735458585e-06, "loss": 1.6835, "step": 11125 }, { "epoch": 0.8515230368896373, "grad_norm": 0.7475079141788933, "learning_rate": 2.4541083141463044e-06, "loss": 1.1913, "step": 11126 }, { "epoch": 0.8515995714067044, "grad_norm": 0.6321605316664762, "learning_rate": 2.4536495726169355e-06, "loss": 1.4079, "step": 11127 }, { "epoch": 0.8516761059237716, "grad_norm": 0.774364716973228, "learning_rate": 2.453191048985939e-06, "loss": 1.6391, "step": 11128 }, { "epoch": 0.8517526404408389, "grad_norm": 0.5297765094358714, "learning_rate": 2.452732743281487e-06, "loss": 1.1148, "step": 11129 }, { "epoch": 0.851829174957906, "grad_norm": 0.6105824333980263, "learning_rate": 2.4522746555317384e-06, "loss": 1.4888, "step": 11130 }, { "epoch": 0.8519057094749732, "grad_norm": 0.5154221529576004, "learning_rate": 2.4518167857648417e-06, "loss": 1.118, "step": 11131 }, { "epoch": 0.8519822439920404, "grad_norm": 0.5648471677550463, "learning_rate": 2.451359134008929e-06, "loss": 1.2995, "step": 11132 }, { "epoch": 0.8520587785091076, "grad_norm": 0.6454337738679529, "learning_rate": 2.4509017002921196e-06, "loss": 1.6325, "step": 11133 }, { "epoch": 0.8521353130261748, "grad_norm": 0.5145186669490518, "learning_rate": 2.4504444846425193e-06, "loss": 1.1571, "step": 11134 }, { "epoch": 0.852211847543242, "grad_norm": 0.5453734983152653, "learning_rate": 2.4499874870882196e-06, "loss": 1.1925, "step": 11135 }, { "epoch": 0.8522883820603092, "grad_norm": 0.6895120590313303, "learning_rate": 2.4495307076573028e-06, "loss": 1.275, "step": 11136 }, { "epoch": 0.8523649165773763, "grad_norm": 0.5864966239830961, "learning_rate": 2.449074146377833e-06, "loss": 1.3548, "step": 11137 }, { "epoch": 0.8524414510944436, "grad_norm": 0.6617446097869272, "learning_rate": 2.448617803277863e-06, "loss": 1.5541, "step": 11138 }, { "epoch": 0.8525179856115108, "grad_norm": 0.7605037064739192, "learning_rate": 2.4481616783854316e-06, "loss": 1.4015, "step": 11139 }, { "epoch": 0.852594520128578, "grad_norm": 0.5881793871916914, "learning_rate": 2.4477057717285645e-06, "loss": 1.0566, "step": 11140 }, { "epoch": 0.8526710546456452, "grad_norm": 0.49592700517920024, "learning_rate": 2.447250083335274e-06, "loss": 1.121, "step": 11141 }, { "epoch": 0.8527475891627124, "grad_norm": 0.49765157766976303, "learning_rate": 2.4467946132335594e-06, "loss": 1.162, "step": 11142 }, { "epoch": 0.8528241236797796, "grad_norm": 0.6149118314063103, "learning_rate": 2.446339361451405e-06, "loss": 1.3791, "step": 11143 }, { "epoch": 0.8529006581968468, "grad_norm": 0.5853070537816417, "learning_rate": 2.445884328016785e-06, "loss": 1.1429, "step": 11144 }, { "epoch": 0.8529771927139139, "grad_norm": 0.839383279419676, "learning_rate": 2.4454295129576565e-06, "loss": 1.4947, "step": 11145 }, { "epoch": 0.8530537272309812, "grad_norm": 0.480645290209718, "learning_rate": 2.4449749163019642e-06, "loss": 1.0435, "step": 11146 }, { "epoch": 0.8531302617480484, "grad_norm": 0.5033457219800452, "learning_rate": 2.444520538077641e-06, "loss": 1.0915, "step": 11147 }, { "epoch": 0.8532067962651155, "grad_norm": 1.2426798253312117, "learning_rate": 2.444066378312605e-06, "loss": 1.5107, "step": 11148 }, { "epoch": 0.8532833307821828, "grad_norm": 0.624706653637152, "learning_rate": 2.4436124370347604e-06, "loss": 1.5701, "step": 11149 }, { "epoch": 0.85335986529925, "grad_norm": 0.9947626187580312, "learning_rate": 2.4431587142719993e-06, "loss": 1.5234, "step": 11150 }, { "epoch": 0.8534363998163171, "grad_norm": 0.8553232801519037, "learning_rate": 2.4427052100521997e-06, "loss": 1.7081, "step": 11151 }, { "epoch": 0.8535129343333844, "grad_norm": 0.6401505290476756, "learning_rate": 2.4422519244032258e-06, "loss": 1.6716, "step": 11152 }, { "epoch": 0.8535894688504515, "grad_norm": 0.7036283958079348, "learning_rate": 2.44179885735293e-06, "loss": 1.2525, "step": 11153 }, { "epoch": 0.8536660033675187, "grad_norm": 0.5234505100700797, "learning_rate": 2.4413460089291492e-06, "loss": 1.2176, "step": 11154 }, { "epoch": 0.853742537884586, "grad_norm": 0.5622551106684437, "learning_rate": 2.440893379159708e-06, "loss": 1.3222, "step": 11155 }, { "epoch": 0.8538190724016531, "grad_norm": 0.5628167944981787, "learning_rate": 2.4404409680724174e-06, "loss": 1.4461, "step": 11156 }, { "epoch": 0.8538956069187204, "grad_norm": 0.5582897553620868, "learning_rate": 2.439988775695074e-06, "loss": 1.1497, "step": 11157 }, { "epoch": 0.8539721414357876, "grad_norm": 0.5909935983827657, "learning_rate": 2.4395368020554633e-06, "loss": 1.243, "step": 11158 }, { "epoch": 0.8540486759528547, "grad_norm": 0.6375549590857812, "learning_rate": 2.439085047181356e-06, "loss": 1.4634, "step": 11159 }, { "epoch": 0.854125210469922, "grad_norm": 0.5200367851075177, "learning_rate": 2.4386335111005076e-06, "loss": 1.128, "step": 11160 }, { "epoch": 0.8542017449869891, "grad_norm": 0.5467560353828406, "learning_rate": 2.438182193840663e-06, "loss": 1.1318, "step": 11161 }, { "epoch": 0.8542782795040563, "grad_norm": 0.6025405462333524, "learning_rate": 2.4377310954295522e-06, "loss": 1.4194, "step": 11162 }, { "epoch": 0.8543548140211236, "grad_norm": 0.6160321543358976, "learning_rate": 2.4372802158948914e-06, "loss": 1.3697, "step": 11163 }, { "epoch": 0.8544313485381907, "grad_norm": 0.5630371389545457, "learning_rate": 2.436829555264385e-06, "loss": 1.4015, "step": 11164 }, { "epoch": 0.8545078830552579, "grad_norm": 0.6817020375855806, "learning_rate": 2.436379113565723e-06, "loss": 1.6119, "step": 11165 }, { "epoch": 0.8545844175723252, "grad_norm": 0.5500660981292858, "learning_rate": 2.4359288908265816e-06, "loss": 1.092, "step": 11166 }, { "epoch": 0.8546609520893923, "grad_norm": 0.6734954241836193, "learning_rate": 2.4354788870746236e-06, "loss": 1.5042, "step": 11167 }, { "epoch": 0.8547374866064595, "grad_norm": 0.7584493424950023, "learning_rate": 2.4350291023374985e-06, "loss": 1.3268, "step": 11168 }, { "epoch": 0.8548140211235267, "grad_norm": 0.5392494251850285, "learning_rate": 2.4345795366428428e-06, "loss": 1.2151, "step": 11169 }, { "epoch": 0.8548905556405939, "grad_norm": 0.5224439486512648, "learning_rate": 2.4341301900182783e-06, "loss": 1.1638, "step": 11170 }, { "epoch": 0.8549670901576611, "grad_norm": 0.5517404461300145, "learning_rate": 2.4336810624914153e-06, "loss": 1.1295, "step": 11171 }, { "epoch": 0.8550436246747283, "grad_norm": 0.5274706223109279, "learning_rate": 2.4332321540898492e-06, "loss": 1.1302, "step": 11172 }, { "epoch": 0.8551201591917955, "grad_norm": 0.7206848408472905, "learning_rate": 2.4327834648411617e-06, "loss": 1.2613, "step": 11173 }, { "epoch": 0.8551966937088626, "grad_norm": 0.7132465577296153, "learning_rate": 2.432334994772922e-06, "loss": 1.4283, "step": 11174 }, { "epoch": 0.8552732282259299, "grad_norm": 0.638824548037196, "learning_rate": 2.431886743912686e-06, "loss": 1.3086, "step": 11175 }, { "epoch": 0.8553497627429971, "grad_norm": 0.5069963574469795, "learning_rate": 2.431438712287995e-06, "loss": 1.0141, "step": 11176 }, { "epoch": 0.8554262972600643, "grad_norm": 0.6517711915930038, "learning_rate": 2.430990899926377e-06, "loss": 1.3549, "step": 11177 }, { "epoch": 0.8555028317771315, "grad_norm": 0.5060012903898912, "learning_rate": 2.430543306855348e-06, "loss": 1.1041, "step": 11178 }, { "epoch": 0.8555793662941987, "grad_norm": 0.6253615779921146, "learning_rate": 2.4300959331024078e-06, "loss": 1.5105, "step": 11179 }, { "epoch": 0.8556559008112659, "grad_norm": 0.5006101735520138, "learning_rate": 2.4296487786950466e-06, "loss": 1.2383, "step": 11180 }, { "epoch": 0.8557324353283331, "grad_norm": 0.7261806031613293, "learning_rate": 2.4292018436607378e-06, "loss": 1.6304, "step": 11181 }, { "epoch": 0.8558089698454002, "grad_norm": 0.5353063323674004, "learning_rate": 2.428755128026942e-06, "loss": 1.1323, "step": 11182 }, { "epoch": 0.8558855043624675, "grad_norm": 1.186281121970952, "learning_rate": 2.4283086318211076e-06, "loss": 1.7498, "step": 11183 }, { "epoch": 0.8559620388795347, "grad_norm": 0.6608071116461002, "learning_rate": 2.427862355070667e-06, "loss": 1.4577, "step": 11184 }, { "epoch": 0.8560385733966018, "grad_norm": 0.5341687602106407, "learning_rate": 2.4274162978030427e-06, "loss": 1.1193, "step": 11185 }, { "epoch": 0.8561151079136691, "grad_norm": 0.5177264119634429, "learning_rate": 2.4269704600456413e-06, "loss": 1.1627, "step": 11186 }, { "epoch": 0.8561916424307363, "grad_norm": 0.6792424041050977, "learning_rate": 2.4265248418258556e-06, "loss": 1.7441, "step": 11187 }, { "epoch": 0.8562681769478034, "grad_norm": 0.7396963597140385, "learning_rate": 2.4260794431710672e-06, "loss": 1.5762, "step": 11188 }, { "epoch": 0.8563447114648707, "grad_norm": 0.5914687750038589, "learning_rate": 2.4256342641086408e-06, "loss": 1.4395, "step": 11189 }, { "epoch": 0.8564212459819378, "grad_norm": 0.9025732241420436, "learning_rate": 2.425189304665931e-06, "loss": 1.4647, "step": 11190 }, { "epoch": 0.856497780499005, "grad_norm": 0.7407139137876086, "learning_rate": 2.4247445648702774e-06, "loss": 1.6265, "step": 11191 }, { "epoch": 0.8565743150160723, "grad_norm": 0.6832542290177788, "learning_rate": 2.424300044749005e-06, "loss": 1.5424, "step": 11192 }, { "epoch": 0.8566508495331394, "grad_norm": 0.5458224290055045, "learning_rate": 2.4238557443294274e-06, "loss": 1.1456, "step": 11193 }, { "epoch": 0.8567273840502067, "grad_norm": 0.5870609844171121, "learning_rate": 2.423411663638844e-06, "loss": 1.2412, "step": 11194 }, { "epoch": 0.8568039185672739, "grad_norm": 0.6313073094975591, "learning_rate": 2.422967802704539e-06, "loss": 1.4073, "step": 11195 }, { "epoch": 0.856880453084341, "grad_norm": 0.6991784036571295, "learning_rate": 2.4225241615537855e-06, "loss": 1.5017, "step": 11196 }, { "epoch": 0.8569569876014083, "grad_norm": 1.1482984143088242, "learning_rate": 2.422080740213843e-06, "loss": 1.6209, "step": 11197 }, { "epoch": 0.8570335221184754, "grad_norm": 0.5970409711079316, "learning_rate": 2.4216375387119556e-06, "loss": 1.4041, "step": 11198 }, { "epoch": 0.8571100566355426, "grad_norm": 0.9695620912326636, "learning_rate": 2.421194557075355e-06, "loss": 1.7298, "step": 11199 }, { "epoch": 0.8571865911526099, "grad_norm": 0.5652673343101584, "learning_rate": 2.4207517953312593e-06, "loss": 1.1929, "step": 11200 }, { "epoch": 0.857263125669677, "grad_norm": 0.6235565122278709, "learning_rate": 2.420309253506872e-06, "loss": 1.3627, "step": 11201 }, { "epoch": 0.8573396601867442, "grad_norm": 0.5486860999590467, "learning_rate": 2.419866931629387e-06, "loss": 1.3015, "step": 11202 }, { "epoch": 0.8574161947038115, "grad_norm": 0.5838347524826529, "learning_rate": 2.41942482972598e-06, "loss": 1.5695, "step": 11203 }, { "epoch": 0.8574927292208786, "grad_norm": 0.5929306883336689, "learning_rate": 2.418982947823815e-06, "loss": 1.276, "step": 11204 }, { "epoch": 0.8575692637379458, "grad_norm": 0.5566806269147947, "learning_rate": 2.418541285950043e-06, "loss": 1.276, "step": 11205 }, { "epoch": 0.857645798255013, "grad_norm": 0.605791587680859, "learning_rate": 2.4180998441318005e-06, "loss": 1.3991, "step": 11206 }, { "epoch": 0.8577223327720802, "grad_norm": 0.6653415257604975, "learning_rate": 2.4176586223962117e-06, "loss": 1.4351, "step": 11207 }, { "epoch": 0.8577988672891474, "grad_norm": 0.6170751945703233, "learning_rate": 2.417217620770387e-06, "loss": 1.5692, "step": 11208 }, { "epoch": 0.8578754018062146, "grad_norm": 0.6128641462080381, "learning_rate": 2.416776839281422e-06, "loss": 1.4561, "step": 11209 }, { "epoch": 0.8579519363232818, "grad_norm": 0.4930119999635218, "learning_rate": 2.416336277956399e-06, "loss": 1.0122, "step": 11210 }, { "epoch": 0.858028470840349, "grad_norm": 0.5188356034690047, "learning_rate": 2.415895936822389e-06, "loss": 1.0874, "step": 11211 }, { "epoch": 0.8581050053574162, "grad_norm": 0.4521086707021159, "learning_rate": 2.4154558159064467e-06, "loss": 0.9876, "step": 11212 }, { "epoch": 0.8581815398744834, "grad_norm": 0.603891257991672, "learning_rate": 2.415015915235615e-06, "loss": 1.4066, "step": 11213 }, { "epoch": 0.8582580743915506, "grad_norm": 0.5554391830566326, "learning_rate": 2.4145762348369225e-06, "loss": 1.1715, "step": 11214 }, { "epoch": 0.8583346089086178, "grad_norm": 3.8382185490175735, "learning_rate": 2.4141367747373846e-06, "loss": 1.7046, "step": 11215 }, { "epoch": 0.858411143425685, "grad_norm": 0.5751272423840762, "learning_rate": 2.4136975349640023e-06, "loss": 1.2021, "step": 11216 }, { "epoch": 0.8584876779427522, "grad_norm": 0.6712737731022017, "learning_rate": 2.4132585155437645e-06, "loss": 1.9631, "step": 11217 }, { "epoch": 0.8585642124598194, "grad_norm": 0.5546568701959341, "learning_rate": 2.4128197165036466e-06, "loss": 1.2641, "step": 11218 }, { "epoch": 0.8586407469768865, "grad_norm": 0.6021431333907895, "learning_rate": 2.4123811378706085e-06, "loss": 1.3393, "step": 11219 }, { "epoch": 0.8587172814939538, "grad_norm": 0.6510686497823504, "learning_rate": 2.4119427796715985e-06, "loss": 1.6181, "step": 11220 }, { "epoch": 0.858793816011021, "grad_norm": 0.5795466760629626, "learning_rate": 2.4115046419335496e-06, "loss": 1.1377, "step": 11221 }, { "epoch": 0.8588703505280881, "grad_norm": 0.5632518325324722, "learning_rate": 2.411066724683384e-06, "loss": 1.2903, "step": 11222 }, { "epoch": 0.8589468850451554, "grad_norm": 0.6199875460158168, "learning_rate": 2.4106290279480067e-06, "loss": 1.6359, "step": 11223 }, { "epoch": 0.8590234195622226, "grad_norm": 0.6098176596652777, "learning_rate": 2.4101915517543127e-06, "loss": 1.577, "step": 11224 }, { "epoch": 0.8590999540792897, "grad_norm": 0.4798846260418684, "learning_rate": 2.409754296129182e-06, "loss": 0.8913, "step": 11225 }, { "epoch": 0.859176488596357, "grad_norm": 3.372324292217871, "learning_rate": 2.4093172610994787e-06, "loss": 1.4812, "step": 11226 }, { "epoch": 0.8592530231134241, "grad_norm": 0.5621329273549144, "learning_rate": 2.408880446692058e-06, "loss": 1.2627, "step": 11227 }, { "epoch": 0.8593295576304913, "grad_norm": 0.5346687460617281, "learning_rate": 2.408443852933757e-06, "loss": 1.2395, "step": 11228 }, { "epoch": 0.8594060921475586, "grad_norm": 0.5253426042153463, "learning_rate": 2.408007479851403e-06, "loss": 1.2902, "step": 11229 }, { "epoch": 0.8594826266646257, "grad_norm": 0.5872490093249251, "learning_rate": 2.407571327471808e-06, "loss": 1.2655, "step": 11230 }, { "epoch": 0.859559161181693, "grad_norm": 0.6094069930618536, "learning_rate": 2.4071353958217693e-06, "loss": 1.2831, "step": 11231 }, { "epoch": 0.8596356956987602, "grad_norm": 0.5587350215239849, "learning_rate": 2.406699684928073e-06, "loss": 1.3366, "step": 11232 }, { "epoch": 0.8597122302158273, "grad_norm": 0.6199081201433114, "learning_rate": 2.40626419481749e-06, "loss": 1.3972, "step": 11233 }, { "epoch": 0.8597887647328946, "grad_norm": 0.6547695587729501, "learning_rate": 2.405828925516778e-06, "loss": 1.5838, "step": 11234 }, { "epoch": 0.8598652992499617, "grad_norm": 0.5652925267290111, "learning_rate": 2.405393877052681e-06, "loss": 1.1447, "step": 11235 }, { "epoch": 0.8599418337670289, "grad_norm": 0.597685962275302, "learning_rate": 2.40495904945193e-06, "loss": 1.4494, "step": 11236 }, { "epoch": 0.8600183682840962, "grad_norm": 0.518270503424825, "learning_rate": 2.4045244427412427e-06, "loss": 1.1916, "step": 11237 }, { "epoch": 0.8600949028011633, "grad_norm": 0.5355196237773261, "learning_rate": 2.4040900569473216e-06, "loss": 1.2134, "step": 11238 }, { "epoch": 0.8601714373182305, "grad_norm": 0.7139132544176422, "learning_rate": 2.4036558920968566e-06, "loss": 1.8975, "step": 11239 }, { "epoch": 0.8602479718352977, "grad_norm": 0.5548464349869265, "learning_rate": 2.4032219482165255e-06, "loss": 1.1989, "step": 11240 }, { "epoch": 0.8603245063523649, "grad_norm": 0.62028393773211, "learning_rate": 2.40278822533299e-06, "loss": 1.2646, "step": 11241 }, { "epoch": 0.8604010408694321, "grad_norm": 0.6282496773469671, "learning_rate": 2.4023547234728996e-06, "loss": 1.4065, "step": 11242 }, { "epoch": 0.8604775753864993, "grad_norm": 0.52496634496388, "learning_rate": 2.401921442662889e-06, "loss": 1.1002, "step": 11243 }, { "epoch": 0.8605541099035665, "grad_norm": 0.6160845346047491, "learning_rate": 2.4014883829295824e-06, "loss": 1.5367, "step": 11244 }, { "epoch": 0.8606306444206338, "grad_norm": 0.5379566390353853, "learning_rate": 2.4010555442995857e-06, "loss": 1.2342, "step": 11245 }, { "epoch": 0.8607071789377009, "grad_norm": 0.5053852514490089, "learning_rate": 2.4006229267994965e-06, "loss": 1.2158, "step": 11246 }, { "epoch": 0.8607837134547681, "grad_norm": 0.6295791262326764, "learning_rate": 2.4001905304558944e-06, "loss": 1.2379, "step": 11247 }, { "epoch": 0.8608602479718352, "grad_norm": 0.7402885483454418, "learning_rate": 2.3997583552953474e-06, "loss": 1.6529, "step": 11248 }, { "epoch": 0.8609367824889025, "grad_norm": 0.6234859543872369, "learning_rate": 2.39932640134441e-06, "loss": 1.4266, "step": 11249 }, { "epoch": 0.8610133170059697, "grad_norm": 0.5686790294136161, "learning_rate": 2.3988946686296213e-06, "loss": 1.387, "step": 11250 }, { "epoch": 0.8610898515230369, "grad_norm": 0.6521336998535009, "learning_rate": 2.3984631571775106e-06, "loss": 1.5375, "step": 11251 }, { "epoch": 0.8611663860401041, "grad_norm": 0.5575487998202823, "learning_rate": 2.3980318670145895e-06, "loss": 1.3434, "step": 11252 }, { "epoch": 0.8612429205571713, "grad_norm": 0.7470869758847066, "learning_rate": 2.397600798167359e-06, "loss": 1.6345, "step": 11253 }, { "epoch": 0.8613194550742385, "grad_norm": 0.8499166951363794, "learning_rate": 2.3971699506623045e-06, "loss": 1.5003, "step": 11254 }, { "epoch": 0.8613959895913057, "grad_norm": 0.6280049137424585, "learning_rate": 2.396739324525898e-06, "loss": 1.5408, "step": 11255 }, { "epoch": 0.8614725241083728, "grad_norm": 0.6260147201457019, "learning_rate": 2.3963089197845987e-06, "loss": 1.4647, "step": 11256 }, { "epoch": 0.8615490586254401, "grad_norm": 0.5720529809431748, "learning_rate": 2.3958787364648535e-06, "loss": 1.3379, "step": 11257 }, { "epoch": 0.8616255931425073, "grad_norm": 0.5561530502185588, "learning_rate": 2.3954487745930925e-06, "loss": 1.188, "step": 11258 }, { "epoch": 0.8617021276595744, "grad_norm": 0.6228685541357704, "learning_rate": 2.395019034195735e-06, "loss": 1.3753, "step": 11259 }, { "epoch": 0.8617786621766417, "grad_norm": 0.5789286797203252, "learning_rate": 2.394589515299185e-06, "loss": 1.2949, "step": 11260 }, { "epoch": 0.8618551966937089, "grad_norm": 0.44371806007336234, "learning_rate": 2.3941602179298316e-06, "loss": 0.8691, "step": 11261 }, { "epoch": 0.861931731210776, "grad_norm": 0.7676272200473223, "learning_rate": 2.3937311421140553e-06, "loss": 1.7491, "step": 11262 }, { "epoch": 0.8620082657278433, "grad_norm": 0.684257738865875, "learning_rate": 2.3933022878782177e-06, "loss": 1.677, "step": 11263 }, { "epoch": 0.8620848002449104, "grad_norm": 0.546746764723464, "learning_rate": 2.3928736552486696e-06, "loss": 1.2303, "step": 11264 }, { "epoch": 0.8621613347619776, "grad_norm": 0.5612706078277494, "learning_rate": 2.3924452442517475e-06, "loss": 1.3767, "step": 11265 }, { "epoch": 0.8622378692790449, "grad_norm": 0.9713391497750917, "learning_rate": 2.392017054913774e-06, "loss": 1.1678, "step": 11266 }, { "epoch": 0.862314403796112, "grad_norm": 0.7252138418642895, "learning_rate": 2.391589087261058e-06, "loss": 1.4349, "step": 11267 }, { "epoch": 0.8623909383131793, "grad_norm": 0.5568899626170792, "learning_rate": 2.3911613413198963e-06, "loss": 1.2914, "step": 11268 }, { "epoch": 0.8624674728302465, "grad_norm": 0.5855871843000455, "learning_rate": 2.39073381711657e-06, "loss": 1.2825, "step": 11269 }, { "epoch": 0.8625440073473136, "grad_norm": 0.6240116709872776, "learning_rate": 2.3903065146773476e-06, "loss": 1.3635, "step": 11270 }, { "epoch": 0.8626205418643809, "grad_norm": 0.8199028482295807, "learning_rate": 2.389879434028484e-06, "loss": 1.2232, "step": 11271 }, { "epoch": 0.862697076381448, "grad_norm": 0.6484557199488692, "learning_rate": 2.3894525751962196e-06, "loss": 1.5697, "step": 11272 }, { "epoch": 0.8627736108985152, "grad_norm": 0.6022494147314363, "learning_rate": 2.3890259382067825e-06, "loss": 1.4281, "step": 11273 }, { "epoch": 0.8628501454155825, "grad_norm": 0.7483397724106822, "learning_rate": 2.388599523086387e-06, "loss": 1.8177, "step": 11274 }, { "epoch": 0.8629266799326496, "grad_norm": 0.5183728362771367, "learning_rate": 2.388173329861233e-06, "loss": 1.1977, "step": 11275 }, { "epoch": 0.8630032144497168, "grad_norm": 0.5806828506611483, "learning_rate": 2.387747358557507e-06, "loss": 1.2958, "step": 11276 }, { "epoch": 0.863079748966784, "grad_norm": 0.6038671233013438, "learning_rate": 2.3873216092013816e-06, "loss": 1.3547, "step": 11277 }, { "epoch": 0.8631562834838512, "grad_norm": 0.7234938740142662, "learning_rate": 2.3868960818190157e-06, "loss": 1.6827, "step": 11278 }, { "epoch": 0.8632328180009184, "grad_norm": 0.6314956513562064, "learning_rate": 2.3864707764365568e-06, "loss": 1.3802, "step": 11279 }, { "epoch": 0.8633093525179856, "grad_norm": 0.48054958875772696, "learning_rate": 2.386045693080135e-06, "loss": 0.7722, "step": 11280 }, { "epoch": 0.8633858870350528, "grad_norm": 0.8263403203552384, "learning_rate": 2.38562083177587e-06, "loss": 1.6999, "step": 11281 }, { "epoch": 0.86346242155212, "grad_norm": 0.6722283746474532, "learning_rate": 2.3851961925498656e-06, "loss": 1.5178, "step": 11282 }, { "epoch": 0.8635389560691872, "grad_norm": 0.5608696462808939, "learning_rate": 2.3847717754282135e-06, "loss": 1.4365, "step": 11283 }, { "epoch": 0.8636154905862544, "grad_norm": 0.5997773930457135, "learning_rate": 2.3843475804369915e-06, "loss": 1.3753, "step": 11284 }, { "epoch": 0.8636920251033215, "grad_norm": 0.5781733380683912, "learning_rate": 2.383923607602262e-06, "loss": 1.3616, "step": 11285 }, { "epoch": 0.8637685596203888, "grad_norm": 0.5810151047536423, "learning_rate": 2.3834998569500767e-06, "loss": 1.4554, "step": 11286 }, { "epoch": 0.863845094137456, "grad_norm": 0.5541384138297468, "learning_rate": 2.3830763285064713e-06, "loss": 1.3406, "step": 11287 }, { "epoch": 0.8639216286545232, "grad_norm": 0.5212962122706429, "learning_rate": 2.382653022297469e-06, "loss": 1.1513, "step": 11288 }, { "epoch": 0.8639981631715904, "grad_norm": 0.566791376760568, "learning_rate": 2.3822299383490777e-06, "loss": 1.2509, "step": 11289 }, { "epoch": 0.8640746976886576, "grad_norm": 0.6817069081192061, "learning_rate": 2.3818070766872957e-06, "loss": 1.3333, "step": 11290 }, { "epoch": 0.8641512322057248, "grad_norm": 0.6567594994345801, "learning_rate": 2.381384437338102e-06, "loss": 1.5676, "step": 11291 }, { "epoch": 0.864227766722792, "grad_norm": 0.540352716013494, "learning_rate": 2.3809620203274668e-06, "loss": 1.0868, "step": 11292 }, { "epoch": 0.8643043012398591, "grad_norm": 0.6056366589573058, "learning_rate": 2.380539825681344e-06, "loss": 1.508, "step": 11293 }, { "epoch": 0.8643808357569264, "grad_norm": 0.5943921376628214, "learning_rate": 2.380117853425673e-06, "loss": 1.2986, "step": 11294 }, { "epoch": 0.8644573702739936, "grad_norm": 0.5348012653286283, "learning_rate": 2.379696103586384e-06, "loss": 1.1581, "step": 11295 }, { "epoch": 0.8645339047910607, "grad_norm": 0.8695560468843201, "learning_rate": 2.379274576189389e-06, "loss": 1.5301, "step": 11296 }, { "epoch": 0.864610439308128, "grad_norm": 0.5574515950466976, "learning_rate": 2.3788532712605874e-06, "loss": 1.2282, "step": 11297 }, { "epoch": 0.8646869738251952, "grad_norm": 0.5695565676053285, "learning_rate": 2.3784321888258663e-06, "loss": 1.2975, "step": 11298 }, { "epoch": 0.8647635083422623, "grad_norm": 0.6708599948353402, "learning_rate": 2.378011328911098e-06, "loss": 1.4754, "step": 11299 }, { "epoch": 0.8648400428593296, "grad_norm": 0.872756563013694, "learning_rate": 2.3775906915421406e-06, "loss": 1.4856, "step": 11300 }, { "epoch": 0.8649165773763967, "grad_norm": 0.9857372997567945, "learning_rate": 2.3771702767448406e-06, "loss": 1.3134, "step": 11301 }, { "epoch": 0.864993111893464, "grad_norm": 0.5671179865564699, "learning_rate": 2.3767500845450293e-06, "loss": 1.3697, "step": 11302 }, { "epoch": 0.8650696464105312, "grad_norm": 0.6071064823030478, "learning_rate": 2.376330114968524e-06, "loss": 1.4804, "step": 11303 }, { "epoch": 0.8651461809275983, "grad_norm": 0.5453392813542755, "learning_rate": 2.375910368041129e-06, "loss": 1.2263, "step": 11304 }, { "epoch": 0.8652227154446656, "grad_norm": 0.5959106729469322, "learning_rate": 2.375490843788635e-06, "loss": 1.3215, "step": 11305 }, { "epoch": 0.8652992499617327, "grad_norm": 0.6104296988255905, "learning_rate": 2.3750715422368192e-06, "loss": 1.3998, "step": 11306 }, { "epoch": 0.8653757844787999, "grad_norm": 0.5870591869836627, "learning_rate": 2.3746524634114445e-06, "loss": 1.4003, "step": 11307 }, { "epoch": 0.8654523189958672, "grad_norm": 0.6021296636904416, "learning_rate": 2.374233607338259e-06, "loss": 1.4257, "step": 11308 }, { "epoch": 0.8655288535129343, "grad_norm": 0.6386103097184509, "learning_rate": 2.3738149740430004e-06, "loss": 1.4946, "step": 11309 }, { "epoch": 0.8656053880300015, "grad_norm": 0.5433739506274682, "learning_rate": 2.37339656355139e-06, "loss": 1.2426, "step": 11310 }, { "epoch": 0.8656819225470688, "grad_norm": 0.5495609656064375, "learning_rate": 2.3729783758891356e-06, "loss": 1.3238, "step": 11311 }, { "epoch": 0.8657584570641359, "grad_norm": 0.5287539697991939, "learning_rate": 2.372560411081933e-06, "loss": 1.1136, "step": 11312 }, { "epoch": 0.8658349915812031, "grad_norm": 0.5025271480415326, "learning_rate": 2.372142669155463e-06, "loss": 1.075, "step": 11313 }, { "epoch": 0.8659115260982703, "grad_norm": 0.6353397116499628, "learning_rate": 2.3717251501353928e-06, "loss": 1.4641, "step": 11314 }, { "epoch": 0.8659880606153375, "grad_norm": 0.5052099873457707, "learning_rate": 2.3713078540473754e-06, "loss": 1.2224, "step": 11315 }, { "epoch": 0.8660645951324047, "grad_norm": 0.6359347965582025, "learning_rate": 2.3708907809170504e-06, "loss": 1.5723, "step": 11316 }, { "epoch": 0.8661411296494719, "grad_norm": 0.5618686588661151, "learning_rate": 2.3704739307700454e-06, "loss": 1.2096, "step": 11317 }, { "epoch": 0.8662176641665391, "grad_norm": 0.6733112252569654, "learning_rate": 2.3700573036319728e-06, "loss": 1.5462, "step": 11318 }, { "epoch": 0.8662941986836064, "grad_norm": 0.6349232758109096, "learning_rate": 2.3696408995284308e-06, "loss": 1.6134, "step": 11319 }, { "epoch": 0.8663707332006735, "grad_norm": 0.5958927872762354, "learning_rate": 2.369224718485005e-06, "loss": 1.5655, "step": 11320 }, { "epoch": 0.8664472677177407, "grad_norm": 0.6678834904148815, "learning_rate": 2.368808760527266e-06, "loss": 1.6754, "step": 11321 }, { "epoch": 0.8665238022348078, "grad_norm": 0.5675407694765133, "learning_rate": 2.368393025680771e-06, "loss": 1.3375, "step": 11322 }, { "epoch": 0.8666003367518751, "grad_norm": 0.5240437586605298, "learning_rate": 2.3679775139710658e-06, "loss": 1.2058, "step": 11323 }, { "epoch": 0.8666768712689423, "grad_norm": 0.5427175428548934, "learning_rate": 2.36756222542368e-06, "loss": 1.2271, "step": 11324 }, { "epoch": 0.8667534057860095, "grad_norm": 0.6575752545849295, "learning_rate": 2.3671471600641303e-06, "loss": 1.2706, "step": 11325 }, { "epoch": 0.8668299403030767, "grad_norm": 0.5545543110206553, "learning_rate": 2.3667323179179185e-06, "loss": 1.2741, "step": 11326 }, { "epoch": 0.8669064748201439, "grad_norm": 0.6015050141926022, "learning_rate": 2.366317699010535e-06, "loss": 1.4188, "step": 11327 }, { "epoch": 0.8669830093372111, "grad_norm": 1.2018849717168363, "learning_rate": 2.3659033033674546e-06, "loss": 1.3761, "step": 11328 }, { "epoch": 0.8670595438542783, "grad_norm": 0.7574783694005363, "learning_rate": 2.365489131014139e-06, "loss": 1.6559, "step": 11329 }, { "epoch": 0.8671360783713454, "grad_norm": 1.162051590651711, "learning_rate": 2.3650751819760366e-06, "loss": 1.6584, "step": 11330 }, { "epoch": 0.8672126128884127, "grad_norm": 1.2641161059348691, "learning_rate": 2.364661456278581e-06, "loss": 1.4817, "step": 11331 }, { "epoch": 0.8672891474054799, "grad_norm": 0.5397981692915615, "learning_rate": 2.364247953947193e-06, "loss": 1.2376, "step": 11332 }, { "epoch": 0.867365681922547, "grad_norm": 0.617142989837031, "learning_rate": 2.3638346750072796e-06, "loss": 1.5953, "step": 11333 }, { "epoch": 0.8674422164396143, "grad_norm": 0.6047149498818485, "learning_rate": 2.363421619484234e-06, "loss": 1.4899, "step": 11334 }, { "epoch": 0.8675187509566815, "grad_norm": 0.5781799235291301, "learning_rate": 2.3630087874034357e-06, "loss": 1.5036, "step": 11335 }, { "epoch": 0.8675952854737486, "grad_norm": 0.6875234094276499, "learning_rate": 2.3625961787902495e-06, "loss": 1.325, "step": 11336 }, { "epoch": 0.8676718199908159, "grad_norm": 0.618698866210584, "learning_rate": 2.3621837936700285e-06, "loss": 1.4025, "step": 11337 }, { "epoch": 0.867748354507883, "grad_norm": 0.5189495267200082, "learning_rate": 2.361771632068109e-06, "loss": 1.2451, "step": 11338 }, { "epoch": 0.8678248890249503, "grad_norm": 0.5122633806974287, "learning_rate": 2.3613596940098176e-06, "loss": 1.0539, "step": 11339 }, { "epoch": 0.8679014235420175, "grad_norm": 0.5867932731116063, "learning_rate": 2.360947979520464e-06, "loss": 1.3512, "step": 11340 }, { "epoch": 0.8679779580590846, "grad_norm": 0.5743828079545744, "learning_rate": 2.3605364886253453e-06, "loss": 1.391, "step": 11341 }, { "epoch": 0.8680544925761519, "grad_norm": 0.6009834312155691, "learning_rate": 2.3601252213497445e-06, "loss": 1.4389, "step": 11342 }, { "epoch": 0.868131027093219, "grad_norm": 0.7890008918275134, "learning_rate": 2.3597141777189308e-06, "loss": 1.4209, "step": 11343 }, { "epoch": 0.8682075616102862, "grad_norm": 0.5540241993336802, "learning_rate": 2.3593033577581608e-06, "loss": 1.094, "step": 11344 }, { "epoch": 0.8682840961273535, "grad_norm": 1.4381695707471267, "learning_rate": 2.358892761492676e-06, "loss": 1.4436, "step": 11345 }, { "epoch": 0.8683606306444206, "grad_norm": 0.9725108515242085, "learning_rate": 2.358482388947705e-06, "loss": 1.4724, "step": 11346 }, { "epoch": 0.8684371651614878, "grad_norm": 0.6352923654433846, "learning_rate": 2.3580722401484624e-06, "loss": 1.5274, "step": 11347 }, { "epoch": 0.8685136996785551, "grad_norm": 0.5131943327537718, "learning_rate": 2.3576623151201483e-06, "loss": 1.1488, "step": 11348 }, { "epoch": 0.8685902341956222, "grad_norm": 0.5442738477926005, "learning_rate": 2.3572526138879504e-06, "loss": 1.3725, "step": 11349 }, { "epoch": 0.8686667687126894, "grad_norm": 0.5044461852354934, "learning_rate": 2.3568431364770416e-06, "loss": 1.0731, "step": 11350 }, { "epoch": 0.8687433032297566, "grad_norm": 0.6633337711252277, "learning_rate": 2.356433882912581e-06, "loss": 1.6006, "step": 11351 }, { "epoch": 0.8688198377468238, "grad_norm": 0.9528873543311437, "learning_rate": 2.356024853219715e-06, "loss": 1.4369, "step": 11352 }, { "epoch": 0.868896372263891, "grad_norm": 0.6084897182953913, "learning_rate": 2.3556160474235756e-06, "loss": 1.5026, "step": 11353 }, { "epoch": 0.8689729067809582, "grad_norm": 0.9533102526748201, "learning_rate": 2.3552074655492803e-06, "loss": 1.3062, "step": 11354 }, { "epoch": 0.8690494412980254, "grad_norm": 0.5980833360389117, "learning_rate": 2.3547991076219345e-06, "loss": 1.319, "step": 11355 }, { "epoch": 0.8691259758150927, "grad_norm": 0.6096179062074473, "learning_rate": 2.3543909736666285e-06, "loss": 1.4115, "step": 11356 }, { "epoch": 0.8692025103321598, "grad_norm": 0.5128190532512443, "learning_rate": 2.35398306370844e-06, "loss": 1.1363, "step": 11357 }, { "epoch": 0.869279044849227, "grad_norm": 1.3521429852211495, "learning_rate": 2.353575377772431e-06, "loss": 1.1753, "step": 11358 }, { "epoch": 0.8693555793662942, "grad_norm": 0.6357027783876154, "learning_rate": 2.353167915883652e-06, "loss": 1.5354, "step": 11359 }, { "epoch": 0.8694321138833614, "grad_norm": 0.662636708013645, "learning_rate": 2.352760678067137e-06, "loss": 1.421, "step": 11360 }, { "epoch": 0.8695086484004286, "grad_norm": 0.5982043960901124, "learning_rate": 2.35235366434791e-06, "loss": 1.3145, "step": 11361 }, { "epoch": 0.8695851829174958, "grad_norm": 0.49572621523289767, "learning_rate": 2.3519468747509784e-06, "loss": 1.0276, "step": 11362 }, { "epoch": 0.869661717434563, "grad_norm": 0.7431027243402972, "learning_rate": 2.351540309301337e-06, "loss": 1.134, "step": 11363 }, { "epoch": 0.8697382519516302, "grad_norm": 0.8094390350558287, "learning_rate": 2.3511339680239645e-06, "loss": 1.824, "step": 11364 }, { "epoch": 0.8698147864686974, "grad_norm": 0.663600633230972, "learning_rate": 2.3507278509438287e-06, "loss": 1.2754, "step": 11365 }, { "epoch": 0.8698913209857646, "grad_norm": 0.549638879420167, "learning_rate": 2.350321958085884e-06, "loss": 1.265, "step": 11366 }, { "epoch": 0.8699678555028317, "grad_norm": 0.5720601919794479, "learning_rate": 2.349916289475069e-06, "loss": 1.2425, "step": 11367 }, { "epoch": 0.870044390019899, "grad_norm": 0.9219046130484971, "learning_rate": 2.3495108451363087e-06, "loss": 1.8091, "step": 11368 }, { "epoch": 0.8701209245369662, "grad_norm": 0.5404552223060267, "learning_rate": 2.3491056250945145e-06, "loss": 1.0946, "step": 11369 }, { "epoch": 0.8701974590540333, "grad_norm": 0.5394062328230439, "learning_rate": 2.348700629374585e-06, "loss": 1.2262, "step": 11370 }, { "epoch": 0.8702739935711006, "grad_norm": 0.7296410789050787, "learning_rate": 2.348295858001404e-06, "loss": 1.4368, "step": 11371 }, { "epoch": 0.8703505280881678, "grad_norm": 0.5828109998707993, "learning_rate": 2.3478913109998424e-06, "loss": 1.3013, "step": 11372 }, { "epoch": 0.870427062605235, "grad_norm": 0.5930109006835385, "learning_rate": 2.347486988394756e-06, "loss": 1.4148, "step": 11373 }, { "epoch": 0.8705035971223022, "grad_norm": 0.572948682951909, "learning_rate": 2.3470828902109878e-06, "loss": 1.3812, "step": 11374 }, { "epoch": 0.8705801316393693, "grad_norm": 0.5820641478043174, "learning_rate": 2.3466790164733678e-06, "loss": 1.3188, "step": 11375 }, { "epoch": 0.8706566661564366, "grad_norm": 0.6312825490768064, "learning_rate": 2.346275367206709e-06, "loss": 1.5201, "step": 11376 }, { "epoch": 0.8707332006735038, "grad_norm": 0.7130196959889005, "learning_rate": 2.345871942435815e-06, "loss": 1.5796, "step": 11377 }, { "epoch": 0.8708097351905709, "grad_norm": 0.7624598705113914, "learning_rate": 2.345468742185473e-06, "loss": 0.9737, "step": 11378 }, { "epoch": 0.8708862697076382, "grad_norm": 0.5782680860409208, "learning_rate": 2.345065766480456e-06, "loss": 1.3169, "step": 11379 }, { "epoch": 0.8709628042247053, "grad_norm": 1.4355163599058554, "learning_rate": 2.3446630153455246e-06, "loss": 1.5932, "step": 11380 }, { "epoch": 0.8710393387417725, "grad_norm": 0.531230871094273, "learning_rate": 2.344260488805425e-06, "loss": 1.3211, "step": 11381 }, { "epoch": 0.8711158732588398, "grad_norm": 0.9179475171001589, "learning_rate": 2.343858186884889e-06, "loss": 1.8192, "step": 11382 }, { "epoch": 0.8711924077759069, "grad_norm": 0.5433236785833517, "learning_rate": 2.343456109608636e-06, "loss": 1.1081, "step": 11383 }, { "epoch": 0.8712689422929741, "grad_norm": 0.5673123961333132, "learning_rate": 2.3430542570013705e-06, "loss": 1.3162, "step": 11384 }, { "epoch": 0.8713454768100414, "grad_norm": 0.6024646242111468, "learning_rate": 2.3426526290877844e-06, "loss": 1.3832, "step": 11385 }, { "epoch": 0.8714220113271085, "grad_norm": 0.6340774362774607, "learning_rate": 2.3422512258925533e-06, "loss": 1.5043, "step": 11386 }, { "epoch": 0.8714985458441757, "grad_norm": 0.5865858860797508, "learning_rate": 2.3418500474403417e-06, "loss": 1.2162, "step": 11387 }, { "epoch": 0.8715750803612429, "grad_norm": 0.5385286142367447, "learning_rate": 2.341449093755799e-06, "loss": 1.2522, "step": 11388 }, { "epoch": 0.8716516148783101, "grad_norm": 0.6115272052558915, "learning_rate": 2.341048364863561e-06, "loss": 1.3003, "step": 11389 }, { "epoch": 0.8717281493953773, "grad_norm": 0.529752210489656, "learning_rate": 2.3406478607882498e-06, "loss": 1.0634, "step": 11390 }, { "epoch": 0.8718046839124445, "grad_norm": 0.662797235480382, "learning_rate": 2.3402475815544735e-06, "loss": 1.3865, "step": 11391 }, { "epoch": 0.8718812184295117, "grad_norm": 0.8245483955691528, "learning_rate": 2.339847527186826e-06, "loss": 1.659, "step": 11392 }, { "epoch": 0.871957752946579, "grad_norm": 0.5997273020998211, "learning_rate": 2.339447697709889e-06, "loss": 1.4525, "step": 11393 }, { "epoch": 0.8720342874636461, "grad_norm": 0.5976342374105994, "learning_rate": 2.339048093148227e-06, "loss": 1.3989, "step": 11394 }, { "epoch": 0.8721108219807133, "grad_norm": 0.6001245848042684, "learning_rate": 2.3386487135263958e-06, "loss": 1.2237, "step": 11395 }, { "epoch": 0.8721873564977805, "grad_norm": 0.8443730879643465, "learning_rate": 2.3382495588689325e-06, "loss": 1.9325, "step": 11396 }, { "epoch": 0.8722638910148477, "grad_norm": 0.7148279913608634, "learning_rate": 2.337850629200363e-06, "loss": 1.333, "step": 11397 }, { "epoch": 0.8723404255319149, "grad_norm": 0.5228698852628746, "learning_rate": 2.3374519245451976e-06, "loss": 1.1307, "step": 11398 }, { "epoch": 0.8724169600489821, "grad_norm": 0.6748128926538216, "learning_rate": 2.3370534449279357e-06, "loss": 1.4308, "step": 11399 }, { "epoch": 0.8724934945660493, "grad_norm": 0.5082491341365161, "learning_rate": 2.33665519037306e-06, "loss": 0.9711, "step": 11400 }, { "epoch": 0.8725700290831165, "grad_norm": 0.8194045596748512, "learning_rate": 2.336257160905041e-06, "loss": 1.3449, "step": 11401 }, { "epoch": 0.8726465636001837, "grad_norm": 0.5656348152003939, "learning_rate": 2.3358593565483347e-06, "loss": 1.2571, "step": 11402 }, { "epoch": 0.8727230981172509, "grad_norm": 0.5368116822584849, "learning_rate": 2.335461777327383e-06, "loss": 1.1184, "step": 11403 }, { "epoch": 0.872799632634318, "grad_norm": 0.6361156264232752, "learning_rate": 2.335064423266614e-06, "loss": 1.5224, "step": 11404 }, { "epoch": 0.8728761671513853, "grad_norm": 0.5809583173030642, "learning_rate": 2.3346672943904432e-06, "loss": 1.3231, "step": 11405 }, { "epoch": 0.8729527016684525, "grad_norm": 0.524410461575977, "learning_rate": 2.334270390723271e-06, "loss": 1.2253, "step": 11406 }, { "epoch": 0.8730292361855196, "grad_norm": 0.626383088792394, "learning_rate": 2.333873712289485e-06, "loss": 1.3233, "step": 11407 }, { "epoch": 0.8731057707025869, "grad_norm": 0.6192573041234669, "learning_rate": 2.333477259113458e-06, "loss": 1.5344, "step": 11408 }, { "epoch": 0.873182305219654, "grad_norm": 0.5985581401609114, "learning_rate": 2.333081031219548e-06, "loss": 1.3909, "step": 11409 }, { "epoch": 0.8732588397367212, "grad_norm": 0.5994053494278971, "learning_rate": 2.3326850286321018e-06, "loss": 1.2899, "step": 11410 }, { "epoch": 0.8733353742537885, "grad_norm": 0.5723347608735828, "learning_rate": 2.3322892513754505e-06, "loss": 1.2839, "step": 11411 }, { "epoch": 0.8734119087708556, "grad_norm": 0.5574080919551764, "learning_rate": 2.3318936994739126e-06, "loss": 1.1815, "step": 11412 }, { "epoch": 0.8734884432879229, "grad_norm": 0.488885372547528, "learning_rate": 2.331498372951791e-06, "loss": 1.0681, "step": 11413 }, { "epoch": 0.8735649778049901, "grad_norm": 0.5319393347223157, "learning_rate": 2.3311032718333764e-06, "loss": 1.2007, "step": 11414 }, { "epoch": 0.8736415123220572, "grad_norm": 0.5813940287152615, "learning_rate": 2.3307083961429444e-06, "loss": 1.3268, "step": 11415 }, { "epoch": 0.8737180468391245, "grad_norm": 0.8139908832767813, "learning_rate": 2.330313745904758e-06, "loss": 1.4843, "step": 11416 }, { "epoch": 0.8737945813561916, "grad_norm": 0.5637149242210192, "learning_rate": 2.3299193211430647e-06, "loss": 1.3294, "step": 11417 }, { "epoch": 0.8738711158732588, "grad_norm": 0.5632925376658197, "learning_rate": 2.3295251218821e-06, "loss": 1.3528, "step": 11418 }, { "epoch": 0.8739476503903261, "grad_norm": 0.5184632999128106, "learning_rate": 2.3291311481460846e-06, "loss": 1.1235, "step": 11419 }, { "epoch": 0.8740241849073932, "grad_norm": 0.4791600237000965, "learning_rate": 2.3287373999592247e-06, "loss": 0.9941, "step": 11420 }, { "epoch": 0.8741007194244604, "grad_norm": 0.8155970353674312, "learning_rate": 2.3283438773457144e-06, "loss": 1.4579, "step": 11421 }, { "epoch": 0.8741772539415277, "grad_norm": 0.5528099430240254, "learning_rate": 2.3279505803297324e-06, "loss": 1.1888, "step": 11422 }, { "epoch": 0.8742537884585948, "grad_norm": 0.5180866044333002, "learning_rate": 2.327557508935444e-06, "loss": 1.1417, "step": 11423 }, { "epoch": 0.874330322975662, "grad_norm": 0.556799829006392, "learning_rate": 2.3271646631870013e-06, "loss": 1.1203, "step": 11424 }, { "epoch": 0.8744068574927292, "grad_norm": 0.5248420734159267, "learning_rate": 2.3267720431085407e-06, "loss": 1.2642, "step": 11425 }, { "epoch": 0.8744833920097964, "grad_norm": 0.5799576809737482, "learning_rate": 2.3263796487241862e-06, "loss": 1.2477, "step": 11426 }, { "epoch": 0.8745599265268637, "grad_norm": 0.6451089667248463, "learning_rate": 2.325987480058049e-06, "loss": 1.3409, "step": 11427 }, { "epoch": 0.8746364610439308, "grad_norm": 0.554856391457219, "learning_rate": 2.3255955371342243e-06, "loss": 1.377, "step": 11428 }, { "epoch": 0.874712995560998, "grad_norm": 0.5406444625029694, "learning_rate": 2.325203819976794e-06, "loss": 1.2255, "step": 11429 }, { "epoch": 0.8747895300780653, "grad_norm": 0.5653293384006727, "learning_rate": 2.324812328609826e-06, "loss": 1.2191, "step": 11430 }, { "epoch": 0.8748660645951324, "grad_norm": 0.6474937863412586, "learning_rate": 2.3244210630573755e-06, "loss": 1.4398, "step": 11431 }, { "epoch": 0.8749425991121996, "grad_norm": 1.4743254575782534, "learning_rate": 2.3240300233434827e-06, "loss": 1.6681, "step": 11432 }, { "epoch": 0.8750191336292668, "grad_norm": 0.5419540560421998, "learning_rate": 2.3236392094921748e-06, "loss": 1.1893, "step": 11433 }, { "epoch": 0.875095668146334, "grad_norm": 0.6574902308540435, "learning_rate": 2.3232486215274636e-06, "loss": 1.4748, "step": 11434 }, { "epoch": 0.8751722026634012, "grad_norm": 0.45120392411325494, "learning_rate": 2.3228582594733492e-06, "loss": 0.8151, "step": 11435 }, { "epoch": 0.8752487371804684, "grad_norm": 0.5520049385934925, "learning_rate": 2.3224681233538153e-06, "loss": 1.3079, "step": 11436 }, { "epoch": 0.8753252716975356, "grad_norm": 0.5490973271763311, "learning_rate": 2.322078213192834e-06, "loss": 1.2067, "step": 11437 }, { "epoch": 0.8754018062146028, "grad_norm": 0.5514221452208012, "learning_rate": 2.3216885290143622e-06, "loss": 1.3562, "step": 11438 }, { "epoch": 0.87547834073167, "grad_norm": 0.5719779841082585, "learning_rate": 2.321299070842343e-06, "loss": 1.388, "step": 11439 }, { "epoch": 0.8755548752487372, "grad_norm": 0.5873272943118364, "learning_rate": 2.3209098387007063e-06, "loss": 1.279, "step": 11440 }, { "epoch": 0.8756314097658043, "grad_norm": 0.6085172104018198, "learning_rate": 2.3205208326133673e-06, "loss": 1.4079, "step": 11441 }, { "epoch": 0.8757079442828716, "grad_norm": 0.6136441239682604, "learning_rate": 2.3201320526042274e-06, "loss": 1.4913, "step": 11442 }, { "epoch": 0.8757844787999388, "grad_norm": 0.5889066791676505, "learning_rate": 2.3197434986971755e-06, "loss": 1.3526, "step": 11443 }, { "epoch": 0.8758610133170059, "grad_norm": 0.5668005873108052, "learning_rate": 2.3193551709160848e-06, "loss": 1.3882, "step": 11444 }, { "epoch": 0.8759375478340732, "grad_norm": 0.6183088307968643, "learning_rate": 2.3189670692848153e-06, "loss": 1.5203, "step": 11445 }, { "epoch": 0.8760140823511403, "grad_norm": 0.5672666967168011, "learning_rate": 2.3185791938272136e-06, "loss": 1.3482, "step": 11446 }, { "epoch": 0.8760906168682076, "grad_norm": 0.6059808551504335, "learning_rate": 2.3181915445671118e-06, "loss": 1.0997, "step": 11447 }, { "epoch": 0.8761671513852748, "grad_norm": 0.5485286766319434, "learning_rate": 2.3178041215283267e-06, "loss": 1.0662, "step": 11448 }, { "epoch": 0.8762436859023419, "grad_norm": 0.8412839136682208, "learning_rate": 2.317416924734665e-06, "loss": 1.3619, "step": 11449 }, { "epoch": 0.8763202204194092, "grad_norm": 0.5672522594844674, "learning_rate": 2.317029954209916e-06, "loss": 1.3387, "step": 11450 }, { "epoch": 0.8763967549364764, "grad_norm": 0.9307674647320859, "learning_rate": 2.3166432099778575e-06, "loss": 1.5429, "step": 11451 }, { "epoch": 0.8764732894535435, "grad_norm": 0.6508138514326681, "learning_rate": 2.3162566920622503e-06, "loss": 1.5783, "step": 11452 }, { "epoch": 0.8765498239706108, "grad_norm": 0.5924597053115339, "learning_rate": 2.315870400486844e-06, "loss": 1.3668, "step": 11453 }, { "epoch": 0.8766263584876779, "grad_norm": 0.8462429850421364, "learning_rate": 2.3154843352753744e-06, "loss": 1.587, "step": 11454 }, { "epoch": 0.8767028930047451, "grad_norm": 0.6051982769937405, "learning_rate": 2.3150984964515623e-06, "loss": 1.534, "step": 11455 }, { "epoch": 0.8767794275218124, "grad_norm": 0.6813207225246918, "learning_rate": 2.3147128840391136e-06, "loss": 1.5174, "step": 11456 }, { "epoch": 0.8768559620388795, "grad_norm": 1.3350862257626475, "learning_rate": 2.3143274980617225e-06, "loss": 1.7392, "step": 11457 }, { "epoch": 0.8769324965559467, "grad_norm": 0.620443643219297, "learning_rate": 2.313942338543068e-06, "loss": 1.5773, "step": 11458 }, { "epoch": 0.877009031073014, "grad_norm": 0.527754622629589, "learning_rate": 2.313557405506815e-06, "loss": 1.1229, "step": 11459 }, { "epoch": 0.8770855655900811, "grad_norm": 0.5534764250333443, "learning_rate": 2.313172698976616e-06, "loss": 1.2289, "step": 11460 }, { "epoch": 0.8771621001071483, "grad_norm": 0.6381536157639484, "learning_rate": 2.3127882189761076e-06, "loss": 1.6054, "step": 11461 }, { "epoch": 0.8772386346242155, "grad_norm": 0.5268397293722039, "learning_rate": 2.312403965528914e-06, "loss": 1.1317, "step": 11462 }, { "epoch": 0.8773151691412827, "grad_norm": 0.527703242141025, "learning_rate": 2.312019938658644e-06, "loss": 1.3996, "step": 11463 }, { "epoch": 0.87739170365835, "grad_norm": 0.7080321029536559, "learning_rate": 2.311636138388894e-06, "loss": 1.7414, "step": 11464 }, { "epoch": 0.8774682381754171, "grad_norm": 0.5832742401204283, "learning_rate": 2.3112525647432457e-06, "loss": 1.5702, "step": 11465 }, { "epoch": 0.8775447726924843, "grad_norm": 0.5608411100496318, "learning_rate": 2.3108692177452672e-06, "loss": 1.3619, "step": 11466 }, { "epoch": 0.8776213072095516, "grad_norm": 0.5775111459922064, "learning_rate": 2.3104860974185126e-06, "loss": 1.3204, "step": 11467 }, { "epoch": 0.8776978417266187, "grad_norm": 1.3534435570721226, "learning_rate": 2.310103203786522e-06, "loss": 1.2618, "step": 11468 }, { "epoch": 0.8777743762436859, "grad_norm": 0.6499963208505106, "learning_rate": 2.309720536872821e-06, "loss": 1.5558, "step": 11469 }, { "epoch": 0.8778509107607531, "grad_norm": 0.5663162538891553, "learning_rate": 2.3093380967009216e-06, "loss": 1.3604, "step": 11470 }, { "epoch": 0.8779274452778203, "grad_norm": 0.5080493036327886, "learning_rate": 2.3089558832943228e-06, "loss": 1.1783, "step": 11471 }, { "epoch": 0.8780039797948875, "grad_norm": 0.5527435999963833, "learning_rate": 2.3085738966765087e-06, "loss": 1.0907, "step": 11472 }, { "epoch": 0.8780805143119547, "grad_norm": 0.5702893782514918, "learning_rate": 2.30819213687095e-06, "loss": 1.3023, "step": 11473 }, { "epoch": 0.8781570488290219, "grad_norm": 0.563472185881505, "learning_rate": 2.307810603901102e-06, "loss": 1.1086, "step": 11474 }, { "epoch": 0.8782335833460891, "grad_norm": 0.6808472537442561, "learning_rate": 2.307429297790408e-06, "loss": 1.4711, "step": 11475 }, { "epoch": 0.8783101178631563, "grad_norm": 0.5887839583551677, "learning_rate": 2.307048218562297e-06, "loss": 1.3277, "step": 11476 }, { "epoch": 0.8783866523802235, "grad_norm": 0.5680253180137839, "learning_rate": 2.3066673662401837e-06, "loss": 1.2255, "step": 11477 }, { "epoch": 0.8784631868972906, "grad_norm": 0.7355218794758104, "learning_rate": 2.306286740847468e-06, "loss": 1.5337, "step": 11478 }, { "epoch": 0.8785397214143579, "grad_norm": 0.5264437306797096, "learning_rate": 2.3059063424075374e-06, "loss": 1.2364, "step": 11479 }, { "epoch": 0.8786162559314251, "grad_norm": 0.6759211238055177, "learning_rate": 2.305526170943764e-06, "loss": 1.1565, "step": 11480 }, { "epoch": 0.8786927904484922, "grad_norm": 0.6385353903075759, "learning_rate": 2.3051462264795066e-06, "loss": 1.5579, "step": 11481 }, { "epoch": 0.8787693249655595, "grad_norm": 0.6950457399043033, "learning_rate": 2.304766509038111e-06, "loss": 1.2974, "step": 11482 }, { "epoch": 0.8788458594826266, "grad_norm": 0.5378752656096893, "learning_rate": 2.3043870186429086e-06, "loss": 1.2499, "step": 11483 }, { "epoch": 0.8789223939996939, "grad_norm": 0.5117582784757354, "learning_rate": 2.304007755317215e-06, "loss": 1.0761, "step": 11484 }, { "epoch": 0.8789989285167611, "grad_norm": 0.5735163747919734, "learning_rate": 2.3036287190843335e-06, "loss": 1.2875, "step": 11485 }, { "epoch": 0.8790754630338282, "grad_norm": 0.6503252571014129, "learning_rate": 2.303249909967553e-06, "loss": 1.2811, "step": 11486 }, { "epoch": 0.8791519975508955, "grad_norm": 0.5946605185033721, "learning_rate": 2.30287132799015e-06, "loss": 1.2931, "step": 11487 }, { "epoch": 0.8792285320679627, "grad_norm": 0.5832176944363382, "learning_rate": 2.3024929731753847e-06, "loss": 1.3296, "step": 11488 }, { "epoch": 0.8793050665850298, "grad_norm": 0.5950794091224635, "learning_rate": 2.3021148455465047e-06, "loss": 1.5486, "step": 11489 }, { "epoch": 0.8793816011020971, "grad_norm": 0.6752077482349489, "learning_rate": 2.3017369451267432e-06, "loss": 1.6212, "step": 11490 }, { "epoch": 0.8794581356191642, "grad_norm": 0.4729994532464845, "learning_rate": 2.3013592719393187e-06, "loss": 1.0422, "step": 11491 }, { "epoch": 0.8795346701362314, "grad_norm": 0.5497294976001941, "learning_rate": 2.3009818260074382e-06, "loss": 1.3813, "step": 11492 }, { "epoch": 0.8796112046532987, "grad_norm": 0.6233916408396414, "learning_rate": 2.300604607354292e-06, "loss": 1.4215, "step": 11493 }, { "epoch": 0.8796877391703658, "grad_norm": 0.6128196978116807, "learning_rate": 2.3002276160030585e-06, "loss": 1.4084, "step": 11494 }, { "epoch": 0.879764273687433, "grad_norm": 0.6222232351972081, "learning_rate": 2.2998508519769e-06, "loss": 1.3932, "step": 11495 }, { "epoch": 0.8798408082045003, "grad_norm": 1.1474222157069047, "learning_rate": 2.299474315298966e-06, "loss": 1.5155, "step": 11496 }, { "epoch": 0.8799173427215674, "grad_norm": 0.6564904167683144, "learning_rate": 2.2990980059923927e-06, "loss": 1.4421, "step": 11497 }, { "epoch": 0.8799938772386346, "grad_norm": 0.7792996622729866, "learning_rate": 2.2987219240803017e-06, "loss": 1.294, "step": 11498 }, { "epoch": 0.8800704117557018, "grad_norm": 0.6304105899037609, "learning_rate": 2.2983460695858006e-06, "loss": 1.5299, "step": 11499 }, { "epoch": 0.880146946272769, "grad_norm": 0.5896748608917188, "learning_rate": 2.297970442531983e-06, "loss": 1.4284, "step": 11500 }, { "epoch": 0.8802234807898363, "grad_norm": 0.6372004771780361, "learning_rate": 2.2975950429419276e-06, "loss": 1.6213, "step": 11501 }, { "epoch": 0.8803000153069034, "grad_norm": 0.5886968750697794, "learning_rate": 2.2972198708387008e-06, "loss": 1.344, "step": 11502 }, { "epoch": 0.8803765498239706, "grad_norm": 0.9193689551460227, "learning_rate": 2.296844926245355e-06, "loss": 1.7769, "step": 11503 }, { "epoch": 0.8804530843410379, "grad_norm": 0.6345363707945073, "learning_rate": 2.296470209184927e-06, "loss": 1.5407, "step": 11504 }, { "epoch": 0.880529618858105, "grad_norm": 0.552685251589556, "learning_rate": 2.2960957196804412e-06, "loss": 1.1954, "step": 11505 }, { "epoch": 0.8806061533751722, "grad_norm": 0.5904271691145225, "learning_rate": 2.2957214577549064e-06, "loss": 1.4334, "step": 11506 }, { "epoch": 0.8806826878922394, "grad_norm": 0.6758909166142759, "learning_rate": 2.2953474234313194e-06, "loss": 1.6431, "step": 11507 }, { "epoch": 0.8807592224093066, "grad_norm": 0.626276576503214, "learning_rate": 2.294973616732661e-06, "loss": 1.446, "step": 11508 }, { "epoch": 0.8808357569263738, "grad_norm": 0.78046303588066, "learning_rate": 2.2946000376819e-06, "loss": 1.7059, "step": 11509 }, { "epoch": 0.880912291443441, "grad_norm": 0.5874427781324475, "learning_rate": 2.2942266863019897e-06, "loss": 1.3022, "step": 11510 }, { "epoch": 0.8809888259605082, "grad_norm": 0.7157400663048196, "learning_rate": 2.2938535626158697e-06, "loss": 1.6899, "step": 11511 }, { "epoch": 0.8810653604775753, "grad_norm": 0.5541531234587634, "learning_rate": 2.2934806666464662e-06, "loss": 1.1972, "step": 11512 }, { "epoch": 0.8811418949946426, "grad_norm": 0.6130067052383339, "learning_rate": 2.2931079984166904e-06, "loss": 1.4692, "step": 11513 }, { "epoch": 0.8812184295117098, "grad_norm": 0.5757011757763184, "learning_rate": 2.2927355579494416e-06, "loss": 1.3963, "step": 11514 }, { "epoch": 0.8812949640287769, "grad_norm": 0.5917492501630098, "learning_rate": 2.292363345267602e-06, "loss": 1.3299, "step": 11515 }, { "epoch": 0.8813714985458442, "grad_norm": 0.5719209218178539, "learning_rate": 2.2919913603940425e-06, "loss": 1.3296, "step": 11516 }, { "epoch": 0.8814480330629114, "grad_norm": 0.6733418680970413, "learning_rate": 2.2916196033516185e-06, "loss": 1.6762, "step": 11517 }, { "epoch": 0.8815245675799785, "grad_norm": 0.647626489538385, "learning_rate": 2.291248074163172e-06, "loss": 1.5155, "step": 11518 }, { "epoch": 0.8816011020970458, "grad_norm": 0.7107060486601001, "learning_rate": 2.2908767728515307e-06, "loss": 1.7647, "step": 11519 }, { "epoch": 0.8816776366141129, "grad_norm": 0.6681526532359973, "learning_rate": 2.2905056994395093e-06, "loss": 1.5866, "step": 11520 }, { "epoch": 0.8817541711311802, "grad_norm": 0.624957457153626, "learning_rate": 2.2901348539499064e-06, "loss": 1.2886, "step": 11521 }, { "epoch": 0.8818307056482474, "grad_norm": 0.49332712630135905, "learning_rate": 2.289764236405509e-06, "loss": 1.0196, "step": 11522 }, { "epoch": 0.8819072401653145, "grad_norm": 0.5147397821768921, "learning_rate": 2.2893938468290877e-06, "loss": 1.1693, "step": 11523 }, { "epoch": 0.8819837746823818, "grad_norm": 0.5674661897833106, "learning_rate": 2.289023685243401e-06, "loss": 1.4641, "step": 11524 }, { "epoch": 0.882060309199449, "grad_norm": 0.6077905625859992, "learning_rate": 2.2886537516711934e-06, "loss": 1.453, "step": 11525 }, { "epoch": 0.8821368437165161, "grad_norm": 0.6333035885217736, "learning_rate": 2.288284046135194e-06, "loss": 1.5239, "step": 11526 }, { "epoch": 0.8822133782335834, "grad_norm": 0.6026721882812218, "learning_rate": 2.287914568658119e-06, "loss": 1.3264, "step": 11527 }, { "epoch": 0.8822899127506505, "grad_norm": 0.5558069644599544, "learning_rate": 2.2875453192626694e-06, "loss": 1.2909, "step": 11528 }, { "epoch": 0.8823664472677177, "grad_norm": 0.6233043931138633, "learning_rate": 2.287176297971534e-06, "loss": 1.3855, "step": 11529 }, { "epoch": 0.882442981784785, "grad_norm": 0.6384804288756644, "learning_rate": 2.286807504807386e-06, "loss": 1.5193, "step": 11530 }, { "epoch": 0.8825195163018521, "grad_norm": 0.4767619598776629, "learning_rate": 2.286438939792885e-06, "loss": 1.0705, "step": 11531 }, { "epoch": 0.8825960508189193, "grad_norm": 0.5520254423051923, "learning_rate": 2.286070602950677e-06, "loss": 1.2848, "step": 11532 }, { "epoch": 0.8826725853359866, "grad_norm": 0.5781386777754004, "learning_rate": 2.2857024943033936e-06, "loss": 1.2815, "step": 11533 }, { "epoch": 0.8827491198530537, "grad_norm": 0.5401485501348384, "learning_rate": 2.2853346138736527e-06, "loss": 1.1209, "step": 11534 }, { "epoch": 0.882825654370121, "grad_norm": 0.565827570822076, "learning_rate": 2.284966961684057e-06, "loss": 1.2644, "step": 11535 }, { "epoch": 0.8829021888871881, "grad_norm": 0.574615257435941, "learning_rate": 2.2845995377571984e-06, "loss": 1.2459, "step": 11536 }, { "epoch": 0.8829787234042553, "grad_norm": 0.5967562462120893, "learning_rate": 2.2842323421156504e-06, "loss": 1.3529, "step": 11537 }, { "epoch": 0.8830552579213226, "grad_norm": 0.5860179555066604, "learning_rate": 2.283865374781976e-06, "loss": 1.4955, "step": 11538 }, { "epoch": 0.8831317924383897, "grad_norm": 0.5921335523463478, "learning_rate": 2.2834986357787215e-06, "loss": 1.3182, "step": 11539 }, { "epoch": 0.8832083269554569, "grad_norm": 0.5883128598400866, "learning_rate": 2.283132125128421e-06, "loss": 1.3317, "step": 11540 }, { "epoch": 0.8832848614725242, "grad_norm": 0.6132319274339619, "learning_rate": 2.2827658428535936e-06, "loss": 1.3795, "step": 11541 }, { "epoch": 0.8833613959895913, "grad_norm": 0.5648844063256695, "learning_rate": 2.282399788976746e-06, "loss": 1.3366, "step": 11542 }, { "epoch": 0.8834379305066585, "grad_norm": 0.5328208343071894, "learning_rate": 2.2820339635203687e-06, "loss": 1.1246, "step": 11543 }, { "epoch": 0.8835144650237257, "grad_norm": 0.6813206412691208, "learning_rate": 2.2816683665069394e-06, "loss": 1.7503, "step": 11544 }, { "epoch": 0.8835909995407929, "grad_norm": 3.7724234363952545, "learning_rate": 2.2813029979589216e-06, "loss": 1.3948, "step": 11545 }, { "epoch": 0.8836675340578601, "grad_norm": 0.5492807470746758, "learning_rate": 2.2809378578987634e-06, "loss": 1.0753, "step": 11546 }, { "epoch": 0.8837440685749273, "grad_norm": 0.5942912511534366, "learning_rate": 2.280572946348902e-06, "loss": 1.5554, "step": 11547 }, { "epoch": 0.8838206030919945, "grad_norm": 1.2373589045631164, "learning_rate": 2.280208263331757e-06, "loss": 1.6483, "step": 11548 }, { "epoch": 0.8838971376090616, "grad_norm": 0.5079577504407947, "learning_rate": 2.279843808869737e-06, "loss": 0.9906, "step": 11549 }, { "epoch": 0.8839736721261289, "grad_norm": 0.6166706805318499, "learning_rate": 2.2794795829852346e-06, "loss": 1.5515, "step": 11550 }, { "epoch": 0.8840502066431961, "grad_norm": 0.5820295692027976, "learning_rate": 2.279115585700629e-06, "loss": 1.3403, "step": 11551 }, { "epoch": 0.8841267411602632, "grad_norm": 0.6000685947044874, "learning_rate": 2.2787518170382843e-06, "loss": 1.242, "step": 11552 }, { "epoch": 0.8842032756773305, "grad_norm": 0.5651699703306873, "learning_rate": 2.2783882770205535e-06, "loss": 1.3062, "step": 11553 }, { "epoch": 0.8842798101943977, "grad_norm": 0.6106437861799119, "learning_rate": 2.2780249656697717e-06, "loss": 1.2858, "step": 11554 }, { "epoch": 0.8843563447114648, "grad_norm": 0.7363640825338749, "learning_rate": 2.277661883008263e-06, "loss": 1.6385, "step": 11555 }, { "epoch": 0.8844328792285321, "grad_norm": 0.5829733710014416, "learning_rate": 2.277299029058335e-06, "loss": 1.3558, "step": 11556 }, { "epoch": 0.8845094137455992, "grad_norm": 0.6435402859470798, "learning_rate": 2.276936403842284e-06, "loss": 1.4456, "step": 11557 }, { "epoch": 0.8845859482626665, "grad_norm": 0.648518473143862, "learning_rate": 2.27657400738239e-06, "loss": 1.3169, "step": 11558 }, { "epoch": 0.8846624827797337, "grad_norm": 0.520666509997643, "learning_rate": 2.27621183970092e-06, "loss": 1.2151, "step": 11559 }, { "epoch": 0.8847390172968008, "grad_norm": 0.5413249898480319, "learning_rate": 2.275849900820127e-06, "loss": 1.1905, "step": 11560 }, { "epoch": 0.8848155518138681, "grad_norm": 0.8688229647446478, "learning_rate": 2.275488190762249e-06, "loss": 1.2682, "step": 11561 }, { "epoch": 0.8848920863309353, "grad_norm": 0.5460372944983958, "learning_rate": 2.2751267095495103e-06, "loss": 1.2394, "step": 11562 }, { "epoch": 0.8849686208480024, "grad_norm": 0.5640872524848869, "learning_rate": 2.2747654572041216e-06, "loss": 1.262, "step": 11563 }, { "epoch": 0.8850451553650697, "grad_norm": 0.6430488912714853, "learning_rate": 2.27440443374828e-06, "loss": 1.4375, "step": 11564 }, { "epoch": 0.8851216898821368, "grad_norm": 0.5469908885372606, "learning_rate": 2.2740436392041677e-06, "loss": 1.157, "step": 11565 }, { "epoch": 0.885198224399204, "grad_norm": 0.6198536813075873, "learning_rate": 2.2736830735939525e-06, "loss": 1.2871, "step": 11566 }, { "epoch": 0.8852747589162713, "grad_norm": 0.6035231006884384, "learning_rate": 2.273322736939789e-06, "loss": 1.407, "step": 11567 }, { "epoch": 0.8853512934333384, "grad_norm": 0.5990195817549181, "learning_rate": 2.2729626292638164e-06, "loss": 1.3214, "step": 11568 }, { "epoch": 0.8854278279504056, "grad_norm": 0.7162628852759215, "learning_rate": 2.272602750588162e-06, "loss": 1.5584, "step": 11569 }, { "epoch": 0.8855043624674729, "grad_norm": 0.8279044857336394, "learning_rate": 2.2722431009349377e-06, "loss": 1.5915, "step": 11570 }, { "epoch": 0.88558089698454, "grad_norm": 0.7287255618625851, "learning_rate": 2.271883680326241e-06, "loss": 1.344, "step": 11571 }, { "epoch": 0.8856574315016073, "grad_norm": 0.7246845195085143, "learning_rate": 2.271524488784156e-06, "loss": 1.5104, "step": 11572 }, { "epoch": 0.8857339660186744, "grad_norm": 0.758840942466027, "learning_rate": 2.271165526330752e-06, "loss": 1.4852, "step": 11573 }, { "epoch": 0.8858105005357416, "grad_norm": 0.6437807488913606, "learning_rate": 2.270806792988086e-06, "loss": 1.5346, "step": 11574 }, { "epoch": 0.8858870350528089, "grad_norm": 0.535931619724943, "learning_rate": 2.2704482887781985e-06, "loss": 1.1461, "step": 11575 }, { "epoch": 0.885963569569876, "grad_norm": 0.6375702746592474, "learning_rate": 2.2700900137231173e-06, "loss": 1.5201, "step": 11576 }, { "epoch": 0.8860401040869432, "grad_norm": 0.53197961697523, "learning_rate": 2.2697319678448557e-06, "loss": 1.0112, "step": 11577 }, { "epoch": 0.8861166386040104, "grad_norm": 0.5617618758807508, "learning_rate": 2.2693741511654134e-06, "loss": 1.2318, "step": 11578 }, { "epoch": 0.8861931731210776, "grad_norm": 0.6460397718810215, "learning_rate": 2.2690165637067756e-06, "loss": 1.5326, "step": 11579 }, { "epoch": 0.8862697076381448, "grad_norm": 0.7547759534728903, "learning_rate": 2.2686592054909137e-06, "loss": 1.4475, "step": 11580 }, { "epoch": 0.886346242155212, "grad_norm": 0.5089242277861802, "learning_rate": 2.268302076539785e-06, "loss": 1.031, "step": 11581 }, { "epoch": 0.8864227766722792, "grad_norm": 0.6010113553895304, "learning_rate": 2.2679451768753326e-06, "loss": 1.4837, "step": 11582 }, { "epoch": 0.8864993111893464, "grad_norm": 0.7343328353113332, "learning_rate": 2.267588506519485e-06, "loss": 1.5518, "step": 11583 }, { "epoch": 0.8865758457064136, "grad_norm": 0.6731316944012293, "learning_rate": 2.2672320654941576e-06, "loss": 1.7689, "step": 11584 }, { "epoch": 0.8866523802234808, "grad_norm": 0.6665921260427103, "learning_rate": 2.26687585382125e-06, "loss": 1.4913, "step": 11585 }, { "epoch": 0.8867289147405479, "grad_norm": 0.5663476856451921, "learning_rate": 2.2665198715226507e-06, "loss": 1.2477, "step": 11586 }, { "epoch": 0.8868054492576152, "grad_norm": 0.7141496813070755, "learning_rate": 2.2661641186202315e-06, "loss": 1.4395, "step": 11587 }, { "epoch": 0.8868819837746824, "grad_norm": 0.4771527157033732, "learning_rate": 2.265808595135851e-06, "loss": 0.8434, "step": 11588 }, { "epoch": 0.8869585182917495, "grad_norm": 0.6060864713266476, "learning_rate": 2.2654533010913526e-06, "loss": 1.4327, "step": 11589 }, { "epoch": 0.8870350528088168, "grad_norm": 0.5992284077396097, "learning_rate": 2.265098236508568e-06, "loss": 1.3505, "step": 11590 }, { "epoch": 0.887111587325884, "grad_norm": 0.696381450262288, "learning_rate": 2.264743401409313e-06, "loss": 1.1653, "step": 11591 }, { "epoch": 0.8871881218429511, "grad_norm": 2.736469339211982, "learning_rate": 2.26438879581539e-06, "loss": 1.7729, "step": 11592 }, { "epoch": 0.8872646563600184, "grad_norm": 0.5877000257078053, "learning_rate": 2.264034419748587e-06, "loss": 1.4978, "step": 11593 }, { "epoch": 0.8873411908770855, "grad_norm": 0.5903576148688605, "learning_rate": 2.2636802732306775e-06, "loss": 1.2281, "step": 11594 }, { "epoch": 0.8874177253941528, "grad_norm": 0.5976020505455181, "learning_rate": 2.2633263562834205e-06, "loss": 1.3746, "step": 11595 }, { "epoch": 0.88749425991122, "grad_norm": 0.654777747052687, "learning_rate": 2.2629726689285634e-06, "loss": 1.6933, "step": 11596 }, { "epoch": 0.8875707944282871, "grad_norm": 0.7088306915490397, "learning_rate": 2.2626192111878377e-06, "loss": 1.5523, "step": 11597 }, { "epoch": 0.8876473289453544, "grad_norm": 0.6916483050543007, "learning_rate": 2.2622659830829596e-06, "loss": 1.8564, "step": 11598 }, { "epoch": 0.8877238634624216, "grad_norm": 0.5780536180019576, "learning_rate": 2.2619129846356334e-06, "loss": 1.3609, "step": 11599 }, { "epoch": 0.8878003979794887, "grad_norm": 0.6944868606067724, "learning_rate": 2.261560215867549e-06, "loss": 1.5364, "step": 11600 }, { "epoch": 0.887876932496556, "grad_norm": 0.6072460963469908, "learning_rate": 2.2612076768003787e-06, "loss": 1.3633, "step": 11601 }, { "epoch": 0.8879534670136231, "grad_norm": 1.1321108870543648, "learning_rate": 2.260855367455787e-06, "loss": 1.4573, "step": 11602 }, { "epoch": 0.8880300015306903, "grad_norm": 0.6935605579689257, "learning_rate": 2.260503287855419e-06, "loss": 1.4233, "step": 11603 }, { "epoch": 0.8881065360477576, "grad_norm": 0.5985869086398483, "learning_rate": 2.2601514380209084e-06, "loss": 1.3946, "step": 11604 }, { "epoch": 0.8881830705648247, "grad_norm": 0.7505231382074002, "learning_rate": 2.259799817973873e-06, "loss": 1.7751, "step": 11605 }, { "epoch": 0.888259605081892, "grad_norm": 0.7237732555191623, "learning_rate": 2.2594484277359187e-06, "loss": 1.3016, "step": 11606 }, { "epoch": 0.8883361395989592, "grad_norm": 0.5555762822521881, "learning_rate": 2.2590972673286336e-06, "loss": 1.2672, "step": 11607 }, { "epoch": 0.8884126741160263, "grad_norm": 0.7818507857522582, "learning_rate": 2.2587463367735965e-06, "loss": 1.5456, "step": 11608 }, { "epoch": 0.8884892086330936, "grad_norm": 0.5801784736115906, "learning_rate": 2.2583956360923686e-06, "loss": 1.1091, "step": 11609 }, { "epoch": 0.8885657431501607, "grad_norm": 0.6670054705931031, "learning_rate": 2.2580451653064985e-06, "loss": 1.6085, "step": 11610 }, { "epoch": 0.8886422776672279, "grad_norm": 0.6271816127314204, "learning_rate": 2.25769492443752e-06, "loss": 1.529, "step": 11611 }, { "epoch": 0.8887188121842952, "grad_norm": 0.5455153023106595, "learning_rate": 2.2573449135069515e-06, "loss": 1.2425, "step": 11612 }, { "epoch": 0.8887953467013623, "grad_norm": 0.5225477808562293, "learning_rate": 2.2569951325363e-06, "loss": 1.2933, "step": 11613 }, { "epoch": 0.8888718812184295, "grad_norm": 0.9921062591328501, "learning_rate": 2.2566455815470577e-06, "loss": 1.5098, "step": 11614 }, { "epoch": 0.8889484157354967, "grad_norm": 0.5558245859127622, "learning_rate": 2.2562962605607014e-06, "loss": 1.3413, "step": 11615 }, { "epoch": 0.8890249502525639, "grad_norm": 0.587877541564106, "learning_rate": 2.255947169598694e-06, "loss": 1.397, "step": 11616 }, { "epoch": 0.8891014847696311, "grad_norm": 0.6263637028808342, "learning_rate": 2.255598308682486e-06, "loss": 1.4647, "step": 11617 }, { "epoch": 0.8891780192866983, "grad_norm": 0.6601381562122046, "learning_rate": 2.255249677833511e-06, "loss": 1.4627, "step": 11618 }, { "epoch": 0.8892545538037655, "grad_norm": 0.6397778863930247, "learning_rate": 2.25490127707319e-06, "loss": 1.5948, "step": 11619 }, { "epoch": 0.8893310883208327, "grad_norm": 0.6382672092610947, "learning_rate": 2.2545531064229307e-06, "loss": 1.5254, "step": 11620 }, { "epoch": 0.8894076228378999, "grad_norm": 0.6576173584076446, "learning_rate": 2.2542051659041255e-06, "loss": 1.3289, "step": 11621 }, { "epoch": 0.8894841573549671, "grad_norm": 0.5645985867755262, "learning_rate": 2.253857455538152e-06, "loss": 1.2085, "step": 11622 }, { "epoch": 0.8895606918720342, "grad_norm": 0.7799535825086387, "learning_rate": 2.2535099753463755e-06, "loss": 1.4919, "step": 11623 }, { "epoch": 0.8896372263891015, "grad_norm": 0.5843409152249616, "learning_rate": 2.2531627253501457e-06, "loss": 1.301, "step": 11624 }, { "epoch": 0.8897137609061687, "grad_norm": 0.4723371885082386, "learning_rate": 2.2528157055708e-06, "loss": 0.8791, "step": 11625 }, { "epoch": 0.8897902954232358, "grad_norm": 0.5685868368622761, "learning_rate": 2.2524689160296584e-06, "loss": 1.1851, "step": 11626 }, { "epoch": 0.8898668299403031, "grad_norm": 0.8810234819441909, "learning_rate": 2.25212235674803e-06, "loss": 1.6077, "step": 11627 }, { "epoch": 0.8899433644573703, "grad_norm": 0.6509425658575136, "learning_rate": 2.251776027747208e-06, "loss": 1.4479, "step": 11628 }, { "epoch": 0.8900198989744375, "grad_norm": 0.6149228762635192, "learning_rate": 2.251429929048471e-06, "loss": 1.4398, "step": 11629 }, { "epoch": 0.8900964334915047, "grad_norm": 0.7789256840436808, "learning_rate": 2.2510840606730864e-06, "loss": 1.546, "step": 11630 }, { "epoch": 0.8901729680085718, "grad_norm": 0.6107917263242166, "learning_rate": 2.250738422642304e-06, "loss": 1.3733, "step": 11631 }, { "epoch": 0.8902495025256391, "grad_norm": 0.49446624003078976, "learning_rate": 2.250393014977361e-06, "loss": 1.1363, "step": 11632 }, { "epoch": 0.8903260370427063, "grad_norm": 0.758772396603022, "learning_rate": 2.2500478376994805e-06, "loss": 1.2213, "step": 11633 }, { "epoch": 0.8904025715597734, "grad_norm": 0.5600745726308726, "learning_rate": 2.2497028908298703e-06, "loss": 1.2831, "step": 11634 }, { "epoch": 0.8904791060768407, "grad_norm": 0.5950844584886082, "learning_rate": 2.2493581743897263e-06, "loss": 1.3237, "step": 11635 }, { "epoch": 0.8905556405939079, "grad_norm": 0.5396483278833076, "learning_rate": 2.2490136884002287e-06, "loss": 1.1794, "step": 11636 }, { "epoch": 0.890632175110975, "grad_norm": 0.5959242225522188, "learning_rate": 2.248669432882543e-06, "loss": 1.3719, "step": 11637 }, { "epoch": 0.8907087096280423, "grad_norm": 0.6597760800998782, "learning_rate": 2.2483254078578217e-06, "loss": 1.6106, "step": 11638 }, { "epoch": 0.8907852441451094, "grad_norm": 0.6488563814745548, "learning_rate": 2.2479816133472026e-06, "loss": 1.7005, "step": 11639 }, { "epoch": 0.8908617786621766, "grad_norm": 0.622328050781568, "learning_rate": 2.24763804937181e-06, "loss": 1.4337, "step": 11640 }, { "epoch": 0.8909383131792439, "grad_norm": 0.61942390282602, "learning_rate": 2.247294715952753e-06, "loss": 1.4936, "step": 11641 }, { "epoch": 0.891014847696311, "grad_norm": 0.532230748612676, "learning_rate": 2.246951613111127e-06, "loss": 1.127, "step": 11642 }, { "epoch": 0.8910913822133782, "grad_norm": 0.5302664855932094, "learning_rate": 2.2466087408680133e-06, "loss": 1.0876, "step": 11643 }, { "epoch": 0.8911679167304455, "grad_norm": 0.5696477645141509, "learning_rate": 2.2462660992444795e-06, "loss": 1.2341, "step": 11644 }, { "epoch": 0.8912444512475126, "grad_norm": 0.5848268551487017, "learning_rate": 2.245923688261577e-06, "loss": 1.3286, "step": 11645 }, { "epoch": 0.8913209857645799, "grad_norm": 0.5887634752564169, "learning_rate": 2.2455815079403464e-06, "loss": 1.3717, "step": 11646 }, { "epoch": 0.891397520281647, "grad_norm": 0.7069695978630975, "learning_rate": 2.2452395583018112e-06, "loss": 1.4127, "step": 11647 }, { "epoch": 0.8914740547987142, "grad_norm": 0.5842712596067046, "learning_rate": 2.2448978393669827e-06, "loss": 1.439, "step": 11648 }, { "epoch": 0.8915505893157815, "grad_norm": 0.6158562881510797, "learning_rate": 2.2445563511568563e-06, "loss": 1.3549, "step": 11649 }, { "epoch": 0.8916271238328486, "grad_norm": 0.48345319573927475, "learning_rate": 2.2442150936924135e-06, "loss": 0.9685, "step": 11650 }, { "epoch": 0.8917036583499158, "grad_norm": 0.6267642799107708, "learning_rate": 2.243874066994624e-06, "loss": 1.5631, "step": 11651 }, { "epoch": 0.891780192866983, "grad_norm": 0.642421476769127, "learning_rate": 2.24353327108444e-06, "loss": 1.6414, "step": 11652 }, { "epoch": 0.8918567273840502, "grad_norm": 0.5641766377280086, "learning_rate": 2.2431927059828022e-06, "loss": 1.2754, "step": 11653 }, { "epoch": 0.8919332619011174, "grad_norm": 0.5891249455172599, "learning_rate": 2.2428523717106347e-06, "loss": 1.469, "step": 11654 }, { "epoch": 0.8920097964181846, "grad_norm": 0.5056922896375038, "learning_rate": 2.2425122682888494e-06, "loss": 1.1019, "step": 11655 }, { "epoch": 0.8920863309352518, "grad_norm": 0.5137752778294061, "learning_rate": 2.2421723957383428e-06, "loss": 1.1751, "step": 11656 }, { "epoch": 0.892162865452319, "grad_norm": 0.5877772180923156, "learning_rate": 2.241832754079998e-06, "loss": 1.2237, "step": 11657 }, { "epoch": 0.8922393999693862, "grad_norm": 0.5052095411395706, "learning_rate": 2.241493343334684e-06, "loss": 1.0731, "step": 11658 }, { "epoch": 0.8923159344864534, "grad_norm": 0.7145943785021517, "learning_rate": 2.2411541635232547e-06, "loss": 1.7288, "step": 11659 }, { "epoch": 0.8923924690035205, "grad_norm": 0.6122962264651626, "learning_rate": 2.2408152146665503e-06, "loss": 1.466, "step": 11660 }, { "epoch": 0.8924690035205878, "grad_norm": 0.5086429472122781, "learning_rate": 2.240476496785397e-06, "loss": 1.001, "step": 11661 }, { "epoch": 0.892545538037655, "grad_norm": 0.6051749276299425, "learning_rate": 2.2401380099006066e-06, "loss": 1.4296, "step": 11662 }, { "epoch": 0.8926220725547221, "grad_norm": 0.6243760771898073, "learning_rate": 2.2397997540329766e-06, "loss": 1.2885, "step": 11663 }, { "epoch": 0.8926986070717894, "grad_norm": 1.0954795408649822, "learning_rate": 2.2394617292032903e-06, "loss": 1.5959, "step": 11664 }, { "epoch": 0.8927751415888566, "grad_norm": 1.3929247000038367, "learning_rate": 2.2391239354323176e-06, "loss": 1.4617, "step": 11665 }, { "epoch": 0.8928516761059238, "grad_norm": 0.5881003151967001, "learning_rate": 2.2387863727408135e-06, "loss": 1.2165, "step": 11666 }, { "epoch": 0.892928210622991, "grad_norm": 0.8067977322803268, "learning_rate": 2.2384490411495174e-06, "loss": 1.5748, "step": 11667 }, { "epoch": 0.8930047451400581, "grad_norm": 0.7716140437731768, "learning_rate": 2.238111940679158e-06, "loss": 1.3836, "step": 11668 }, { "epoch": 0.8930812796571254, "grad_norm": 0.5436440587346529, "learning_rate": 2.2377750713504466e-06, "loss": 1.1626, "step": 11669 }, { "epoch": 0.8931578141741926, "grad_norm": 0.5609492003515139, "learning_rate": 2.2374384331840822e-06, "loss": 1.3283, "step": 11670 }, { "epoch": 0.8932343486912597, "grad_norm": 0.5586684175276043, "learning_rate": 2.237102026200748e-06, "loss": 1.2332, "step": 11671 }, { "epoch": 0.893310883208327, "grad_norm": 0.6126031502425909, "learning_rate": 2.2367658504211138e-06, "loss": 1.6306, "step": 11672 }, { "epoch": 0.8933874177253942, "grad_norm": 0.6697031968656986, "learning_rate": 2.236429905865836e-06, "loss": 1.5636, "step": 11673 }, { "epoch": 0.8934639522424613, "grad_norm": 0.5474727829940773, "learning_rate": 2.236094192555556e-06, "loss": 1.4353, "step": 11674 }, { "epoch": 0.8935404867595286, "grad_norm": 0.5339949544160303, "learning_rate": 2.2357587105109004e-06, "loss": 1.2076, "step": 11675 }, { "epoch": 0.8936170212765957, "grad_norm": 0.882241617934221, "learning_rate": 2.235423459752483e-06, "loss": 1.2353, "step": 11676 }, { "epoch": 0.8936935557936629, "grad_norm": 0.5749555850669672, "learning_rate": 2.235088440300902e-06, "loss": 1.257, "step": 11677 }, { "epoch": 0.8937700903107302, "grad_norm": 0.6110574024192922, "learning_rate": 2.2347536521767417e-06, "loss": 1.4867, "step": 11678 }, { "epoch": 0.8938466248277973, "grad_norm": 0.6422450414598734, "learning_rate": 2.234419095400573e-06, "loss": 1.1554, "step": 11679 }, { "epoch": 0.8939231593448645, "grad_norm": 0.6044935266804832, "learning_rate": 2.2340847699929526e-06, "loss": 1.4974, "step": 11680 }, { "epoch": 0.8939996938619317, "grad_norm": 0.9258060143005707, "learning_rate": 2.233750675974422e-06, "loss": 1.5404, "step": 11681 }, { "epoch": 0.8940762283789989, "grad_norm": 0.762427427463425, "learning_rate": 2.2334168133655087e-06, "loss": 1.6289, "step": 11682 }, { "epoch": 0.8941527628960662, "grad_norm": 0.5627159907498002, "learning_rate": 2.233083182186726e-06, "loss": 1.3622, "step": 11683 }, { "epoch": 0.8942292974131333, "grad_norm": 0.4530379397063759, "learning_rate": 2.2327497824585737e-06, "loss": 0.9077, "step": 11684 }, { "epoch": 0.8943058319302005, "grad_norm": 0.5541368796799406, "learning_rate": 2.2324166142015366e-06, "loss": 1.344, "step": 11685 }, { "epoch": 0.8943823664472678, "grad_norm": 0.5351673204668553, "learning_rate": 2.2320836774360858e-06, "loss": 1.1341, "step": 11686 }, { "epoch": 0.8944589009643349, "grad_norm": 0.5357300868925344, "learning_rate": 2.231750972182678e-06, "loss": 1.344, "step": 11687 }, { "epoch": 0.8945354354814021, "grad_norm": 0.5892547916863953, "learning_rate": 2.231418498461755e-06, "loss": 1.3046, "step": 11688 }, { "epoch": 0.8946119699984693, "grad_norm": 0.5859408352925657, "learning_rate": 2.231086256293745e-06, "loss": 1.302, "step": 11689 }, { "epoch": 0.8946885045155365, "grad_norm": 0.7366600460812496, "learning_rate": 2.230754245699063e-06, "loss": 1.2883, "step": 11690 }, { "epoch": 0.8947650390326037, "grad_norm": 0.5629459124915043, "learning_rate": 2.2304224666981074e-06, "loss": 1.3137, "step": 11691 }, { "epoch": 0.8948415735496709, "grad_norm": 2.434885830711829, "learning_rate": 2.2300909193112645e-06, "loss": 1.3124, "step": 11692 }, { "epoch": 0.8949181080667381, "grad_norm": 1.3332399382839162, "learning_rate": 2.229759603558906e-06, "loss": 1.2621, "step": 11693 }, { "epoch": 0.8949946425838053, "grad_norm": 0.5318079137450592, "learning_rate": 2.2294285194613873e-06, "loss": 1.2124, "step": 11694 }, { "epoch": 0.8950711771008725, "grad_norm": 0.5692125564125922, "learning_rate": 2.229097667039053e-06, "loss": 1.2573, "step": 11695 }, { "epoch": 0.8951477116179397, "grad_norm": 0.6240218303564385, "learning_rate": 2.228767046312231e-06, "loss": 1.5547, "step": 11696 }, { "epoch": 0.8952242461350068, "grad_norm": 0.5811818502572252, "learning_rate": 2.228436657301235e-06, "loss": 1.2276, "step": 11697 }, { "epoch": 0.8953007806520741, "grad_norm": 1.6606756773901699, "learning_rate": 2.228106500026366e-06, "loss": 1.647, "step": 11698 }, { "epoch": 0.8953773151691413, "grad_norm": 0.556001521371838, "learning_rate": 2.2277765745079098e-06, "loss": 1.3347, "step": 11699 }, { "epoch": 0.8954538496862084, "grad_norm": 0.49576102808925376, "learning_rate": 2.2274468807661365e-06, "loss": 1.2141, "step": 11700 }, { "epoch": 0.8955303842032757, "grad_norm": 0.6136868063521304, "learning_rate": 2.2271174188213053e-06, "loss": 1.3422, "step": 11701 }, { "epoch": 0.8956069187203429, "grad_norm": 0.6469571500826301, "learning_rate": 2.226788188693659e-06, "loss": 1.5072, "step": 11702 }, { "epoch": 0.89568345323741, "grad_norm": 0.502367713406526, "learning_rate": 2.226459190403426e-06, "loss": 1.1728, "step": 11703 }, { "epoch": 0.8957599877544773, "grad_norm": 0.7046410034956215, "learning_rate": 2.2261304239708206e-06, "loss": 1.3689, "step": 11704 }, { "epoch": 0.8958365222715444, "grad_norm": 0.6286528776525803, "learning_rate": 2.225801889416044e-06, "loss": 1.2516, "step": 11705 }, { "epoch": 0.8959130567886117, "grad_norm": 0.5319862913244991, "learning_rate": 2.225473586759282e-06, "loss": 1.1085, "step": 11706 }, { "epoch": 0.8959895913056789, "grad_norm": 0.53788508432203, "learning_rate": 2.225145516020706e-06, "loss": 1.1829, "step": 11707 }, { "epoch": 0.896066125822746, "grad_norm": 0.567937771796732, "learning_rate": 2.2248176772204754e-06, "loss": 1.2828, "step": 11708 }, { "epoch": 0.8961426603398133, "grad_norm": 0.6927134958367099, "learning_rate": 2.224490070378732e-06, "loss": 1.6671, "step": 11709 }, { "epoch": 0.8962191948568805, "grad_norm": 0.5902350506266113, "learning_rate": 2.2241626955156047e-06, "loss": 1.4513, "step": 11710 }, { "epoch": 0.8962957293739476, "grad_norm": 0.6786343407651116, "learning_rate": 2.223835552651209e-06, "loss": 1.5088, "step": 11711 }, { "epoch": 0.8963722638910149, "grad_norm": 0.656337827997032, "learning_rate": 2.223508641805646e-06, "loss": 1.4234, "step": 11712 }, { "epoch": 0.896448798408082, "grad_norm": 0.5947706708705702, "learning_rate": 2.2231819629990014e-06, "loss": 1.4839, "step": 11713 }, { "epoch": 0.8965253329251492, "grad_norm": 0.5310289521922331, "learning_rate": 2.2228555162513476e-06, "loss": 1.1216, "step": 11714 }, { "epoch": 0.8966018674422165, "grad_norm": 0.6108152563486019, "learning_rate": 2.222529301582742e-06, "loss": 1.4711, "step": 11715 }, { "epoch": 0.8966784019592836, "grad_norm": 0.6964953251519814, "learning_rate": 2.2222033190132285e-06, "loss": 1.4186, "step": 11716 }, { "epoch": 0.8967549364763509, "grad_norm": 0.5536968769244425, "learning_rate": 2.2218775685628367e-06, "loss": 1.0852, "step": 11717 }, { "epoch": 0.896831470993418, "grad_norm": 0.5981983536246022, "learning_rate": 2.221552050251582e-06, "loss": 1.2026, "step": 11718 }, { "epoch": 0.8969080055104852, "grad_norm": 0.5964782189073996, "learning_rate": 2.2212267640994646e-06, "loss": 1.3571, "step": 11719 }, { "epoch": 0.8969845400275525, "grad_norm": 1.617282592922193, "learning_rate": 2.2209017101264716e-06, "loss": 1.4087, "step": 11720 }, { "epoch": 0.8970610745446196, "grad_norm": 0.5837608072365131, "learning_rate": 2.220576888352574e-06, "loss": 1.3879, "step": 11721 }, { "epoch": 0.8971376090616868, "grad_norm": 0.6580663536469001, "learning_rate": 2.2202522987977302e-06, "loss": 1.4677, "step": 11722 }, { "epoch": 0.8972141435787541, "grad_norm": 0.6516925836831565, "learning_rate": 2.219927941481886e-06, "loss": 1.5554, "step": 11723 }, { "epoch": 0.8972906780958212, "grad_norm": 0.7320813066796236, "learning_rate": 2.219603816424969e-06, "loss": 1.6797, "step": 11724 }, { "epoch": 0.8973672126128884, "grad_norm": 0.7643340644323688, "learning_rate": 2.219279923646894e-06, "loss": 1.4899, "step": 11725 }, { "epoch": 0.8974437471299556, "grad_norm": 0.5270903716485006, "learning_rate": 2.218956263167563e-06, "loss": 1.1465, "step": 11726 }, { "epoch": 0.8975202816470228, "grad_norm": 0.5666594600350079, "learning_rate": 2.2186328350068623e-06, "loss": 1.2616, "step": 11727 }, { "epoch": 0.89759681616409, "grad_norm": 0.6218076193952339, "learning_rate": 2.2183096391846645e-06, "loss": 1.586, "step": 11728 }, { "epoch": 0.8976733506811572, "grad_norm": 0.5867005634553768, "learning_rate": 2.217986675720828e-06, "loss": 1.3052, "step": 11729 }, { "epoch": 0.8977498851982244, "grad_norm": 0.7543061581166158, "learning_rate": 2.217663944635196e-06, "loss": 1.5405, "step": 11730 }, { "epoch": 0.8978264197152916, "grad_norm": 0.5669263785556854, "learning_rate": 2.217341445947598e-06, "loss": 1.3079, "step": 11731 }, { "epoch": 0.8979029542323588, "grad_norm": 0.70849952260487, "learning_rate": 2.2170191796778495e-06, "loss": 1.5204, "step": 11732 }, { "epoch": 0.897979488749426, "grad_norm": 0.5071125248118028, "learning_rate": 2.216697145845752e-06, "loss": 1.0911, "step": 11733 }, { "epoch": 0.8980560232664931, "grad_norm": 0.5676345797542701, "learning_rate": 2.2163753444710913e-06, "loss": 1.2719, "step": 11734 }, { "epoch": 0.8981325577835604, "grad_norm": 0.5233623193815302, "learning_rate": 2.2160537755736403e-06, "loss": 1.1868, "step": 11735 }, { "epoch": 0.8982090923006276, "grad_norm": 0.5883632353216712, "learning_rate": 2.215732439173158e-06, "loss": 1.0743, "step": 11736 }, { "epoch": 0.8982856268176947, "grad_norm": 0.6200711505114096, "learning_rate": 2.215411335289387e-06, "loss": 1.6147, "step": 11737 }, { "epoch": 0.898362161334762, "grad_norm": 0.6445123614285502, "learning_rate": 2.2150904639420565e-06, "loss": 1.6881, "step": 11738 }, { "epoch": 0.8984386958518292, "grad_norm": 0.6051895971786836, "learning_rate": 2.2147698251508833e-06, "loss": 1.2726, "step": 11739 }, { "epoch": 0.8985152303688964, "grad_norm": 0.5814089982869314, "learning_rate": 2.2144494189355676e-06, "loss": 1.476, "step": 11740 }, { "epoch": 0.8985917648859636, "grad_norm": 0.635671685607053, "learning_rate": 2.2141292453157965e-06, "loss": 1.449, "step": 11741 }, { "epoch": 0.8986682994030307, "grad_norm": 0.650744570508919, "learning_rate": 2.213809304311242e-06, "loss": 1.1854, "step": 11742 }, { "epoch": 0.898744833920098, "grad_norm": 0.5091349285883346, "learning_rate": 2.213489595941563e-06, "loss": 1.1836, "step": 11743 }, { "epoch": 0.8988213684371652, "grad_norm": 0.5187206931464637, "learning_rate": 2.213170120226401e-06, "loss": 1.1597, "step": 11744 }, { "epoch": 0.8988979029542323, "grad_norm": 0.5671386877249387, "learning_rate": 2.212850877185389e-06, "loss": 1.3208, "step": 11745 }, { "epoch": 0.8989744374712996, "grad_norm": 0.6310109503074501, "learning_rate": 2.2125318668381397e-06, "loss": 1.5256, "step": 11746 }, { "epoch": 0.8990509719883667, "grad_norm": 0.5804549269542764, "learning_rate": 2.2122130892042553e-06, "loss": 1.3122, "step": 11747 }, { "epoch": 0.8991275065054339, "grad_norm": 0.8978188730805021, "learning_rate": 2.211894544303322e-06, "loss": 1.5907, "step": 11748 }, { "epoch": 0.8992040410225012, "grad_norm": 0.609245555921343, "learning_rate": 2.211576232154911e-06, "loss": 1.4837, "step": 11749 }, { "epoch": 0.8992805755395683, "grad_norm": 0.5976530906164464, "learning_rate": 2.211258152778583e-06, "loss": 1.3446, "step": 11750 }, { "epoch": 0.8993571100566355, "grad_norm": 0.5737395866653395, "learning_rate": 2.21094030619388e-06, "loss": 1.2808, "step": 11751 }, { "epoch": 0.8994336445737028, "grad_norm": 0.4803386928004548, "learning_rate": 2.210622692420332e-06, "loss": 1.029, "step": 11752 }, { "epoch": 0.8995101790907699, "grad_norm": 0.5687186616928709, "learning_rate": 2.210305311477453e-06, "loss": 1.3286, "step": 11753 }, { "epoch": 0.8995867136078372, "grad_norm": 0.6095800393063427, "learning_rate": 2.209988163384745e-06, "loss": 1.3592, "step": 11754 }, { "epoch": 0.8996632481249043, "grad_norm": 0.5208647972376503, "learning_rate": 2.2096712481616944e-06, "loss": 1.1182, "step": 11755 }, { "epoch": 0.8997397826419715, "grad_norm": 0.598635547368062, "learning_rate": 2.2093545658277733e-06, "loss": 1.4674, "step": 11756 }, { "epoch": 0.8998163171590388, "grad_norm": 0.6041153497053292, "learning_rate": 2.20903811640244e-06, "loss": 1.5312, "step": 11757 }, { "epoch": 0.8998928516761059, "grad_norm": 0.5985568364941315, "learning_rate": 2.208721899905137e-06, "loss": 1.533, "step": 11758 }, { "epoch": 0.8999693861931731, "grad_norm": 0.5000485390093333, "learning_rate": 2.208405916355294e-06, "loss": 0.9866, "step": 11759 }, { "epoch": 0.9000459207102404, "grad_norm": 0.5444385020188511, "learning_rate": 2.2080901657723262e-06, "loss": 1.324, "step": 11760 }, { "epoch": 0.9001224552273075, "grad_norm": 0.5788422565410763, "learning_rate": 2.2077746481756346e-06, "loss": 1.3301, "step": 11761 }, { "epoch": 0.9001989897443747, "grad_norm": 0.5271026180372611, "learning_rate": 2.2074593635846053e-06, "loss": 1.1352, "step": 11762 }, { "epoch": 0.9002755242614419, "grad_norm": 0.5620407022677918, "learning_rate": 2.20714431201861e-06, "loss": 1.3856, "step": 11763 }, { "epoch": 0.9003520587785091, "grad_norm": 0.6322463198946485, "learning_rate": 2.206829493497007e-06, "loss": 1.4507, "step": 11764 }, { "epoch": 0.9004285932955763, "grad_norm": 0.5966076355206952, "learning_rate": 2.206514908039139e-06, "loss": 1.5232, "step": 11765 }, { "epoch": 0.9005051278126435, "grad_norm": 0.5986127651629625, "learning_rate": 2.2062005556643353e-06, "loss": 1.5461, "step": 11766 }, { "epoch": 0.9005816623297107, "grad_norm": 0.9345432997060545, "learning_rate": 2.205886436391911e-06, "loss": 1.7051, "step": 11767 }, { "epoch": 0.900658196846778, "grad_norm": 0.855093090206946, "learning_rate": 2.2055725502411666e-06, "loss": 1.4926, "step": 11768 }, { "epoch": 0.9007347313638451, "grad_norm": 0.6592671795025961, "learning_rate": 2.2052588972313875e-06, "loss": 1.2595, "step": 11769 }, { "epoch": 0.9008112658809123, "grad_norm": 0.6242378210421589, "learning_rate": 2.204945477381846e-06, "loss": 1.3774, "step": 11770 }, { "epoch": 0.9008878003979794, "grad_norm": 0.633449311108087, "learning_rate": 2.204632290711799e-06, "loss": 1.5865, "step": 11771 }, { "epoch": 0.9009643349150467, "grad_norm": 0.722246617093964, "learning_rate": 2.2043193372404904e-06, "loss": 1.6403, "step": 11772 }, { "epoch": 0.9010408694321139, "grad_norm": 0.6080409595292524, "learning_rate": 2.204006616987149e-06, "loss": 1.4341, "step": 11773 }, { "epoch": 0.901117403949181, "grad_norm": 0.7150279641203582, "learning_rate": 2.203694129970988e-06, "loss": 1.3502, "step": 11774 }, { "epoch": 0.9011939384662483, "grad_norm": 0.5633606288205081, "learning_rate": 2.203381876211209e-06, "loss": 1.1783, "step": 11775 }, { "epoch": 0.9012704729833155, "grad_norm": 0.5814072395878508, "learning_rate": 2.203069855726998e-06, "loss": 1.2345, "step": 11776 }, { "epoch": 0.9013470075003827, "grad_norm": 0.5439314743523286, "learning_rate": 2.202758068537525e-06, "loss": 1.2768, "step": 11777 }, { "epoch": 0.9014235420174499, "grad_norm": 1.0664237129307281, "learning_rate": 2.2024465146619476e-06, "loss": 1.6584, "step": 11778 }, { "epoch": 0.901500076534517, "grad_norm": 0.6312129452912791, "learning_rate": 2.2021351941194092e-06, "loss": 1.5888, "step": 11779 }, { "epoch": 0.9015766110515843, "grad_norm": 0.8003937543389361, "learning_rate": 2.201824106929038e-06, "loss": 1.6552, "step": 11780 }, { "epoch": 0.9016531455686515, "grad_norm": 0.5136534600701547, "learning_rate": 2.201513253109948e-06, "loss": 1.1996, "step": 11781 }, { "epoch": 0.9017296800857186, "grad_norm": 0.6017831892581635, "learning_rate": 2.2012026326812385e-06, "loss": 1.2187, "step": 11782 }, { "epoch": 0.9018062146027859, "grad_norm": 0.6108489511617816, "learning_rate": 2.2008922456619957e-06, "loss": 1.5792, "step": 11783 }, { "epoch": 0.901882749119853, "grad_norm": 0.5710505247221063, "learning_rate": 2.200582092071291e-06, "loss": 1.2331, "step": 11784 }, { "epoch": 0.9019592836369202, "grad_norm": 0.6162802562593158, "learning_rate": 2.2002721719281807e-06, "loss": 1.4588, "step": 11785 }, { "epoch": 0.9020358181539875, "grad_norm": 0.5512835201543468, "learning_rate": 2.1999624852517067e-06, "loss": 1.2137, "step": 11786 }, { "epoch": 0.9021123526710546, "grad_norm": 0.5979211216013942, "learning_rate": 2.199653032060897e-06, "loss": 1.48, "step": 11787 }, { "epoch": 0.9021888871881218, "grad_norm": 0.5912521665741716, "learning_rate": 2.1993438123747667e-06, "loss": 1.3601, "step": 11788 }, { "epoch": 0.9022654217051891, "grad_norm": 0.614189671176686, "learning_rate": 2.1990348262123147e-06, "loss": 1.3706, "step": 11789 }, { "epoch": 0.9023419562222562, "grad_norm": 0.5376821705049049, "learning_rate": 2.1987260735925252e-06, "loss": 1.2405, "step": 11790 }, { "epoch": 0.9024184907393235, "grad_norm": 0.9820614743709307, "learning_rate": 2.1984175545343697e-06, "loss": 1.4258, "step": 11791 }, { "epoch": 0.9024950252563906, "grad_norm": 0.7051732581213502, "learning_rate": 2.198109269056804e-06, "loss": 1.418, "step": 11792 }, { "epoch": 0.9025715597734578, "grad_norm": 0.5666288250636557, "learning_rate": 2.19780121717877e-06, "loss": 1.2548, "step": 11793 }, { "epoch": 0.9026480942905251, "grad_norm": 0.587188580830128, "learning_rate": 2.1974933989191956e-06, "loss": 1.4228, "step": 11794 }, { "epoch": 0.9027246288075922, "grad_norm": 0.5542234696343524, "learning_rate": 2.1971858142969946e-06, "loss": 1.2583, "step": 11795 }, { "epoch": 0.9028011633246594, "grad_norm": 2.3480108113037317, "learning_rate": 2.1968784633310654e-06, "loss": 1.8231, "step": 11796 }, { "epoch": 0.9028776978417267, "grad_norm": 0.6082428438996156, "learning_rate": 2.196571346040293e-06, "loss": 1.545, "step": 11797 }, { "epoch": 0.9029542323587938, "grad_norm": 0.5860414716825418, "learning_rate": 2.196264462443546e-06, "loss": 1.2982, "step": 11798 }, { "epoch": 0.903030766875861, "grad_norm": 0.5935299807446387, "learning_rate": 2.195957812559682e-06, "loss": 1.3724, "step": 11799 }, { "epoch": 0.9031073013929282, "grad_norm": 0.6137701163424766, "learning_rate": 2.195651396407542e-06, "loss": 1.3687, "step": 11800 }, { "epoch": 0.9031838359099954, "grad_norm": 0.6962404449898927, "learning_rate": 2.1953452140059532e-06, "loss": 1.5925, "step": 11801 }, { "epoch": 0.9032603704270626, "grad_norm": 0.6255197207104793, "learning_rate": 2.195039265373728e-06, "loss": 1.5859, "step": 11802 }, { "epoch": 0.9033369049441298, "grad_norm": 0.574162847469134, "learning_rate": 2.1947335505296652e-06, "loss": 1.3791, "step": 11803 }, { "epoch": 0.903413439461197, "grad_norm": 0.5457256638365947, "learning_rate": 2.194428069492548e-06, "loss": 1.2066, "step": 11804 }, { "epoch": 0.9034899739782642, "grad_norm": 0.5621899754151742, "learning_rate": 2.194122822281147e-06, "loss": 1.2894, "step": 11805 }, { "epoch": 0.9035665084953314, "grad_norm": 0.8795530891303841, "learning_rate": 2.1938178089142178e-06, "loss": 1.1301, "step": 11806 }, { "epoch": 0.9036430430123986, "grad_norm": 0.5181018739343839, "learning_rate": 2.1935130294105e-06, "loss": 1.1858, "step": 11807 }, { "epoch": 0.9037195775294657, "grad_norm": 0.5179862486980868, "learning_rate": 2.1932084837887215e-06, "loss": 1.1286, "step": 11808 }, { "epoch": 0.903796112046533, "grad_norm": 0.6240323337197606, "learning_rate": 2.192904172067593e-06, "loss": 1.475, "step": 11809 }, { "epoch": 0.9038726465636002, "grad_norm": 0.6102346664079594, "learning_rate": 2.192600094265814e-06, "loss": 1.506, "step": 11810 }, { "epoch": 0.9039491810806674, "grad_norm": 1.0685465281574764, "learning_rate": 2.192296250402067e-06, "loss": 1.1413, "step": 11811 }, { "epoch": 0.9040257155977346, "grad_norm": 0.5261777546889408, "learning_rate": 2.191992640495021e-06, "loss": 1.2805, "step": 11812 }, { "epoch": 0.9041022501148018, "grad_norm": 0.606180088573511, "learning_rate": 2.1916892645633313e-06, "loss": 1.476, "step": 11813 }, { "epoch": 0.904178784631869, "grad_norm": 0.7111702769207323, "learning_rate": 2.1913861226256378e-06, "loss": 1.4468, "step": 11814 }, { "epoch": 0.9042553191489362, "grad_norm": 0.6005067179159365, "learning_rate": 2.191083214700566e-06, "loss": 1.3384, "step": 11815 }, { "epoch": 0.9043318536660033, "grad_norm": 0.9313056697839666, "learning_rate": 2.1907805408067286e-06, "loss": 1.8726, "step": 11816 }, { "epoch": 0.9044083881830706, "grad_norm": 0.5527120384645814, "learning_rate": 2.1904781009627218e-06, "loss": 1.1845, "step": 11817 }, { "epoch": 0.9044849227001378, "grad_norm": 0.6866268909464373, "learning_rate": 2.190175895187129e-06, "loss": 1.3979, "step": 11818 }, { "epoch": 0.9045614572172049, "grad_norm": 0.6271202371378511, "learning_rate": 2.1898739234985186e-06, "loss": 1.1674, "step": 11819 }, { "epoch": 0.9046379917342722, "grad_norm": 0.5344799437196882, "learning_rate": 2.189572185915444e-06, "loss": 1.1601, "step": 11820 }, { "epoch": 0.9047145262513393, "grad_norm": 0.706809971987979, "learning_rate": 2.189270682456446e-06, "loss": 1.4008, "step": 11821 }, { "epoch": 0.9047910607684065, "grad_norm": 0.7813532154854735, "learning_rate": 2.1889694131400485e-06, "loss": 1.9083, "step": 11822 }, { "epoch": 0.9048675952854738, "grad_norm": 2.21714927205071, "learning_rate": 2.1886683779847633e-06, "loss": 1.3634, "step": 11823 }, { "epoch": 0.9049441298025409, "grad_norm": 0.5679929570803456, "learning_rate": 2.1883675770090867e-06, "loss": 1.2506, "step": 11824 }, { "epoch": 0.9050206643196081, "grad_norm": 0.5127670804402987, "learning_rate": 2.1880670102315007e-06, "loss": 1.2076, "step": 11825 }, { "epoch": 0.9050971988366754, "grad_norm": 0.9615193339111727, "learning_rate": 2.187766677670473e-06, "loss": 1.3819, "step": 11826 }, { "epoch": 0.9051737333537425, "grad_norm": 0.5779165091222375, "learning_rate": 2.1874665793444568e-06, "loss": 1.2902, "step": 11827 }, { "epoch": 0.9052502678708098, "grad_norm": 0.5909274003830187, "learning_rate": 2.187166715271892e-06, "loss": 1.4458, "step": 11828 }, { "epoch": 0.9053268023878769, "grad_norm": 0.6948049688883948, "learning_rate": 2.186867085471202e-06, "loss": 1.4877, "step": 11829 }, { "epoch": 0.9054033369049441, "grad_norm": 0.5359722595088287, "learning_rate": 2.1865676899607975e-06, "loss": 1.2039, "step": 11830 }, { "epoch": 0.9054798714220114, "grad_norm": 0.5562535329662404, "learning_rate": 2.1862685287590736e-06, "loss": 1.2068, "step": 11831 }, { "epoch": 0.9055564059390785, "grad_norm": 0.7786201977483405, "learning_rate": 2.1859696018844126e-06, "loss": 1.4079, "step": 11832 }, { "epoch": 0.9056329404561457, "grad_norm": 1.6687726104619842, "learning_rate": 2.1856709093551816e-06, "loss": 1.7478, "step": 11833 }, { "epoch": 0.905709474973213, "grad_norm": 0.653251225489228, "learning_rate": 2.185372451189732e-06, "loss": 1.4512, "step": 11834 }, { "epoch": 0.9057860094902801, "grad_norm": 0.6130999787706393, "learning_rate": 2.1850742274064024e-06, "loss": 1.3775, "step": 11835 }, { "epoch": 0.9058625440073473, "grad_norm": 0.5514240830001588, "learning_rate": 2.1847762380235173e-06, "loss": 1.1492, "step": 11836 }, { "epoch": 0.9059390785244145, "grad_norm": 0.5940711944342911, "learning_rate": 2.184478483059384e-06, "loss": 1.2636, "step": 11837 }, { "epoch": 0.9060156130414817, "grad_norm": 0.5416114718692198, "learning_rate": 2.1841809625323004e-06, "loss": 1.1814, "step": 11838 }, { "epoch": 0.9060921475585489, "grad_norm": 0.5684921042233341, "learning_rate": 2.1838836764605454e-06, "loss": 1.1492, "step": 11839 }, { "epoch": 0.9061686820756161, "grad_norm": 0.4909972834972819, "learning_rate": 2.1835866248623853e-06, "loss": 1.0626, "step": 11840 }, { "epoch": 0.9062452165926833, "grad_norm": 0.5823750163110005, "learning_rate": 2.183289807756072e-06, "loss": 1.1275, "step": 11841 }, { "epoch": 0.9063217511097506, "grad_norm": 0.6249664195922999, "learning_rate": 2.1829932251598427e-06, "loss": 1.3356, "step": 11842 }, { "epoch": 0.9063982856268177, "grad_norm": 0.7034753344445148, "learning_rate": 2.18269687709192e-06, "loss": 1.6998, "step": 11843 }, { "epoch": 0.9064748201438849, "grad_norm": 0.6044694865628785, "learning_rate": 2.182400763570513e-06, "loss": 1.5236, "step": 11844 }, { "epoch": 0.906551354660952, "grad_norm": 0.6667659178242649, "learning_rate": 2.1821048846138153e-06, "loss": 1.2852, "step": 11845 }, { "epoch": 0.9066278891780193, "grad_norm": 0.48088434191873125, "learning_rate": 2.181809240240007e-06, "loss": 1.0181, "step": 11846 }, { "epoch": 0.9067044236950865, "grad_norm": 0.5853987718632789, "learning_rate": 2.181513830467253e-06, "loss": 1.4127, "step": 11847 }, { "epoch": 0.9067809582121537, "grad_norm": 0.5190749379788578, "learning_rate": 2.181218655313704e-06, "loss": 1.1187, "step": 11848 }, { "epoch": 0.9068574927292209, "grad_norm": 0.6136867638936029, "learning_rate": 2.1809237147974976e-06, "loss": 1.5012, "step": 11849 }, { "epoch": 0.906934027246288, "grad_norm": 0.5705656819773687, "learning_rate": 2.1806290089367546e-06, "loss": 1.3115, "step": 11850 }, { "epoch": 0.9070105617633553, "grad_norm": 0.6400074708006736, "learning_rate": 2.1803345377495834e-06, "loss": 1.6152, "step": 11851 }, { "epoch": 0.9070870962804225, "grad_norm": 0.6285221158487304, "learning_rate": 2.1800403012540763e-06, "loss": 1.4283, "step": 11852 }, { "epoch": 0.9071636307974896, "grad_norm": 0.6831224554326979, "learning_rate": 2.1797462994683126e-06, "loss": 1.4246, "step": 11853 }, { "epoch": 0.9072401653145569, "grad_norm": 0.5211521239421437, "learning_rate": 2.1794525324103568e-06, "loss": 1.2618, "step": 11854 }, { "epoch": 0.9073166998316241, "grad_norm": 0.6270853142678046, "learning_rate": 2.1791590000982585e-06, "loss": 1.4581, "step": 11855 }, { "epoch": 0.9073932343486912, "grad_norm": 0.8460307615236571, "learning_rate": 2.1788657025500538e-06, "loss": 1.4812, "step": 11856 }, { "epoch": 0.9074697688657585, "grad_norm": 0.5179251225864394, "learning_rate": 2.1785726397837636e-06, "loss": 1.0554, "step": 11857 }, { "epoch": 0.9075463033828256, "grad_norm": 0.5479526353122457, "learning_rate": 2.1782798118173937e-06, "loss": 1.1891, "step": 11858 }, { "epoch": 0.9076228378998928, "grad_norm": 0.5299144850456063, "learning_rate": 2.1779872186689364e-06, "loss": 1.156, "step": 11859 }, { "epoch": 0.9076993724169601, "grad_norm": 0.5392155950095643, "learning_rate": 2.1776948603563712e-06, "loss": 1.2327, "step": 11860 }, { "epoch": 0.9077759069340272, "grad_norm": 0.6070704067164412, "learning_rate": 2.1774027368976596e-06, "loss": 1.4762, "step": 11861 }, { "epoch": 0.9078524414510944, "grad_norm": 0.5648470501767552, "learning_rate": 2.1771108483107517e-06, "loss": 1.3791, "step": 11862 }, { "epoch": 0.9079289759681617, "grad_norm": 0.5130889236211149, "learning_rate": 2.1768191946135814e-06, "loss": 1.1975, "step": 11863 }, { "epoch": 0.9080055104852288, "grad_norm": 0.5524628681807259, "learning_rate": 2.176527775824069e-06, "loss": 1.1496, "step": 11864 }, { "epoch": 0.9080820450022961, "grad_norm": 0.5673434686139782, "learning_rate": 2.17623659196012e-06, "loss": 1.39, "step": 11865 }, { "epoch": 0.9081585795193632, "grad_norm": 0.6269179102772177, "learning_rate": 2.1759456430396258e-06, "loss": 1.5256, "step": 11866 }, { "epoch": 0.9082351140364304, "grad_norm": 1.5708222170793646, "learning_rate": 2.1756549290804627e-06, "loss": 1.7135, "step": 11867 }, { "epoch": 0.9083116485534977, "grad_norm": 0.5483288059592745, "learning_rate": 2.1753644501004932e-06, "loss": 1.3956, "step": 11868 }, { "epoch": 0.9083881830705648, "grad_norm": 0.6124370635499853, "learning_rate": 2.175074206117566e-06, "loss": 1.7232, "step": 11869 }, { "epoch": 0.908464717587632, "grad_norm": 0.6712141260577471, "learning_rate": 2.174784197149513e-06, "loss": 1.4744, "step": 11870 }, { "epoch": 0.9085412521046993, "grad_norm": 0.894807261326486, "learning_rate": 2.174494423214155e-06, "loss": 1.4593, "step": 11871 }, { "epoch": 0.9086177866217664, "grad_norm": 0.6223291402606951, "learning_rate": 2.1742048843292953e-06, "loss": 1.5185, "step": 11872 }, { "epoch": 0.9086943211388336, "grad_norm": 0.6034976377900618, "learning_rate": 2.1739155805127245e-06, "loss": 1.2932, "step": 11873 }, { "epoch": 0.9087708556559008, "grad_norm": 0.9413421272496939, "learning_rate": 2.1736265117822184e-06, "loss": 1.7213, "step": 11874 }, { "epoch": 0.908847390172968, "grad_norm": 0.5395347257474191, "learning_rate": 2.173337678155538e-06, "loss": 1.1901, "step": 11875 }, { "epoch": 0.9089239246900352, "grad_norm": 0.9459244678868446, "learning_rate": 2.17304907965043e-06, "loss": 1.5471, "step": 11876 }, { "epoch": 0.9090004592071024, "grad_norm": 0.9720757500870257, "learning_rate": 2.172760716284627e-06, "loss": 1.3312, "step": 11877 }, { "epoch": 0.9090769937241696, "grad_norm": 0.5797852123931749, "learning_rate": 2.1724725880758476e-06, "loss": 1.2772, "step": 11878 }, { "epoch": 0.9091535282412369, "grad_norm": 0.568822034974707, "learning_rate": 2.172184695041794e-06, "loss": 1.4269, "step": 11879 }, { "epoch": 0.909230062758304, "grad_norm": 0.7105041552055082, "learning_rate": 2.1718970372001556e-06, "loss": 1.4047, "step": 11880 }, { "epoch": 0.9093065972753712, "grad_norm": 1.4848178862342805, "learning_rate": 2.171609614568607e-06, "loss": 1.5533, "step": 11881 }, { "epoch": 0.9093831317924383, "grad_norm": 0.6552995727927948, "learning_rate": 2.1713224271648085e-06, "loss": 1.1661, "step": 11882 }, { "epoch": 0.9094596663095056, "grad_norm": 0.49557967871936015, "learning_rate": 2.1710354750064054e-06, "loss": 1.0257, "step": 11883 }, { "epoch": 0.9095362008265728, "grad_norm": 0.5644195580849064, "learning_rate": 2.1707487581110294e-06, "loss": 1.3175, "step": 11884 }, { "epoch": 0.90961273534364, "grad_norm": 0.5449630666585099, "learning_rate": 2.1704622764962975e-06, "loss": 1.3013, "step": 11885 }, { "epoch": 0.9096892698607072, "grad_norm": 0.6340453330689133, "learning_rate": 2.1701760301798108e-06, "loss": 1.4092, "step": 11886 }, { "epoch": 0.9097658043777743, "grad_norm": 0.5801197914071606, "learning_rate": 2.169890019179158e-06, "loss": 1.3092, "step": 11887 }, { "epoch": 0.9098423388948416, "grad_norm": 0.5858950611199405, "learning_rate": 2.1696042435119126e-06, "loss": 1.3192, "step": 11888 }, { "epoch": 0.9099188734119088, "grad_norm": 0.6431113471448395, "learning_rate": 2.1693187031956327e-06, "loss": 1.3614, "step": 11889 }, { "epoch": 0.9099954079289759, "grad_norm": 0.6534846932805831, "learning_rate": 2.1690333982478636e-06, "loss": 1.701, "step": 11890 }, { "epoch": 0.9100719424460432, "grad_norm": 0.5141537689406769, "learning_rate": 2.168748328686135e-06, "loss": 1.2038, "step": 11891 }, { "epoch": 0.9101484769631104, "grad_norm": 0.9314047897680412, "learning_rate": 2.168463494527961e-06, "loss": 1.3673, "step": 11892 }, { "epoch": 0.9102250114801775, "grad_norm": 0.5970057205009103, "learning_rate": 2.1681788957908455e-06, "loss": 1.4412, "step": 11893 }, { "epoch": 0.9103015459972448, "grad_norm": 0.6109953966345267, "learning_rate": 2.1678945324922736e-06, "loss": 1.4006, "step": 11894 }, { "epoch": 0.9103780805143119, "grad_norm": 0.5497891946234946, "learning_rate": 2.167610404649717e-06, "loss": 1.2507, "step": 11895 }, { "epoch": 0.9104546150313791, "grad_norm": 0.7238580876970154, "learning_rate": 2.1673265122806335e-06, "loss": 1.4558, "step": 11896 }, { "epoch": 0.9105311495484464, "grad_norm": 0.5253846545460683, "learning_rate": 2.1670428554024667e-06, "loss": 1.179, "step": 11897 }, { "epoch": 0.9106076840655135, "grad_norm": 0.5498224783978576, "learning_rate": 2.1667594340326453e-06, "loss": 1.2136, "step": 11898 }, { "epoch": 0.9106842185825808, "grad_norm": 0.5705884292559813, "learning_rate": 2.1664762481885836e-06, "loss": 1.2815, "step": 11899 }, { "epoch": 0.910760753099648, "grad_norm": 0.5432301695608797, "learning_rate": 2.1661932978876815e-06, "loss": 1.2947, "step": 11900 }, { "epoch": 0.9108372876167151, "grad_norm": 0.6444221508651092, "learning_rate": 2.165910583147324e-06, "loss": 1.5063, "step": 11901 }, { "epoch": 0.9109138221337824, "grad_norm": 0.6401827333570186, "learning_rate": 2.165628103984882e-06, "loss": 1.7314, "step": 11902 }, { "epoch": 0.9109903566508495, "grad_norm": 0.5475327325286863, "learning_rate": 2.165345860417711e-06, "loss": 1.2475, "step": 11903 }, { "epoch": 0.9110668911679167, "grad_norm": 0.5386391313658555, "learning_rate": 2.1650638524631546e-06, "loss": 1.0518, "step": 11904 }, { "epoch": 0.911143425684984, "grad_norm": 0.4967105737996188, "learning_rate": 2.1647820801385397e-06, "loss": 1.1421, "step": 11905 }, { "epoch": 0.9112199602020511, "grad_norm": 0.7068153023594779, "learning_rate": 2.1645005434611783e-06, "loss": 1.6492, "step": 11906 }, { "epoch": 0.9112964947191183, "grad_norm": 0.5752058544876105, "learning_rate": 2.1642192424483698e-06, "loss": 1.0617, "step": 11907 }, { "epoch": 0.9113730292361856, "grad_norm": 0.6668007755871759, "learning_rate": 2.1639381771173975e-06, "loss": 1.4371, "step": 11908 }, { "epoch": 0.9114495637532527, "grad_norm": 0.5418182004441444, "learning_rate": 2.1636573474855317e-06, "loss": 1.1211, "step": 11909 }, { "epoch": 0.9115260982703199, "grad_norm": 0.5857441572375479, "learning_rate": 2.1633767535700267e-06, "loss": 1.3044, "step": 11910 }, { "epoch": 0.9116026327873871, "grad_norm": 0.6506659256061582, "learning_rate": 2.1630963953881235e-06, "loss": 1.529, "step": 11911 }, { "epoch": 0.9116791673044543, "grad_norm": 0.6311537932405573, "learning_rate": 2.1628162729570477e-06, "loss": 1.5711, "step": 11912 }, { "epoch": 0.9117557018215215, "grad_norm": 0.5572107629834024, "learning_rate": 2.162536386294011e-06, "loss": 1.4396, "step": 11913 }, { "epoch": 0.9118322363385887, "grad_norm": 0.5100882443334306, "learning_rate": 2.1622567354162107e-06, "loss": 0.9936, "step": 11914 }, { "epoch": 0.9119087708556559, "grad_norm": 0.46290652794082277, "learning_rate": 2.1619773203408292e-06, "loss": 1.057, "step": 11915 }, { "epoch": 0.911985305372723, "grad_norm": 0.5904613443708783, "learning_rate": 2.161698141085035e-06, "loss": 1.3308, "step": 11916 }, { "epoch": 0.9120618398897903, "grad_norm": 0.541212564985319, "learning_rate": 2.1614191976659808e-06, "loss": 1.2783, "step": 11917 }, { "epoch": 0.9121383744068575, "grad_norm": 0.5435376786657435, "learning_rate": 2.1611404901008067e-06, "loss": 1.1882, "step": 11918 }, { "epoch": 0.9122149089239246, "grad_norm": 0.5433830534754553, "learning_rate": 2.1608620184066364e-06, "loss": 1.2729, "step": 11919 }, { "epoch": 0.9122914434409919, "grad_norm": 0.7669658380690956, "learning_rate": 2.1605837826005806e-06, "loss": 1.297, "step": 11920 }, { "epoch": 0.9123679779580591, "grad_norm": 0.5687123212857269, "learning_rate": 2.160305782699735e-06, "loss": 1.1919, "step": 11921 }, { "epoch": 0.9124445124751263, "grad_norm": 0.5809591575082802, "learning_rate": 2.1600280187211806e-06, "loss": 1.3492, "step": 11922 }, { "epoch": 0.9125210469921935, "grad_norm": 0.5644191403474973, "learning_rate": 2.159750490681984e-06, "loss": 1.2759, "step": 11923 }, { "epoch": 0.9125975815092606, "grad_norm": 0.6868732071337227, "learning_rate": 2.159473198599198e-06, "loss": 1.495, "step": 11924 }, { "epoch": 0.9126741160263279, "grad_norm": 0.4733960525745392, "learning_rate": 2.1591961424898584e-06, "loss": 0.9347, "step": 11925 }, { "epoch": 0.9127506505433951, "grad_norm": 0.49811416439150125, "learning_rate": 2.158919322370991e-06, "loss": 1.0251, "step": 11926 }, { "epoch": 0.9128271850604622, "grad_norm": 0.5803794136667704, "learning_rate": 2.1586427382596022e-06, "loss": 1.3991, "step": 11927 }, { "epoch": 0.9129037195775295, "grad_norm": 0.8939605127907733, "learning_rate": 2.1583663901726878e-06, "loss": 1.6527, "step": 11928 }, { "epoch": 0.9129802540945967, "grad_norm": 0.6438682967569701, "learning_rate": 2.1580902781272263e-06, "loss": 1.6503, "step": 11929 }, { "epoch": 0.9130567886116638, "grad_norm": 0.6162668464877366, "learning_rate": 2.1578144021401825e-06, "loss": 1.3343, "step": 11930 }, { "epoch": 0.9131333231287311, "grad_norm": 4.503725840140087, "learning_rate": 2.1575387622285086e-06, "loss": 1.8283, "step": 11931 }, { "epoch": 0.9132098576457982, "grad_norm": 0.5990721815302363, "learning_rate": 2.1572633584091406e-06, "loss": 1.2701, "step": 11932 }, { "epoch": 0.9132863921628654, "grad_norm": 0.6401171409201744, "learning_rate": 2.1569881906989983e-06, "loss": 1.2447, "step": 11933 }, { "epoch": 0.9133629266799327, "grad_norm": 1.0277296549326425, "learning_rate": 2.1567132591149906e-06, "loss": 1.433, "step": 11934 }, { "epoch": 0.9134394611969998, "grad_norm": 0.7002559762279277, "learning_rate": 2.156438563674009e-06, "loss": 1.5958, "step": 11935 }, { "epoch": 0.913515995714067, "grad_norm": 1.1891249070163736, "learning_rate": 2.1561641043929322e-06, "loss": 1.5312, "step": 11936 }, { "epoch": 0.9135925302311343, "grad_norm": 0.6639457565950877, "learning_rate": 2.155889881288624e-06, "loss": 1.494, "step": 11937 }, { "epoch": 0.9136690647482014, "grad_norm": 0.7045922985445887, "learning_rate": 2.155615894377933e-06, "loss": 1.3096, "step": 11938 }, { "epoch": 0.9137455992652687, "grad_norm": 0.6503516967776548, "learning_rate": 2.1553421436776937e-06, "loss": 1.6251, "step": 11939 }, { "epoch": 0.9138221337823358, "grad_norm": 0.5356885824367327, "learning_rate": 2.155068629204727e-06, "loss": 1.1696, "step": 11940 }, { "epoch": 0.913898668299403, "grad_norm": 0.5131683255216126, "learning_rate": 2.1547953509758374e-06, "loss": 1.0923, "step": 11941 }, { "epoch": 0.9139752028164703, "grad_norm": 0.4977679841780326, "learning_rate": 2.1545223090078166e-06, "loss": 0.9997, "step": 11942 }, { "epoch": 0.9140517373335374, "grad_norm": 0.5803789838302174, "learning_rate": 2.1542495033174414e-06, "loss": 1.4163, "step": 11943 }, { "epoch": 0.9141282718506046, "grad_norm": 0.6710081124915587, "learning_rate": 2.153976933921473e-06, "loss": 1.3229, "step": 11944 }, { "epoch": 0.9142048063676719, "grad_norm": 0.6648738066199618, "learning_rate": 2.1537046008366585e-06, "loss": 1.6842, "step": 11945 }, { "epoch": 0.914281340884739, "grad_norm": 0.6172598920828898, "learning_rate": 2.153432504079732e-06, "loss": 1.3662, "step": 11946 }, { "epoch": 0.9143578754018062, "grad_norm": 0.562689547087083, "learning_rate": 2.1531606436674116e-06, "loss": 1.1934, "step": 11947 }, { "epoch": 0.9144344099188734, "grad_norm": 0.5521188002553791, "learning_rate": 2.152889019616401e-06, "loss": 1.1471, "step": 11948 }, { "epoch": 0.9145109444359406, "grad_norm": 0.6209373772713561, "learning_rate": 2.1526176319433898e-06, "loss": 1.3189, "step": 11949 }, { "epoch": 0.9145874789530078, "grad_norm": 0.5423630711084381, "learning_rate": 2.152346480665053e-06, "loss": 1.202, "step": 11950 }, { "epoch": 0.914664013470075, "grad_norm": 0.5855576980387412, "learning_rate": 2.1520755657980506e-06, "loss": 1.3185, "step": 11951 }, { "epoch": 0.9147405479871422, "grad_norm": 0.745580637926328, "learning_rate": 2.1518048873590284e-06, "loss": 1.1976, "step": 11952 }, { "epoch": 0.9148170825042093, "grad_norm": 0.8216500658766275, "learning_rate": 2.151534445364618e-06, "loss": 1.5203, "step": 11953 }, { "epoch": 0.9148936170212766, "grad_norm": 0.5866944095521124, "learning_rate": 2.151264239831436e-06, "loss": 1.0731, "step": 11954 }, { "epoch": 0.9149701515383438, "grad_norm": 0.5803271216037919, "learning_rate": 2.1509942707760846e-06, "loss": 1.2947, "step": 11955 }, { "epoch": 0.915046686055411, "grad_norm": 0.6527693400952201, "learning_rate": 2.1507245382151513e-06, "loss": 1.5646, "step": 11956 }, { "epoch": 0.9151232205724782, "grad_norm": 0.6166253553879683, "learning_rate": 2.15045504216521e-06, "loss": 1.3344, "step": 11957 }, { "epoch": 0.9151997550895454, "grad_norm": 0.7586075212408817, "learning_rate": 2.1501857826428184e-06, "loss": 1.4283, "step": 11958 }, { "epoch": 0.9152762896066126, "grad_norm": 0.7307479808995961, "learning_rate": 2.1499167596645213e-06, "loss": 1.6627, "step": 11959 }, { "epoch": 0.9153528241236798, "grad_norm": 0.5964968802181564, "learning_rate": 2.149647973246848e-06, "loss": 1.2889, "step": 11960 }, { "epoch": 0.9154293586407469, "grad_norm": 0.5144355953017795, "learning_rate": 2.1493794234063138e-06, "loss": 1.1238, "step": 11961 }, { "epoch": 0.9155058931578142, "grad_norm": 0.5881070897547287, "learning_rate": 2.149111110159419e-06, "loss": 1.3869, "step": 11962 }, { "epoch": 0.9155824276748814, "grad_norm": 0.5458124667988349, "learning_rate": 2.1488430335226483e-06, "loss": 1.179, "step": 11963 }, { "epoch": 0.9156589621919485, "grad_norm": 1.2700985410151324, "learning_rate": 2.148575193512476e-06, "loss": 1.6459, "step": 11964 }, { "epoch": 0.9157354967090158, "grad_norm": 0.5529199397255437, "learning_rate": 2.1483075901453564e-06, "loss": 1.1712, "step": 11965 }, { "epoch": 0.915812031226083, "grad_norm": 0.6456107594940788, "learning_rate": 2.148040223437734e-06, "loss": 1.5787, "step": 11966 }, { "epoch": 0.9158885657431501, "grad_norm": 0.547077108723805, "learning_rate": 2.1477730934060347e-06, "loss": 1.2099, "step": 11967 }, { "epoch": 0.9159651002602174, "grad_norm": 0.589115348119674, "learning_rate": 2.1475062000666715e-06, "loss": 1.3567, "step": 11968 }, { "epoch": 0.9160416347772845, "grad_norm": 0.589690870527357, "learning_rate": 2.147239543436045e-06, "loss": 1.3319, "step": 11969 }, { "epoch": 0.9161181692943517, "grad_norm": 0.551481401298428, "learning_rate": 2.1469731235305386e-06, "loss": 1.3711, "step": 11970 }, { "epoch": 0.916194703811419, "grad_norm": 0.6011252899952575, "learning_rate": 2.146706940366522e-06, "loss": 1.4279, "step": 11971 }, { "epoch": 0.9162712383284861, "grad_norm": 0.5647494173816453, "learning_rate": 2.14644099396035e-06, "loss": 1.4285, "step": 11972 }, { "epoch": 0.9163477728455534, "grad_norm": 1.1662979695106412, "learning_rate": 2.1461752843283633e-06, "loss": 1.4125, "step": 11973 }, { "epoch": 0.9164243073626206, "grad_norm": 0.6394532194814594, "learning_rate": 2.1459098114868874e-06, "loss": 1.4961, "step": 11974 }, { "epoch": 0.9165008418796877, "grad_norm": 0.594672929558799, "learning_rate": 2.145644575452235e-06, "loss": 1.3243, "step": 11975 }, { "epoch": 0.916577376396755, "grad_norm": 0.585765881322926, "learning_rate": 2.145379576240702e-06, "loss": 1.2431, "step": 11976 }, { "epoch": 0.9166539109138221, "grad_norm": 0.6994817131876041, "learning_rate": 2.1451148138685707e-06, "loss": 1.4493, "step": 11977 }, { "epoch": 0.9167304454308893, "grad_norm": 0.5757208323574096, "learning_rate": 2.144850288352109e-06, "loss": 1.3321, "step": 11978 }, { "epoch": 0.9168069799479566, "grad_norm": 0.561175300283466, "learning_rate": 2.144585999707571e-06, "loss": 1.3561, "step": 11979 }, { "epoch": 0.9168835144650237, "grad_norm": 0.6824827138834498, "learning_rate": 2.1443219479511943e-06, "loss": 1.8597, "step": 11980 }, { "epoch": 0.9169600489820909, "grad_norm": 0.6517970329358829, "learning_rate": 2.1440581330992037e-06, "loss": 1.6097, "step": 11981 }, { "epoch": 0.9170365834991582, "grad_norm": 0.49624158317216466, "learning_rate": 2.143794555167808e-06, "loss": 1.1571, "step": 11982 }, { "epoch": 0.9171131180162253, "grad_norm": 0.8967168737701015, "learning_rate": 2.1435312141732024e-06, "loss": 1.7001, "step": 11983 }, { "epoch": 0.9171896525332925, "grad_norm": 0.6633661114971067, "learning_rate": 2.1432681101315684e-06, "loss": 1.7045, "step": 11984 }, { "epoch": 0.9172661870503597, "grad_norm": 0.6504513706800984, "learning_rate": 2.1430052430590705e-06, "loss": 1.6478, "step": 11985 }, { "epoch": 0.9173427215674269, "grad_norm": 0.5806182069761077, "learning_rate": 2.142742612971861e-06, "loss": 1.3729, "step": 11986 }, { "epoch": 0.9174192560844942, "grad_norm": 0.5951006447158245, "learning_rate": 2.1424802198860762e-06, "loss": 1.4277, "step": 11987 }, { "epoch": 0.9174957906015613, "grad_norm": 0.5059806017610544, "learning_rate": 2.1422180638178387e-06, "loss": 1.1177, "step": 11988 }, { "epoch": 0.9175723251186285, "grad_norm": 0.48439138809013493, "learning_rate": 2.141956144783256e-06, "loss": 0.9473, "step": 11989 }, { "epoch": 0.9176488596356956, "grad_norm": 0.6742567610224897, "learning_rate": 2.1416944627984207e-06, "loss": 1.4392, "step": 11990 }, { "epoch": 0.9177253941527629, "grad_norm": 0.6862514792168255, "learning_rate": 2.141433017879412e-06, "loss": 1.5282, "step": 11991 }, { "epoch": 0.9178019286698301, "grad_norm": 0.5004871059693818, "learning_rate": 2.1411718100422936e-06, "loss": 1.086, "step": 11992 }, { "epoch": 0.9178784631868973, "grad_norm": 0.5811168541387556, "learning_rate": 2.140910839303115e-06, "loss": 1.4349, "step": 11993 }, { "epoch": 0.9179549977039645, "grad_norm": 0.6031700076669122, "learning_rate": 2.140650105677911e-06, "loss": 1.3465, "step": 11994 }, { "epoch": 0.9180315322210317, "grad_norm": 0.5693019171546897, "learning_rate": 2.1403896091827014e-06, "loss": 1.2961, "step": 11995 }, { "epoch": 0.9181080667380989, "grad_norm": 0.5963504112430156, "learning_rate": 2.1401293498334924e-06, "loss": 1.3218, "step": 11996 }, { "epoch": 0.9181846012551661, "grad_norm": 0.6888371296207462, "learning_rate": 2.139869327646275e-06, "loss": 1.7035, "step": 11997 }, { "epoch": 0.9182611357722332, "grad_norm": 0.5535575021621131, "learning_rate": 2.1396095426370254e-06, "loss": 1.1579, "step": 11998 }, { "epoch": 0.9183376702893005, "grad_norm": 0.5532031148416484, "learning_rate": 2.1393499948217065e-06, "loss": 1.3334, "step": 11999 }, { "epoch": 0.9184142048063677, "grad_norm": 0.6366154721078775, "learning_rate": 2.139090684216265e-06, "loss": 1.2679, "step": 12000 }, { "epoch": 0.9184907393234348, "grad_norm": 0.6475722299894631, "learning_rate": 2.1388316108366334e-06, "loss": 1.4724, "step": 12001 }, { "epoch": 0.9185672738405021, "grad_norm": 0.5328405254852757, "learning_rate": 2.1385727746987305e-06, "loss": 1.082, "step": 12002 }, { "epoch": 0.9186438083575693, "grad_norm": 0.6111830979781909, "learning_rate": 2.13831417581846e-06, "loss": 1.2886, "step": 12003 }, { "epoch": 0.9187203428746364, "grad_norm": 0.6445024137496533, "learning_rate": 2.1380558142117107e-06, "loss": 1.6438, "step": 12004 }, { "epoch": 0.9187968773917037, "grad_norm": 0.5147685994897825, "learning_rate": 2.1377976898943568e-06, "loss": 1.0351, "step": 12005 }, { "epoch": 0.9188734119087708, "grad_norm": 0.7645853283331604, "learning_rate": 2.137539802882259e-06, "loss": 1.682, "step": 12006 }, { "epoch": 0.918949946425838, "grad_norm": 0.5846405485888296, "learning_rate": 2.137282153191262e-06, "loss": 1.2939, "step": 12007 }, { "epoch": 0.9190264809429053, "grad_norm": 0.5356076017266111, "learning_rate": 2.1370247408371968e-06, "loss": 1.0059, "step": 12008 }, { "epoch": 0.9191030154599724, "grad_norm": 0.6611444025399432, "learning_rate": 2.1367675658358796e-06, "loss": 1.1143, "step": 12009 }, { "epoch": 0.9191795499770397, "grad_norm": 0.574675886027034, "learning_rate": 2.136510628203112e-06, "loss": 1.3786, "step": 12010 }, { "epoch": 0.9192560844941069, "grad_norm": 0.6257902663751741, "learning_rate": 2.1362539279546816e-06, "loss": 1.5981, "step": 12011 }, { "epoch": 0.919332619011174, "grad_norm": 0.8307769093669803, "learning_rate": 2.1359974651063594e-06, "loss": 1.3244, "step": 12012 }, { "epoch": 0.9194091535282413, "grad_norm": 0.719644854086383, "learning_rate": 2.135741239673905e-06, "loss": 1.4173, "step": 12013 }, { "epoch": 0.9194856880453084, "grad_norm": 0.5784904503124401, "learning_rate": 2.1354852516730605e-06, "loss": 1.2833, "step": 12014 }, { "epoch": 0.9195622225623756, "grad_norm": 0.6281869155750182, "learning_rate": 2.1352295011195546e-06, "loss": 1.5601, "step": 12015 }, { "epoch": 0.9196387570794429, "grad_norm": 0.7386485617353119, "learning_rate": 2.134973988029102e-06, "loss": 1.542, "step": 12016 }, { "epoch": 0.91971529159651, "grad_norm": 0.5469825294449409, "learning_rate": 2.1347187124174015e-06, "loss": 1.3765, "step": 12017 }, { "epoch": 0.9197918261135772, "grad_norm": 0.6716629166849372, "learning_rate": 2.1344636743001374e-06, "loss": 1.6997, "step": 12018 }, { "epoch": 0.9198683606306444, "grad_norm": 0.5361461003022872, "learning_rate": 2.134208873692982e-06, "loss": 1.223, "step": 12019 }, { "epoch": 0.9199448951477116, "grad_norm": 0.6275817776759324, "learning_rate": 2.13395431061159e-06, "loss": 1.4897, "step": 12020 }, { "epoch": 0.9200214296647788, "grad_norm": 3.012934004291611, "learning_rate": 2.133699985071602e-06, "loss": 1.7053, "step": 12021 }, { "epoch": 0.920097964181846, "grad_norm": 0.6073749078789575, "learning_rate": 2.133445897088645e-06, "loss": 1.4573, "step": 12022 }, { "epoch": 0.9201744986989132, "grad_norm": 0.7140507957958091, "learning_rate": 2.1331920466783314e-06, "loss": 1.7168, "step": 12023 }, { "epoch": 0.9202510332159805, "grad_norm": 0.7051547853243196, "learning_rate": 2.132938433856258e-06, "loss": 1.8828, "step": 12024 }, { "epoch": 0.9203275677330476, "grad_norm": 0.5569027988178141, "learning_rate": 2.132685058638007e-06, "loss": 1.3191, "step": 12025 }, { "epoch": 0.9204041022501148, "grad_norm": 0.5643858676613943, "learning_rate": 2.132431921039147e-06, "loss": 1.2981, "step": 12026 }, { "epoch": 0.920480636767182, "grad_norm": 1.426963896250995, "learning_rate": 2.1321790210752323e-06, "loss": 1.3606, "step": 12027 }, { "epoch": 0.9205571712842492, "grad_norm": 0.5916301125582483, "learning_rate": 2.1319263587618007e-06, "loss": 1.3769, "step": 12028 }, { "epoch": 0.9206337058013164, "grad_norm": 0.5948394438866598, "learning_rate": 2.1316739341143765e-06, "loss": 1.2602, "step": 12029 }, { "epoch": 0.9207102403183836, "grad_norm": 0.5514897137026735, "learning_rate": 2.1314217471484703e-06, "loss": 1.0769, "step": 12030 }, { "epoch": 0.9207867748354508, "grad_norm": 0.6956424492124775, "learning_rate": 2.131169797879577e-06, "loss": 1.2813, "step": 12031 }, { "epoch": 0.920863309352518, "grad_norm": 0.8640251952336492, "learning_rate": 2.1309180863231765e-06, "loss": 1.4159, "step": 12032 }, { "epoch": 0.9209398438695852, "grad_norm": 0.981788598507757, "learning_rate": 2.1306666124947356e-06, "loss": 1.1604, "step": 12033 }, { "epoch": 0.9210163783866524, "grad_norm": 0.6923975566919626, "learning_rate": 2.130415376409704e-06, "loss": 1.6559, "step": 12034 }, { "epoch": 0.9210929129037195, "grad_norm": 0.7291527059473704, "learning_rate": 2.1301643780835205e-06, "loss": 1.5723, "step": 12035 }, { "epoch": 0.9211694474207868, "grad_norm": 0.8296400571558066, "learning_rate": 2.1299136175316066e-06, "loss": 1.689, "step": 12036 }, { "epoch": 0.921245981937854, "grad_norm": 0.6486430305362766, "learning_rate": 2.129663094769369e-06, "loss": 1.5896, "step": 12037 }, { "epoch": 0.9213225164549211, "grad_norm": 0.6302304626053883, "learning_rate": 2.129412809812201e-06, "loss": 1.4846, "step": 12038 }, { "epoch": 0.9213990509719884, "grad_norm": 0.5573212733290411, "learning_rate": 2.1291627626754805e-06, "loss": 0.9553, "step": 12039 }, { "epoch": 0.9214755854890556, "grad_norm": 0.9252370257689135, "learning_rate": 2.1289129533745715e-06, "loss": 1.6619, "step": 12040 }, { "epoch": 0.9215521200061227, "grad_norm": 0.7047035928058802, "learning_rate": 2.1286633819248233e-06, "loss": 1.5072, "step": 12041 }, { "epoch": 0.92162865452319, "grad_norm": 0.5669746734228631, "learning_rate": 2.12841404834157e-06, "loss": 1.3781, "step": 12042 }, { "epoch": 0.9217051890402571, "grad_norm": 0.7405488785493965, "learning_rate": 2.1281649526401313e-06, "loss": 1.6443, "step": 12043 }, { "epoch": 0.9217817235573244, "grad_norm": 1.1478829078796307, "learning_rate": 2.1279160948358124e-06, "loss": 1.551, "step": 12044 }, { "epoch": 0.9218582580743916, "grad_norm": 0.5397717589799876, "learning_rate": 2.127667474943904e-06, "loss": 1.2138, "step": 12045 }, { "epoch": 0.9219347925914587, "grad_norm": 0.6831271867098435, "learning_rate": 2.127419092979682e-06, "loss": 1.7089, "step": 12046 }, { "epoch": 0.922011327108526, "grad_norm": 0.5070822786117318, "learning_rate": 2.127170948958408e-06, "loss": 1.066, "step": 12047 }, { "epoch": 0.9220878616255932, "grad_norm": 1.255229358887073, "learning_rate": 2.1269230428953285e-06, "loss": 1.6564, "step": 12048 }, { "epoch": 0.9221643961426603, "grad_norm": 0.6469844159915862, "learning_rate": 2.126675374805675e-06, "loss": 1.4519, "step": 12049 }, { "epoch": 0.9222409306597276, "grad_norm": 0.6547788654151475, "learning_rate": 2.1264279447046656e-06, "loss": 1.6795, "step": 12050 }, { "epoch": 0.9223174651767947, "grad_norm": 0.504591004851214, "learning_rate": 2.126180752607503e-06, "loss": 1.1635, "step": 12051 }, { "epoch": 0.9223939996938619, "grad_norm": 0.5456403121144139, "learning_rate": 2.1259337985293755e-06, "loss": 1.3415, "step": 12052 }, { "epoch": 0.9224705342109292, "grad_norm": 0.5087239862255262, "learning_rate": 2.1256870824854572e-06, "loss": 1.0775, "step": 12053 }, { "epoch": 0.9225470687279963, "grad_norm": 0.620115376684061, "learning_rate": 2.1254406044909063e-06, "loss": 1.5078, "step": 12054 }, { "epoch": 0.9226236032450635, "grad_norm": 0.5837093491805923, "learning_rate": 2.1251943645608674e-06, "loss": 1.2871, "step": 12055 }, { "epoch": 0.9227001377621307, "grad_norm": 0.519168181134821, "learning_rate": 2.1249483627104693e-06, "loss": 1.0432, "step": 12056 }, { "epoch": 0.9227766722791979, "grad_norm": 0.5709622871711262, "learning_rate": 2.1247025989548294e-06, "loss": 1.2825, "step": 12057 }, { "epoch": 0.9228532067962651, "grad_norm": 0.6186354720826863, "learning_rate": 2.124457073309046e-06, "loss": 1.4463, "step": 12058 }, { "epoch": 0.9229297413133323, "grad_norm": 0.600601541941483, "learning_rate": 2.124211785788206e-06, "loss": 1.2343, "step": 12059 }, { "epoch": 0.9230062758303995, "grad_norm": 0.5508886530464019, "learning_rate": 2.12396673640738e-06, "loss": 1.3291, "step": 12060 }, { "epoch": 0.9230828103474668, "grad_norm": 0.6038261532779597, "learning_rate": 2.1237219251816254e-06, "loss": 1.5305, "step": 12061 }, { "epoch": 0.9231593448645339, "grad_norm": 0.589797938575114, "learning_rate": 2.1234773521259823e-06, "loss": 1.4875, "step": 12062 }, { "epoch": 0.9232358793816011, "grad_norm": 0.7335374623399098, "learning_rate": 2.1232330172554805e-06, "loss": 1.3075, "step": 12063 }, { "epoch": 0.9233124138986682, "grad_norm": 0.515761120174319, "learning_rate": 2.122988920585131e-06, "loss": 1.092, "step": 12064 }, { "epoch": 0.9233889484157355, "grad_norm": 0.5517412520939798, "learning_rate": 2.1227450621299325e-06, "loss": 1.3195, "step": 12065 }, { "epoch": 0.9234654829328027, "grad_norm": 0.5805231399485372, "learning_rate": 2.122501441904868e-06, "loss": 1.2903, "step": 12066 }, { "epoch": 0.9235420174498699, "grad_norm": 0.7016594271175867, "learning_rate": 2.1222580599249066e-06, "loss": 1.7243, "step": 12067 }, { "epoch": 0.9236185519669371, "grad_norm": 0.6259247685910112, "learning_rate": 2.122014916205002e-06, "loss": 1.5394, "step": 12068 }, { "epoch": 0.9236950864840043, "grad_norm": 0.7850014132753206, "learning_rate": 2.1217720107600944e-06, "loss": 1.8713, "step": 12069 }, { "epoch": 0.9237716210010715, "grad_norm": 0.6595346350816644, "learning_rate": 2.121529343605108e-06, "loss": 1.5166, "step": 12070 }, { "epoch": 0.9238481555181387, "grad_norm": 0.5785499756086107, "learning_rate": 2.121286914754953e-06, "loss": 1.3788, "step": 12071 }, { "epoch": 0.9239246900352058, "grad_norm": 0.6648930926187419, "learning_rate": 2.121044724224525e-06, "loss": 1.527, "step": 12072 }, { "epoch": 0.9240012245522731, "grad_norm": 0.6619741004531968, "learning_rate": 2.1208027720287042e-06, "loss": 1.6305, "step": 12073 }, { "epoch": 0.9240777590693403, "grad_norm": 0.6852633981420866, "learning_rate": 2.1205610581823586e-06, "loss": 1.8015, "step": 12074 }, { "epoch": 0.9241542935864074, "grad_norm": 0.5944523317255906, "learning_rate": 2.120319582700339e-06, "loss": 1.4166, "step": 12075 }, { "epoch": 0.9242308281034747, "grad_norm": 0.5039406407722407, "learning_rate": 2.120078345597482e-06, "loss": 1.1007, "step": 12076 }, { "epoch": 0.9243073626205419, "grad_norm": 0.5694134538518713, "learning_rate": 2.11983734688861e-06, "loss": 1.4426, "step": 12077 }, { "epoch": 0.924383897137609, "grad_norm": 0.5993691758438761, "learning_rate": 2.1195965865885313e-06, "loss": 1.5278, "step": 12078 }, { "epoch": 0.9244604316546763, "grad_norm": 1.5110412668144537, "learning_rate": 2.119356064712038e-06, "loss": 1.1799, "step": 12079 }, { "epoch": 0.9245369661717434, "grad_norm": 0.593096350644422, "learning_rate": 2.11911578127391e-06, "loss": 1.5582, "step": 12080 }, { "epoch": 0.9246135006888107, "grad_norm": 1.372866115973166, "learning_rate": 2.1188757362889096e-06, "loss": 1.2765, "step": 12081 }, { "epoch": 0.9246900352058779, "grad_norm": 0.7643374974468893, "learning_rate": 2.118635929771786e-06, "loss": 1.5695, "step": 12082 }, { "epoch": 0.924766569722945, "grad_norm": 0.5508735117747517, "learning_rate": 2.118396361737274e-06, "loss": 1.3087, "step": 12083 }, { "epoch": 0.9248431042400123, "grad_norm": 0.5786668371442953, "learning_rate": 2.118157032200093e-06, "loss": 1.3395, "step": 12084 }, { "epoch": 0.9249196387570794, "grad_norm": 0.5977956523077871, "learning_rate": 2.117917941174949e-06, "loss": 1.3545, "step": 12085 }, { "epoch": 0.9249961732741466, "grad_norm": 0.5951459772301697, "learning_rate": 2.1176790886765316e-06, "loss": 1.4616, "step": 12086 }, { "epoch": 0.9250727077912139, "grad_norm": 0.5314922368542587, "learning_rate": 2.1174404747195175e-06, "loss": 1.1771, "step": 12087 }, { "epoch": 0.925149242308281, "grad_norm": 0.6467185394948168, "learning_rate": 2.117202099318567e-06, "loss": 1.5773, "step": 12088 }, { "epoch": 0.9252257768253482, "grad_norm": 0.9948459347551294, "learning_rate": 2.1169639624883265e-06, "loss": 1.5767, "step": 12089 }, { "epoch": 0.9253023113424155, "grad_norm": 0.5950741784256472, "learning_rate": 2.116726064243429e-06, "loss": 1.4984, "step": 12090 }, { "epoch": 0.9253788458594826, "grad_norm": 0.7365593692702571, "learning_rate": 2.11648840459849e-06, "loss": 1.3424, "step": 12091 }, { "epoch": 0.9254553803765498, "grad_norm": 0.6659285078152125, "learning_rate": 2.116250983568113e-06, "loss": 1.6725, "step": 12092 }, { "epoch": 0.925531914893617, "grad_norm": 0.6140517179685449, "learning_rate": 2.1160138011668864e-06, "loss": 1.3003, "step": 12093 }, { "epoch": 0.9256084494106842, "grad_norm": 0.6377717164965768, "learning_rate": 2.115776857409382e-06, "loss": 1.6038, "step": 12094 }, { "epoch": 0.9256849839277514, "grad_norm": 0.5668174538524067, "learning_rate": 2.115540152310159e-06, "loss": 1.268, "step": 12095 }, { "epoch": 0.9257615184448186, "grad_norm": 0.5894239302595133, "learning_rate": 2.1153036858837624e-06, "loss": 1.3998, "step": 12096 }, { "epoch": 0.9258380529618858, "grad_norm": 0.7027140257287036, "learning_rate": 2.1150674581447196e-06, "loss": 1.5306, "step": 12097 }, { "epoch": 0.925914587478953, "grad_norm": 0.6763459081213906, "learning_rate": 2.1148314691075463e-06, "loss": 1.6164, "step": 12098 }, { "epoch": 0.9259911219960202, "grad_norm": 0.5389603239186186, "learning_rate": 2.1145957187867424e-06, "loss": 0.9928, "step": 12099 }, { "epoch": 0.9260676565130874, "grad_norm": 0.5472492567587373, "learning_rate": 2.1143602071967914e-06, "loss": 1.3015, "step": 12100 }, { "epoch": 0.9261441910301546, "grad_norm": 0.6852999966205796, "learning_rate": 2.1141249343521656e-06, "loss": 1.6985, "step": 12101 }, { "epoch": 0.9262207255472218, "grad_norm": 0.5925373412566918, "learning_rate": 2.113889900267321e-06, "loss": 1.4435, "step": 12102 }, { "epoch": 0.926297260064289, "grad_norm": 0.7326406043554358, "learning_rate": 2.1136551049566974e-06, "loss": 1.6721, "step": 12103 }, { "epoch": 0.9263737945813562, "grad_norm": 0.5776918517690571, "learning_rate": 2.1134205484347227e-06, "loss": 1.3317, "step": 12104 }, { "epoch": 0.9264503290984234, "grad_norm": 0.6905370076437766, "learning_rate": 2.113186230715808e-06, "loss": 1.6195, "step": 12105 }, { "epoch": 0.9265268636154906, "grad_norm": 0.6244454784449075, "learning_rate": 2.1129521518143503e-06, "loss": 1.6224, "step": 12106 }, { "epoch": 0.9266033981325578, "grad_norm": 0.6377076976361349, "learning_rate": 2.1127183117447327e-06, "loss": 1.5268, "step": 12107 }, { "epoch": 0.926679932649625, "grad_norm": 0.6392137994871516, "learning_rate": 2.112484710521323e-06, "loss": 1.6418, "step": 12108 }, { "epoch": 0.9267564671666921, "grad_norm": 0.5570024608573197, "learning_rate": 2.112251348158474e-06, "loss": 1.2936, "step": 12109 }, { "epoch": 0.9268330016837594, "grad_norm": 0.5824920312854663, "learning_rate": 2.1120182246705243e-06, "loss": 1.3337, "step": 12110 }, { "epoch": 0.9269095362008266, "grad_norm": 0.8086191156163055, "learning_rate": 2.111785340071798e-06, "loss": 1.6795, "step": 12111 }, { "epoch": 0.9269860707178937, "grad_norm": 0.6316215510169432, "learning_rate": 2.111552694376604e-06, "loss": 1.2666, "step": 12112 }, { "epoch": 0.927062605234961, "grad_norm": 0.5772661613936598, "learning_rate": 2.111320287599236e-06, "loss": 1.368, "step": 12113 }, { "epoch": 0.9271391397520282, "grad_norm": 0.6170788072642218, "learning_rate": 2.1110881197539756e-06, "loss": 1.1426, "step": 12114 }, { "epoch": 0.9272156742690953, "grad_norm": 0.7460259082769233, "learning_rate": 2.1108561908550858e-06, "loss": 1.4351, "step": 12115 }, { "epoch": 0.9272922087861626, "grad_norm": 0.5702697720510843, "learning_rate": 2.1106245009168176e-06, "loss": 1.3157, "step": 12116 }, { "epoch": 0.9273687433032297, "grad_norm": 0.5265225864504148, "learning_rate": 2.110393049953408e-06, "loss": 1.2824, "step": 12117 }, { "epoch": 0.927445277820297, "grad_norm": 0.5204960832881237, "learning_rate": 2.1101618379790766e-06, "loss": 1.2538, "step": 12118 }, { "epoch": 0.9275218123373642, "grad_norm": 0.7633105482944135, "learning_rate": 2.1099308650080307e-06, "loss": 1.5562, "step": 12119 }, { "epoch": 0.9275983468544313, "grad_norm": 0.6227704845211262, "learning_rate": 2.1097001310544613e-06, "loss": 1.6376, "step": 12120 }, { "epoch": 0.9276748813714986, "grad_norm": 0.6598464866398612, "learning_rate": 2.1094696361325457e-06, "loss": 0.9996, "step": 12121 }, { "epoch": 0.9277514158885657, "grad_norm": 0.5731712488096453, "learning_rate": 2.1092393802564458e-06, "loss": 1.4078, "step": 12122 }, { "epoch": 0.9278279504056329, "grad_norm": 0.6492184120611725, "learning_rate": 2.1090093634403096e-06, "loss": 1.2129, "step": 12123 }, { "epoch": 0.9279044849227002, "grad_norm": 1.8507046574554635, "learning_rate": 2.1087795856982702e-06, "loss": 1.111, "step": 12124 }, { "epoch": 0.9279810194397673, "grad_norm": 0.7344650692773156, "learning_rate": 2.1085500470444457e-06, "loss": 1.5673, "step": 12125 }, { "epoch": 0.9280575539568345, "grad_norm": 0.6875614550648805, "learning_rate": 2.1083207474929395e-06, "loss": 1.3405, "step": 12126 }, { "epoch": 0.9281340884739018, "grad_norm": 0.8317019543222788, "learning_rate": 2.10809168705784e-06, "loss": 1.5013, "step": 12127 }, { "epoch": 0.9282106229909689, "grad_norm": 0.701365176707081, "learning_rate": 2.1078628657532224e-06, "loss": 1.6131, "step": 12128 }, { "epoch": 0.9282871575080361, "grad_norm": 0.5740822155792467, "learning_rate": 2.1076342835931453e-06, "loss": 1.3636, "step": 12129 }, { "epoch": 0.9283636920251033, "grad_norm": 0.6060652978010319, "learning_rate": 2.107405940591654e-06, "loss": 1.2321, "step": 12130 }, { "epoch": 0.9284402265421705, "grad_norm": 0.7073598179266078, "learning_rate": 2.107177836762778e-06, "loss": 1.6154, "step": 12131 }, { "epoch": 0.9285167610592377, "grad_norm": 0.6590581630983526, "learning_rate": 2.1069499721205337e-06, "loss": 1.5558, "step": 12132 }, { "epoch": 0.9285932955763049, "grad_norm": 0.6499484608090323, "learning_rate": 2.1067223466789207e-06, "loss": 1.3518, "step": 12133 }, { "epoch": 0.9286698300933721, "grad_norm": 0.5733254169464621, "learning_rate": 2.1064949604519255e-06, "loss": 1.466, "step": 12134 }, { "epoch": 0.9287463646104394, "grad_norm": 0.6237010162971196, "learning_rate": 2.1062678134535193e-06, "loss": 1.4141, "step": 12135 }, { "epoch": 0.9288228991275065, "grad_norm": 0.5789486952145407, "learning_rate": 2.106040905697658e-06, "loss": 1.2464, "step": 12136 }, { "epoch": 0.9288994336445737, "grad_norm": 0.746910489906974, "learning_rate": 2.105814237198285e-06, "loss": 1.4595, "step": 12137 }, { "epoch": 0.9289759681616409, "grad_norm": 0.6341081507916373, "learning_rate": 2.1055878079693264e-06, "loss": 1.3625, "step": 12138 }, { "epoch": 0.9290525026787081, "grad_norm": 0.6160728371905065, "learning_rate": 2.1053616180246945e-06, "loss": 1.5415, "step": 12139 }, { "epoch": 0.9291290371957753, "grad_norm": 0.6546780562778777, "learning_rate": 2.105135667378288e-06, "loss": 1.432, "step": 12140 }, { "epoch": 0.9292055717128425, "grad_norm": 1.462382580239777, "learning_rate": 2.1049099560439903e-06, "loss": 1.7339, "step": 12141 }, { "epoch": 0.9292821062299097, "grad_norm": 0.7842528932710817, "learning_rate": 2.104684484035668e-06, "loss": 1.7059, "step": 12142 }, { "epoch": 0.9293586407469769, "grad_norm": 0.5639458205402381, "learning_rate": 2.1044592513671765e-06, "loss": 1.256, "step": 12143 }, { "epoch": 0.9294351752640441, "grad_norm": 0.5540664311581037, "learning_rate": 2.1042342580523526e-06, "loss": 1.3268, "step": 12144 }, { "epoch": 0.9295117097811113, "grad_norm": 0.592902951236761, "learning_rate": 2.1040095041050235e-06, "loss": 1.4931, "step": 12145 }, { "epoch": 0.9295882442981784, "grad_norm": 0.5611586033983454, "learning_rate": 2.103784989538997e-06, "loss": 1.306, "step": 12146 }, { "epoch": 0.9296647788152457, "grad_norm": 0.4855877623189091, "learning_rate": 2.1035607143680677e-06, "loss": 0.9961, "step": 12147 }, { "epoch": 0.9297413133323129, "grad_norm": 0.5074330688496853, "learning_rate": 2.103336678606016e-06, "loss": 1.0993, "step": 12148 }, { "epoch": 0.92981784784938, "grad_norm": 0.6082984783657687, "learning_rate": 2.103112882266608e-06, "loss": 1.3476, "step": 12149 }, { "epoch": 0.9298943823664473, "grad_norm": 8.41843295950283, "learning_rate": 2.102889325363594e-06, "loss": 2.1884, "step": 12150 }, { "epoch": 0.9299709168835145, "grad_norm": 0.6051944375864845, "learning_rate": 2.1026660079107106e-06, "loss": 1.5229, "step": 12151 }, { "epoch": 0.9300474514005816, "grad_norm": 0.8378342622082141, "learning_rate": 2.102442929921678e-06, "loss": 1.7874, "step": 12152 }, { "epoch": 0.9301239859176489, "grad_norm": 0.6759633140772741, "learning_rate": 2.102220091410203e-06, "loss": 1.6073, "step": 12153 }, { "epoch": 0.930200520434716, "grad_norm": 0.7293284331789338, "learning_rate": 2.101997492389978e-06, "loss": 1.2598, "step": 12154 }, { "epoch": 0.9302770549517833, "grad_norm": 0.667525039905748, "learning_rate": 2.101775132874679e-06, "loss": 1.4125, "step": 12155 }, { "epoch": 0.9303535894688505, "grad_norm": 0.5232585925215726, "learning_rate": 2.1015530128779705e-06, "loss": 1.3046, "step": 12156 }, { "epoch": 0.9304301239859176, "grad_norm": 0.5713640226269249, "learning_rate": 2.101331132413499e-06, "loss": 1.2537, "step": 12157 }, { "epoch": 0.9305066585029849, "grad_norm": 0.6513888557200744, "learning_rate": 2.1011094914948975e-06, "loss": 1.5139, "step": 12158 }, { "epoch": 0.930583193020052, "grad_norm": 0.5600360777066177, "learning_rate": 2.1008880901357835e-06, "loss": 1.2661, "step": 12159 }, { "epoch": 0.9306597275371192, "grad_norm": 0.5764994761781387, "learning_rate": 2.100666928349762e-06, "loss": 1.3512, "step": 12160 }, { "epoch": 0.9307362620541865, "grad_norm": 0.5830280785646218, "learning_rate": 2.100446006150421e-06, "loss": 1.2533, "step": 12161 }, { "epoch": 0.9308127965712536, "grad_norm": 0.5169313975397846, "learning_rate": 2.1002253235513345e-06, "loss": 1.1466, "step": 12162 }, { "epoch": 0.9308893310883208, "grad_norm": 0.6557748517499995, "learning_rate": 2.100004880566063e-06, "loss": 1.5214, "step": 12163 }, { "epoch": 0.9309658656053881, "grad_norm": 1.5499682797597594, "learning_rate": 2.0997846772081503e-06, "loss": 1.5453, "step": 12164 }, { "epoch": 0.9310424001224552, "grad_norm": 1.2938973461793324, "learning_rate": 2.0995647134911258e-06, "loss": 1.6191, "step": 12165 }, { "epoch": 0.9311189346395224, "grad_norm": 0.5390626654986482, "learning_rate": 2.0993449894285055e-06, "loss": 1.22, "step": 12166 }, { "epoch": 0.9311954691565896, "grad_norm": 0.6637824595962956, "learning_rate": 2.0991255050337903e-06, "loss": 1.4556, "step": 12167 }, { "epoch": 0.9312720036736568, "grad_norm": 0.5964503471365715, "learning_rate": 2.0989062603204654e-06, "loss": 1.3598, "step": 12168 }, { "epoch": 0.931348538190724, "grad_norm": 0.5129359509835035, "learning_rate": 2.0986872553020015e-06, "loss": 1.0735, "step": 12169 }, { "epoch": 0.9314250727077912, "grad_norm": 0.6281089673065676, "learning_rate": 2.0984684899918555e-06, "loss": 1.3613, "step": 12170 }, { "epoch": 0.9315016072248584, "grad_norm": 0.5886402254700053, "learning_rate": 2.0982499644034685e-06, "loss": 1.3395, "step": 12171 }, { "epoch": 0.9315781417419257, "grad_norm": 0.5294987966852395, "learning_rate": 2.098031678550268e-06, "loss": 1.2346, "step": 12172 }, { "epoch": 0.9316546762589928, "grad_norm": 0.6488431937444281, "learning_rate": 2.097813632445666e-06, "loss": 1.4252, "step": 12173 }, { "epoch": 0.93173121077606, "grad_norm": 0.6943901188079422, "learning_rate": 2.0975958261030597e-06, "loss": 1.7797, "step": 12174 }, { "epoch": 0.9318077452931272, "grad_norm": 0.7420388573931714, "learning_rate": 2.097378259535832e-06, "loss": 1.7657, "step": 12175 }, { "epoch": 0.9318842798101944, "grad_norm": 0.74745430812936, "learning_rate": 2.0971609327573505e-06, "loss": 1.7575, "step": 12176 }, { "epoch": 0.9319608143272616, "grad_norm": 0.6647192983224456, "learning_rate": 2.096943845780968e-06, "loss": 1.511, "step": 12177 }, { "epoch": 0.9320373488443288, "grad_norm": 0.6137658476143626, "learning_rate": 2.0967269986200244e-06, "loss": 1.5144, "step": 12178 }, { "epoch": 0.932113883361396, "grad_norm": 0.54838358627906, "learning_rate": 2.096510391287842e-06, "loss": 1.2876, "step": 12179 }, { "epoch": 0.9321904178784632, "grad_norm": 0.5197530687321088, "learning_rate": 2.0962940237977303e-06, "loss": 0.9031, "step": 12180 }, { "epoch": 0.9322669523955304, "grad_norm": 0.7821289013243363, "learning_rate": 2.096077896162984e-06, "loss": 1.6779, "step": 12181 }, { "epoch": 0.9323434869125976, "grad_norm": 0.621803161658027, "learning_rate": 2.0958620083968823e-06, "loss": 1.3679, "step": 12182 }, { "epoch": 0.9324200214296647, "grad_norm": 0.8462913769093164, "learning_rate": 2.09564636051269e-06, "loss": 1.5734, "step": 12183 }, { "epoch": 0.932496555946732, "grad_norm": 0.7057709238882012, "learning_rate": 2.095430952523657e-06, "loss": 1.7149, "step": 12184 }, { "epoch": 0.9325730904637992, "grad_norm": 0.5576676050704716, "learning_rate": 2.0952157844430185e-06, "loss": 1.2846, "step": 12185 }, { "epoch": 0.9326496249808663, "grad_norm": 0.7953396138750864, "learning_rate": 2.0950008562839953e-06, "loss": 1.8372, "step": 12186 }, { "epoch": 0.9327261594979336, "grad_norm": 0.5393430780635126, "learning_rate": 2.094786168059794e-06, "loss": 1.2765, "step": 12187 }, { "epoch": 0.9328026940150007, "grad_norm": 0.5195825089309984, "learning_rate": 2.0945717197836036e-06, "loss": 1.159, "step": 12188 }, { "epoch": 0.932879228532068, "grad_norm": 0.6370927569769614, "learning_rate": 2.0943575114686023e-06, "loss": 1.4717, "step": 12189 }, { "epoch": 0.9329557630491352, "grad_norm": 0.5630122565905838, "learning_rate": 2.094143543127951e-06, "loss": 1.2246, "step": 12190 }, { "epoch": 0.9330322975662023, "grad_norm": 0.6802466933220098, "learning_rate": 2.093929814774797e-06, "loss": 1.3196, "step": 12191 }, { "epoch": 0.9331088320832696, "grad_norm": 0.6489582030525745, "learning_rate": 2.0937163264222723e-06, "loss": 1.6331, "step": 12192 }, { "epoch": 0.9331853666003368, "grad_norm": 0.5043608518085886, "learning_rate": 2.0935030780834936e-06, "loss": 1.1991, "step": 12193 }, { "epoch": 0.9332619011174039, "grad_norm": 0.6270933995958713, "learning_rate": 2.0932900697715642e-06, "loss": 1.5049, "step": 12194 }, { "epoch": 0.9333384356344712, "grad_norm": 0.5964750260164128, "learning_rate": 2.0930773014995716e-06, "loss": 1.244, "step": 12195 }, { "epoch": 0.9334149701515383, "grad_norm": 0.6524620635167714, "learning_rate": 2.0928647732805893e-06, "loss": 1.5961, "step": 12196 }, { "epoch": 0.9334915046686055, "grad_norm": 0.6413877881881238, "learning_rate": 2.0926524851276756e-06, "loss": 1.4906, "step": 12197 }, { "epoch": 0.9335680391856728, "grad_norm": 0.543823343237875, "learning_rate": 2.092440437053874e-06, "loss": 1.3037, "step": 12198 }, { "epoch": 0.9336445737027399, "grad_norm": 0.6232887748613054, "learning_rate": 2.0922286290722126e-06, "loss": 1.5116, "step": 12199 }, { "epoch": 0.9337211082198071, "grad_norm": 0.6931381085190116, "learning_rate": 2.0920170611957065e-06, "loss": 1.6702, "step": 12200 }, { "epoch": 0.9337976427368744, "grad_norm": 0.6308256672413567, "learning_rate": 2.0918057334373553e-06, "loss": 1.5492, "step": 12201 }, { "epoch": 0.9338741772539415, "grad_norm": 0.5739408068255178, "learning_rate": 2.0915946458101425e-06, "loss": 1.4297, "step": 12202 }, { "epoch": 0.9339507117710087, "grad_norm": 0.6418228158780448, "learning_rate": 2.091383798327039e-06, "loss": 1.6172, "step": 12203 }, { "epoch": 0.9340272462880759, "grad_norm": 0.530209222598515, "learning_rate": 2.0911731910009993e-06, "loss": 1.158, "step": 12204 }, { "epoch": 0.9341037808051431, "grad_norm": 0.7093739844579581, "learning_rate": 2.090962823844964e-06, "loss": 1.715, "step": 12205 }, { "epoch": 0.9341803153222104, "grad_norm": 0.5569633918657694, "learning_rate": 2.0907526968718585e-06, "loss": 1.3869, "step": 12206 }, { "epoch": 0.9342568498392775, "grad_norm": 0.5596948138237399, "learning_rate": 2.0905428100945943e-06, "loss": 1.2897, "step": 12207 }, { "epoch": 0.9343333843563447, "grad_norm": 0.6436425509795233, "learning_rate": 2.090333163526066e-06, "loss": 1.6447, "step": 12208 }, { "epoch": 0.934409918873412, "grad_norm": 0.6083434705720716, "learning_rate": 2.090123757179156e-06, "loss": 1.294, "step": 12209 }, { "epoch": 0.9344864533904791, "grad_norm": 0.6090119320971243, "learning_rate": 2.089914591066731e-06, "loss": 1.535, "step": 12210 }, { "epoch": 0.9345629879075463, "grad_norm": 0.595696917509832, "learning_rate": 2.089705665201642e-06, "loss": 1.4049, "step": 12211 }, { "epoch": 0.9346395224246135, "grad_norm": 0.5976999926100895, "learning_rate": 2.0894969795967273e-06, "loss": 1.3822, "step": 12212 }, { "epoch": 0.9347160569416807, "grad_norm": 0.6574122687216616, "learning_rate": 2.0892885342648077e-06, "loss": 1.3715, "step": 12213 }, { "epoch": 0.9347925914587479, "grad_norm": 0.9530122749923853, "learning_rate": 2.0890803292186914e-06, "loss": 1.2161, "step": 12214 }, { "epoch": 0.9348691259758151, "grad_norm": 0.5739497001559297, "learning_rate": 2.0888723644711707e-06, "loss": 1.2582, "step": 12215 }, { "epoch": 0.9349456604928823, "grad_norm": 0.5907405847705969, "learning_rate": 2.088664640035025e-06, "loss": 1.4587, "step": 12216 }, { "epoch": 0.9350221950099495, "grad_norm": 0.5328577574129023, "learning_rate": 2.0884571559230157e-06, "loss": 1.2945, "step": 12217 }, { "epoch": 0.9350987295270167, "grad_norm": 2.877084325331732, "learning_rate": 2.0882499121478923e-06, "loss": 1.8677, "step": 12218 }, { "epoch": 0.9351752640440839, "grad_norm": 0.5944348763567973, "learning_rate": 2.0880429087223886e-06, "loss": 1.43, "step": 12219 }, { "epoch": 0.935251798561151, "grad_norm": 0.6340566179320634, "learning_rate": 2.0878361456592227e-06, "loss": 1.1647, "step": 12220 }, { "epoch": 0.9353283330782183, "grad_norm": 0.5242195537394486, "learning_rate": 2.087629622971099e-06, "loss": 1.1936, "step": 12221 }, { "epoch": 0.9354048675952855, "grad_norm": 0.6609908169336235, "learning_rate": 2.087423340670707e-06, "loss": 1.5753, "step": 12222 }, { "epoch": 0.9354814021123526, "grad_norm": 0.5866645208955048, "learning_rate": 2.0872172987707216e-06, "loss": 1.3464, "step": 12223 }, { "epoch": 0.9355579366294199, "grad_norm": 0.6108021655853725, "learning_rate": 2.087011497283802e-06, "loss": 1.3668, "step": 12224 }, { "epoch": 0.935634471146487, "grad_norm": 0.5501692293543736, "learning_rate": 2.0868059362225943e-06, "loss": 1.2014, "step": 12225 }, { "epoch": 0.9357110056635543, "grad_norm": 0.5887441378043666, "learning_rate": 2.086600615599727e-06, "loss": 1.3673, "step": 12226 }, { "epoch": 0.9357875401806215, "grad_norm": 0.6711834782313949, "learning_rate": 2.086395535427818e-06, "loss": 1.6317, "step": 12227 }, { "epoch": 0.9358640746976886, "grad_norm": 0.5872532822644666, "learning_rate": 2.0861906957194662e-06, "loss": 1.3172, "step": 12228 }, { "epoch": 0.9359406092147559, "grad_norm": 0.5943022772315025, "learning_rate": 2.0859860964872582e-06, "loss": 1.4231, "step": 12229 }, { "epoch": 0.9360171437318231, "grad_norm": 0.5764503275162781, "learning_rate": 2.085781737743765e-06, "loss": 1.3367, "step": 12230 }, { "epoch": 0.9360936782488902, "grad_norm": 0.6379279860421306, "learning_rate": 2.0855776195015433e-06, "loss": 1.6393, "step": 12231 }, { "epoch": 0.9361702127659575, "grad_norm": 0.7033031709925568, "learning_rate": 2.085373741773134e-06, "loss": 1.659, "step": 12232 }, { "epoch": 0.9362467472830246, "grad_norm": 0.4969369519122522, "learning_rate": 2.085170104571065e-06, "loss": 1.1304, "step": 12233 }, { "epoch": 0.9363232818000918, "grad_norm": 0.5254075172106979, "learning_rate": 2.0849667079078474e-06, "loss": 1.1173, "step": 12234 }, { "epoch": 0.9363998163171591, "grad_norm": 0.5494384808218525, "learning_rate": 2.0847635517959793e-06, "loss": 1.2769, "step": 12235 }, { "epoch": 0.9364763508342262, "grad_norm": 0.6154736321327421, "learning_rate": 2.084560636247943e-06, "loss": 1.493, "step": 12236 }, { "epoch": 0.9365528853512934, "grad_norm": 0.6183065435895272, "learning_rate": 2.084357961276206e-06, "loss": 1.4503, "step": 12237 }, { "epoch": 0.9366294198683607, "grad_norm": 0.6455302534374938, "learning_rate": 2.0841555268932214e-06, "loss": 1.6236, "step": 12238 }, { "epoch": 0.9367059543854278, "grad_norm": 0.5537670629751158, "learning_rate": 2.0839533331114275e-06, "loss": 1.2844, "step": 12239 }, { "epoch": 0.936782488902495, "grad_norm": 0.7155621033841352, "learning_rate": 2.083751379943247e-06, "loss": 1.6034, "step": 12240 }, { "epoch": 0.9368590234195622, "grad_norm": 0.8461726873109041, "learning_rate": 2.08354966740109e-06, "loss": 1.5544, "step": 12241 }, { "epoch": 0.9369355579366294, "grad_norm": 0.5006035864772531, "learning_rate": 2.083348195497349e-06, "loss": 1.0721, "step": 12242 }, { "epoch": 0.9370120924536967, "grad_norm": 0.7328689346594528, "learning_rate": 2.0831469642444034e-06, "loss": 1.5965, "step": 12243 }, { "epoch": 0.9370886269707638, "grad_norm": 1.3172381810179266, "learning_rate": 2.0829459736546173e-06, "loss": 1.6468, "step": 12244 }, { "epoch": 0.937165161487831, "grad_norm": 0.5927044291186745, "learning_rate": 2.0827452237403405e-06, "loss": 1.2683, "step": 12245 }, { "epoch": 0.9372416960048983, "grad_norm": 0.521732135841404, "learning_rate": 2.082544714513908e-06, "loss": 1.2189, "step": 12246 }, { "epoch": 0.9373182305219654, "grad_norm": 0.5418044539399525, "learning_rate": 2.082344445987639e-06, "loss": 1.263, "step": 12247 }, { "epoch": 0.9373947650390326, "grad_norm": 0.5894515025294811, "learning_rate": 2.082144418173838e-06, "loss": 1.3989, "step": 12248 }, { "epoch": 0.9374712995560998, "grad_norm": 0.6160802642141443, "learning_rate": 2.081944631084797e-06, "loss": 1.2891, "step": 12249 }, { "epoch": 0.937547834073167, "grad_norm": 0.544830267875606, "learning_rate": 2.08174508473279e-06, "loss": 1.2106, "step": 12250 }, { "epoch": 0.9376243685902342, "grad_norm": 0.7936921674233508, "learning_rate": 2.081545779130079e-06, "loss": 1.6948, "step": 12251 }, { "epoch": 0.9377009031073014, "grad_norm": 0.6881851439269797, "learning_rate": 2.0813467142889087e-06, "loss": 1.4748, "step": 12252 }, { "epoch": 0.9377774376243686, "grad_norm": 0.5788418524021062, "learning_rate": 2.0811478902215107e-06, "loss": 1.246, "step": 12253 }, { "epoch": 0.9378539721414357, "grad_norm": 0.7955316952939712, "learning_rate": 2.080949306940102e-06, "loss": 1.5348, "step": 12254 }, { "epoch": 0.937930506658503, "grad_norm": 0.7757387196263704, "learning_rate": 2.0807509644568833e-06, "loss": 1.9883, "step": 12255 }, { "epoch": 0.9380070411755702, "grad_norm": 0.5375002517032227, "learning_rate": 2.080552862784042e-06, "loss": 1.2143, "step": 12256 }, { "epoch": 0.9380835756926373, "grad_norm": 0.6269430688386051, "learning_rate": 2.0803550019337494e-06, "loss": 1.4253, "step": 12257 }, { "epoch": 0.9381601102097046, "grad_norm": 0.7902678637504276, "learning_rate": 2.0801573819181633e-06, "loss": 1.9897, "step": 12258 }, { "epoch": 0.9382366447267718, "grad_norm": 0.6219822216388056, "learning_rate": 2.0799600027494255e-06, "loss": 1.5475, "step": 12259 }, { "epoch": 0.938313179243839, "grad_norm": 0.5929282736020831, "learning_rate": 2.079762864439664e-06, "loss": 1.322, "step": 12260 }, { "epoch": 0.9383897137609062, "grad_norm": 0.5997994290064002, "learning_rate": 2.079565967000991e-06, "loss": 1.5205, "step": 12261 }, { "epoch": 0.9384662482779733, "grad_norm": 0.6085510786021066, "learning_rate": 2.0793693104455052e-06, "loss": 1.3895, "step": 12262 }, { "epoch": 0.9385427827950406, "grad_norm": 0.5759115771237313, "learning_rate": 2.0791728947852894e-06, "loss": 1.2807, "step": 12263 }, { "epoch": 0.9386193173121078, "grad_norm": 0.6398925577333839, "learning_rate": 2.0789767200324115e-06, "loss": 1.7129, "step": 12264 }, { "epoch": 0.9386958518291749, "grad_norm": 0.6116009457718848, "learning_rate": 2.078780786198926e-06, "loss": 1.4535, "step": 12265 }, { "epoch": 0.9387723863462422, "grad_norm": 0.6431849704173704, "learning_rate": 2.078585093296871e-06, "loss": 1.5928, "step": 12266 }, { "epoch": 0.9388489208633094, "grad_norm": 1.6169023410002628, "learning_rate": 2.078389641338271e-06, "loss": 1.2325, "step": 12267 }, { "epoch": 0.9389254553803765, "grad_norm": 0.6060973763457959, "learning_rate": 2.0781944303351347e-06, "loss": 1.4631, "step": 12268 }, { "epoch": 0.9390019898974438, "grad_norm": 0.5788294568478385, "learning_rate": 2.0779994602994566e-06, "loss": 1.318, "step": 12269 }, { "epoch": 0.9390785244145109, "grad_norm": 0.9879159792688971, "learning_rate": 2.0778047312432163e-06, "loss": 1.6316, "step": 12270 }, { "epoch": 0.9391550589315781, "grad_norm": 0.6037649698763923, "learning_rate": 2.077610243178378e-06, "loss": 1.4704, "step": 12271 }, { "epoch": 0.9392315934486454, "grad_norm": 0.4995728845064838, "learning_rate": 2.0774159961168925e-06, "loss": 1.0335, "step": 12272 }, { "epoch": 0.9393081279657125, "grad_norm": 0.5979151079921533, "learning_rate": 2.0772219900706944e-06, "loss": 1.3217, "step": 12273 }, { "epoch": 0.9393846624827797, "grad_norm": 1.1439146566391831, "learning_rate": 2.0770282250517037e-06, "loss": 1.707, "step": 12274 }, { "epoch": 0.939461196999847, "grad_norm": 0.5358475927268561, "learning_rate": 2.076834701071827e-06, "loss": 1.2723, "step": 12275 }, { "epoch": 0.9395377315169141, "grad_norm": 0.5694241486981579, "learning_rate": 2.0766414181429536e-06, "loss": 1.1177, "step": 12276 }, { "epoch": 0.9396142660339813, "grad_norm": 0.6441520424254974, "learning_rate": 2.07644837627696e-06, "loss": 1.4994, "step": 12277 }, { "epoch": 0.9396908005510485, "grad_norm": 0.5921230102567193, "learning_rate": 2.076255575485708e-06, "loss": 1.3149, "step": 12278 }, { "epoch": 0.9397673350681157, "grad_norm": 0.8261353720820687, "learning_rate": 2.0760630157810433e-06, "loss": 1.5949, "step": 12279 }, { "epoch": 0.939843869585183, "grad_norm": 0.6351065386071164, "learning_rate": 2.0758706971747973e-06, "loss": 1.5424, "step": 12280 }, { "epoch": 0.9399204041022501, "grad_norm": 0.6685142308705759, "learning_rate": 2.0756786196787855e-06, "loss": 1.6261, "step": 12281 }, { "epoch": 0.9399969386193173, "grad_norm": 0.6609458256614693, "learning_rate": 2.075486783304812e-06, "loss": 1.3867, "step": 12282 }, { "epoch": 0.9400734731363846, "grad_norm": 0.5176964327905063, "learning_rate": 2.075295188064662e-06, "loss": 1.171, "step": 12283 }, { "epoch": 0.9401500076534517, "grad_norm": 0.6735851083864218, "learning_rate": 2.0751038339701078e-06, "loss": 1.6939, "step": 12284 }, { "epoch": 0.9402265421705189, "grad_norm": 0.7167527834967268, "learning_rate": 2.0749127210329086e-06, "loss": 1.5054, "step": 12285 }, { "epoch": 0.9403030766875861, "grad_norm": 0.592752181357556, "learning_rate": 2.0747218492648043e-06, "loss": 1.3594, "step": 12286 }, { "epoch": 0.9403796112046533, "grad_norm": 0.5951218347149754, "learning_rate": 2.0745312186775245e-06, "loss": 1.38, "step": 12287 }, { "epoch": 0.9404561457217205, "grad_norm": 0.5914826923597006, "learning_rate": 2.0743408292827815e-06, "loss": 1.4129, "step": 12288 }, { "epoch": 0.9405326802387877, "grad_norm": 0.6528913821220095, "learning_rate": 2.0741506810922733e-06, "loss": 1.6775, "step": 12289 }, { "epoch": 0.9406092147558549, "grad_norm": 0.5734246062675666, "learning_rate": 2.0739607741176836e-06, "loss": 1.2425, "step": 12290 }, { "epoch": 0.940685749272922, "grad_norm": 0.5896763526434479, "learning_rate": 2.07377110837068e-06, "loss": 1.3172, "step": 12291 }, { "epoch": 0.9407622837899893, "grad_norm": 0.6580914302073603, "learning_rate": 2.073581683862917e-06, "loss": 1.3723, "step": 12292 }, { "epoch": 0.9408388183070565, "grad_norm": 0.575849786285326, "learning_rate": 2.0733925006060337e-06, "loss": 1.3871, "step": 12293 }, { "epoch": 0.9409153528241236, "grad_norm": 0.7928031139001955, "learning_rate": 2.073203558611653e-06, "loss": 1.4463, "step": 12294 }, { "epoch": 0.9409918873411909, "grad_norm": 0.6209194203581027, "learning_rate": 2.0730148578913848e-06, "loss": 1.5112, "step": 12295 }, { "epoch": 0.9410684218582581, "grad_norm": 0.6675800194748694, "learning_rate": 2.0728263984568228e-06, "loss": 1.458, "step": 12296 }, { "epoch": 0.9411449563753252, "grad_norm": 0.9111220463928855, "learning_rate": 2.0726381803195464e-06, "loss": 1.4925, "step": 12297 }, { "epoch": 0.9412214908923925, "grad_norm": 0.5473158082885097, "learning_rate": 2.072450203491122e-06, "loss": 1.1742, "step": 12298 }, { "epoch": 0.9412980254094596, "grad_norm": 0.6740310161457065, "learning_rate": 2.0722624679830973e-06, "loss": 1.5189, "step": 12299 }, { "epoch": 0.9413745599265269, "grad_norm": 0.6500357651401485, "learning_rate": 2.0720749738070087e-06, "loss": 1.7068, "step": 12300 }, { "epoch": 0.9414510944435941, "grad_norm": 0.547770850228023, "learning_rate": 2.071887720974376e-06, "loss": 1.2674, "step": 12301 }, { "epoch": 0.9415276289606612, "grad_norm": 0.5619852579051156, "learning_rate": 2.0717007094967043e-06, "loss": 1.0202, "step": 12302 }, { "epoch": 0.9416041634777285, "grad_norm": 0.524669976615568, "learning_rate": 2.071513939385484e-06, "loss": 1.1208, "step": 12303 }, { "epoch": 0.9416806979947957, "grad_norm": 0.6218277379723918, "learning_rate": 2.0713274106521917e-06, "loss": 1.4813, "step": 12304 }, { "epoch": 0.9417572325118628, "grad_norm": 0.5861309308073809, "learning_rate": 2.071141123308288e-06, "loss": 1.4513, "step": 12305 }, { "epoch": 0.9418337670289301, "grad_norm": 0.6515163950489934, "learning_rate": 2.0709550773652183e-06, "loss": 1.5609, "step": 12306 }, { "epoch": 0.9419103015459972, "grad_norm": 0.5665177050309019, "learning_rate": 2.0707692728344143e-06, "loss": 1.4149, "step": 12307 }, { "epoch": 0.9419868360630644, "grad_norm": 0.620686810671482, "learning_rate": 2.0705837097272928e-06, "loss": 1.4731, "step": 12308 }, { "epoch": 0.9420633705801317, "grad_norm": 0.5299751021316909, "learning_rate": 2.0703983880552543e-06, "loss": 1.2663, "step": 12309 }, { "epoch": 0.9421399050971988, "grad_norm": 1.017454782876199, "learning_rate": 2.0702133078296868e-06, "loss": 1.4668, "step": 12310 }, { "epoch": 0.942216439614266, "grad_norm": 0.5688224166839317, "learning_rate": 2.070028469061961e-06, "loss": 1.3015, "step": 12311 }, { "epoch": 0.9422929741313333, "grad_norm": 0.6781919821948269, "learning_rate": 2.0698438717634347e-06, "loss": 1.5267, "step": 12312 }, { "epoch": 0.9423695086484004, "grad_norm": 0.702956944872617, "learning_rate": 2.069659515945449e-06, "loss": 1.3911, "step": 12313 }, { "epoch": 0.9424460431654677, "grad_norm": 1.1097544771313903, "learning_rate": 2.069475401619333e-06, "loss": 1.5965, "step": 12314 }, { "epoch": 0.9425225776825348, "grad_norm": 0.9118857860420334, "learning_rate": 2.0692915287963986e-06, "loss": 1.5009, "step": 12315 }, { "epoch": 0.942599112199602, "grad_norm": 0.6043433724061877, "learning_rate": 2.0691078974879426e-06, "loss": 1.4171, "step": 12316 }, { "epoch": 0.9426756467166693, "grad_norm": 0.5006270542382053, "learning_rate": 2.068924507705249e-06, "loss": 1.0798, "step": 12317 }, { "epoch": 0.9427521812337364, "grad_norm": 0.7107411169505505, "learning_rate": 2.068741359459585e-06, "loss": 1.6806, "step": 12318 }, { "epoch": 0.9428287157508036, "grad_norm": 0.590292920134989, "learning_rate": 2.0685584527622035e-06, "loss": 1.3568, "step": 12319 }, { "epoch": 0.9429052502678709, "grad_norm": 0.6698602723988409, "learning_rate": 2.0683757876243445e-06, "loss": 1.2124, "step": 12320 }, { "epoch": 0.942981784784938, "grad_norm": 0.65186088996709, "learning_rate": 2.0681933640572302e-06, "loss": 1.1377, "step": 12321 }, { "epoch": 0.9430583193020052, "grad_norm": 0.7308456635320701, "learning_rate": 2.068011182072069e-06, "loss": 1.7056, "step": 12322 }, { "epoch": 0.9431348538190724, "grad_norm": 0.6851739910521358, "learning_rate": 2.0678292416800557e-06, "loss": 1.4472, "step": 12323 }, { "epoch": 0.9432113883361396, "grad_norm": 0.6190129651064662, "learning_rate": 2.0676475428923685e-06, "loss": 1.4484, "step": 12324 }, { "epoch": 0.9432879228532068, "grad_norm": 0.5794889492324986, "learning_rate": 2.0674660857201713e-06, "loss": 1.4222, "step": 12325 }, { "epoch": 0.943364457370274, "grad_norm": 0.7405605476306689, "learning_rate": 2.0672848701746146e-06, "loss": 1.6408, "step": 12326 }, { "epoch": 0.9434409918873412, "grad_norm": 0.6284020313569523, "learning_rate": 2.0671038962668313e-06, "loss": 1.6191, "step": 12327 }, { "epoch": 0.9435175264044083, "grad_norm": 0.6277651940904386, "learning_rate": 2.0669231640079424e-06, "loss": 1.4358, "step": 12328 }, { "epoch": 0.9435940609214756, "grad_norm": 1.3054756764951156, "learning_rate": 2.066742673409052e-06, "loss": 1.2462, "step": 12329 }, { "epoch": 0.9436705954385428, "grad_norm": 0.578053644546718, "learning_rate": 2.0665624244812497e-06, "loss": 1.3202, "step": 12330 }, { "epoch": 0.9437471299556099, "grad_norm": 0.6451092165391437, "learning_rate": 2.06638241723561e-06, "loss": 1.5482, "step": 12331 }, { "epoch": 0.9438236644726772, "grad_norm": 0.5364341698818067, "learning_rate": 2.066202651683195e-06, "loss": 1.2308, "step": 12332 }, { "epoch": 0.9439001989897444, "grad_norm": 0.5271164706306607, "learning_rate": 2.0660231278350486e-06, "loss": 1.0778, "step": 12333 }, { "epoch": 0.9439767335068115, "grad_norm": 0.7084102093528503, "learning_rate": 2.0658438457022016e-06, "loss": 1.4291, "step": 12334 }, { "epoch": 0.9440532680238788, "grad_norm": 0.5142860251751374, "learning_rate": 2.0656648052956697e-06, "loss": 0.9383, "step": 12335 }, { "epoch": 0.9441298025409459, "grad_norm": 0.5764586016883854, "learning_rate": 2.0654860066264535e-06, "loss": 1.1916, "step": 12336 }, { "epoch": 0.9442063370580132, "grad_norm": 0.5577334541579847, "learning_rate": 2.065307449705539e-06, "loss": 1.2446, "step": 12337 }, { "epoch": 0.9442828715750804, "grad_norm": 0.6558565425682041, "learning_rate": 2.065129134543897e-06, "loss": 1.707, "step": 12338 }, { "epoch": 0.9443594060921475, "grad_norm": 0.6532091551904047, "learning_rate": 2.064951061152484e-06, "loss": 1.5323, "step": 12339 }, { "epoch": 0.9444359406092148, "grad_norm": 0.6116953708812841, "learning_rate": 2.064773229542242e-06, "loss": 1.3808, "step": 12340 }, { "epoch": 0.944512475126282, "grad_norm": 0.652006670027058, "learning_rate": 2.064595639724096e-06, "loss": 1.2728, "step": 12341 }, { "epoch": 0.9445890096433491, "grad_norm": 0.5539994741107273, "learning_rate": 2.064418291708959e-06, "loss": 1.2299, "step": 12342 }, { "epoch": 0.9446655441604164, "grad_norm": 0.6425982061363897, "learning_rate": 2.0642411855077275e-06, "loss": 1.5921, "step": 12343 }, { "epoch": 0.9447420786774835, "grad_norm": 0.5656005808275119, "learning_rate": 2.064064321131283e-06, "loss": 1.3619, "step": 12344 }, { "epoch": 0.9448186131945507, "grad_norm": 0.5566883447437119, "learning_rate": 2.063887698590493e-06, "loss": 1.2495, "step": 12345 }, { "epoch": 0.944895147711618, "grad_norm": 0.7943397254912291, "learning_rate": 2.063711317896209e-06, "loss": 1.7267, "step": 12346 }, { "epoch": 0.9449716822286851, "grad_norm": 0.6225009564201521, "learning_rate": 2.063535179059269e-06, "loss": 1.4133, "step": 12347 }, { "epoch": 0.9450482167457523, "grad_norm": 0.7318587580611278, "learning_rate": 2.0633592820904956e-06, "loss": 1.6655, "step": 12348 }, { "epoch": 0.9451247512628196, "grad_norm": 0.5309271229820963, "learning_rate": 2.0631836270006963e-06, "loss": 1.2036, "step": 12349 }, { "epoch": 0.9452012857798867, "grad_norm": 0.5606633729681715, "learning_rate": 2.063008213800664e-06, "loss": 1.4031, "step": 12350 }, { "epoch": 0.945277820296954, "grad_norm": 0.5480985312513997, "learning_rate": 2.0628330425011765e-06, "loss": 1.1401, "step": 12351 }, { "epoch": 0.9453543548140211, "grad_norm": 0.7704887966915325, "learning_rate": 2.0626581131129956e-06, "loss": 1.7435, "step": 12352 }, { "epoch": 0.9454308893310883, "grad_norm": 0.5548438723812719, "learning_rate": 2.062483425646872e-06, "loss": 1.2437, "step": 12353 }, { "epoch": 0.9455074238481556, "grad_norm": 0.6691036081566792, "learning_rate": 2.062308980113537e-06, "loss": 1.5924, "step": 12354 }, { "epoch": 0.9455839583652227, "grad_norm": 0.5799945542966491, "learning_rate": 2.0621347765237095e-06, "loss": 1.1597, "step": 12355 }, { "epoch": 0.9456604928822899, "grad_norm": 0.5529868656554351, "learning_rate": 2.061960814888094e-06, "loss": 1.142, "step": 12356 }, { "epoch": 0.945737027399357, "grad_norm": 0.5217813958697375, "learning_rate": 2.0617870952173777e-06, "loss": 1.2159, "step": 12357 }, { "epoch": 0.9458135619164243, "grad_norm": 0.8598292785562752, "learning_rate": 2.061613617522236e-06, "loss": 1.8003, "step": 12358 }, { "epoch": 0.9458900964334915, "grad_norm": 0.6607327871579836, "learning_rate": 2.061440381813327e-06, "loss": 1.0875, "step": 12359 }, { "epoch": 0.9459666309505587, "grad_norm": 0.5067107261827842, "learning_rate": 2.0612673881012953e-06, "loss": 1.1022, "step": 12360 }, { "epoch": 0.9460431654676259, "grad_norm": 0.5392034800152096, "learning_rate": 2.0610946363967694e-06, "loss": 1.2205, "step": 12361 }, { "epoch": 0.9461196999846931, "grad_norm": 0.5703321631558587, "learning_rate": 2.0609221267103642e-06, "loss": 1.2956, "step": 12362 }, { "epoch": 0.9461962345017603, "grad_norm": 0.5600203792625597, "learning_rate": 2.060749859052679e-06, "loss": 1.3005, "step": 12363 }, { "epoch": 0.9462727690188275, "grad_norm": 0.708452321496981, "learning_rate": 2.060577833434298e-06, "loss": 1.6063, "step": 12364 }, { "epoch": 0.9463493035358946, "grad_norm": 0.5761653803755986, "learning_rate": 2.0604060498657926e-06, "loss": 1.3608, "step": 12365 }, { "epoch": 0.9464258380529619, "grad_norm": 0.5226834527607309, "learning_rate": 2.0602345083577164e-06, "loss": 1.0777, "step": 12366 }, { "epoch": 0.9465023725700291, "grad_norm": 0.6026510317151762, "learning_rate": 2.0600632089206095e-06, "loss": 1.3933, "step": 12367 }, { "epoch": 0.9465789070870962, "grad_norm": 0.6478015320168629, "learning_rate": 2.0598921515649976e-06, "loss": 1.3979, "step": 12368 }, { "epoch": 0.9466554416041635, "grad_norm": 0.5712555085618181, "learning_rate": 2.05972133630139e-06, "loss": 1.5096, "step": 12369 }, { "epoch": 0.9467319761212307, "grad_norm": 0.564328671981039, "learning_rate": 2.0595507631402826e-06, "loss": 1.2015, "step": 12370 }, { "epoch": 0.9468085106382979, "grad_norm": 0.6056662788029037, "learning_rate": 2.0593804320921566e-06, "loss": 1.5276, "step": 12371 }, { "epoch": 0.9468850451553651, "grad_norm": 0.701426399899545, "learning_rate": 2.0592103431674767e-06, "loss": 1.6612, "step": 12372 }, { "epoch": 0.9469615796724322, "grad_norm": 0.5690052990138871, "learning_rate": 2.0590404963766942e-06, "loss": 1.3995, "step": 12373 }, { "epoch": 0.9470381141894995, "grad_norm": 0.5156511580291279, "learning_rate": 2.0588708917302443e-06, "loss": 1.0919, "step": 12374 }, { "epoch": 0.9471146487065667, "grad_norm": 0.7024772312254178, "learning_rate": 2.0587015292385497e-06, "loss": 1.4556, "step": 12375 }, { "epoch": 0.9471911832236338, "grad_norm": 0.5162358753951847, "learning_rate": 2.0585324089120145e-06, "loss": 1.2028, "step": 12376 }, { "epoch": 0.9472677177407011, "grad_norm": 0.7521303227471281, "learning_rate": 2.0583635307610313e-06, "loss": 1.6412, "step": 12377 }, { "epoch": 0.9473442522577683, "grad_norm": 0.6451967910458843, "learning_rate": 2.0581948947959754e-06, "loss": 1.5173, "step": 12378 }, { "epoch": 0.9474207867748354, "grad_norm": 0.5455778851915413, "learning_rate": 2.058026501027209e-06, "loss": 1.2643, "step": 12379 }, { "epoch": 0.9474973212919027, "grad_norm": 0.590125271837188, "learning_rate": 2.0578583494650793e-06, "loss": 1.3343, "step": 12380 }, { "epoch": 0.9475738558089698, "grad_norm": 0.5231666342593454, "learning_rate": 2.057690440119917e-06, "loss": 0.9408, "step": 12381 }, { "epoch": 0.947650390326037, "grad_norm": 0.6170471613531584, "learning_rate": 2.0575227730020394e-06, "loss": 1.5524, "step": 12382 }, { "epoch": 0.9477269248431043, "grad_norm": 0.6498034919197937, "learning_rate": 2.0573553481217485e-06, "loss": 1.7105, "step": 12383 }, { "epoch": 0.9478034593601714, "grad_norm": 0.5467552232248734, "learning_rate": 2.057188165489331e-06, "loss": 1.2404, "step": 12384 }, { "epoch": 0.9478799938772386, "grad_norm": 0.6158702522537695, "learning_rate": 2.0570212251150594e-06, "loss": 1.4736, "step": 12385 }, { "epoch": 0.9479565283943059, "grad_norm": 0.5821412821821806, "learning_rate": 2.056854527009191e-06, "loss": 1.3375, "step": 12386 }, { "epoch": 0.948033062911373, "grad_norm": 0.6057769431522761, "learning_rate": 2.0566880711819683e-06, "loss": 1.4914, "step": 12387 }, { "epoch": 0.9481095974284403, "grad_norm": 0.6790993678420666, "learning_rate": 2.056521857643619e-06, "loss": 1.4877, "step": 12388 }, { "epoch": 0.9481861319455074, "grad_norm": 0.7161543648387225, "learning_rate": 2.056355886404355e-06, "loss": 1.7772, "step": 12389 }, { "epoch": 0.9482626664625746, "grad_norm": 0.563874287820415, "learning_rate": 2.0561901574743745e-06, "loss": 1.3262, "step": 12390 }, { "epoch": 0.9483392009796419, "grad_norm": 0.61533404797886, "learning_rate": 2.0560246708638607e-06, "loss": 1.6888, "step": 12391 }, { "epoch": 0.948415735496709, "grad_norm": 0.629953620091132, "learning_rate": 2.055859426582982e-06, "loss": 1.6398, "step": 12392 }, { "epoch": 0.9484922700137762, "grad_norm": 0.6538011319121335, "learning_rate": 2.05569442464189e-06, "loss": 1.3977, "step": 12393 }, { "epoch": 0.9485688045308434, "grad_norm": 0.5988665382947977, "learning_rate": 2.055529665050724e-06, "loss": 1.4228, "step": 12394 }, { "epoch": 0.9486453390479106, "grad_norm": 0.576931912008875, "learning_rate": 2.0553651478196074e-06, "loss": 1.3403, "step": 12395 }, { "epoch": 0.9487218735649778, "grad_norm": 0.5446292914939294, "learning_rate": 2.055200872958648e-06, "loss": 1.3012, "step": 12396 }, { "epoch": 0.948798408082045, "grad_norm": 0.595070750464721, "learning_rate": 2.05503684047794e-06, "loss": 1.2078, "step": 12397 }, { "epoch": 0.9488749425991122, "grad_norm": 0.646834331854908, "learning_rate": 2.0548730503875615e-06, "loss": 1.491, "step": 12398 }, { "epoch": 0.9489514771161794, "grad_norm": 0.6576735713722311, "learning_rate": 2.0547095026975765e-06, "loss": 1.4561, "step": 12399 }, { "epoch": 0.9490280116332466, "grad_norm": 0.6098612446775923, "learning_rate": 2.054546197418034e-06, "loss": 1.2626, "step": 12400 }, { "epoch": 0.9491045461503138, "grad_norm": 0.6024325722083151, "learning_rate": 2.054383134558967e-06, "loss": 1.4286, "step": 12401 }, { "epoch": 0.9491810806673809, "grad_norm": 0.5703981908261188, "learning_rate": 2.0542203141303964e-06, "loss": 1.3167, "step": 12402 }, { "epoch": 0.9492576151844482, "grad_norm": 0.5464487262598602, "learning_rate": 2.0540577361423248e-06, "loss": 1.1236, "step": 12403 }, { "epoch": 0.9493341497015154, "grad_norm": 0.5861526514131592, "learning_rate": 2.053895400604742e-06, "loss": 1.4217, "step": 12404 }, { "epoch": 0.9494106842185825, "grad_norm": 0.5109798342488497, "learning_rate": 2.053733307527623e-06, "loss": 1.0463, "step": 12405 }, { "epoch": 0.9494872187356498, "grad_norm": 0.7462818663809779, "learning_rate": 2.0535714569209263e-06, "loss": 1.6181, "step": 12406 }, { "epoch": 0.949563753252717, "grad_norm": 0.5300729138161745, "learning_rate": 2.0534098487945968e-06, "loss": 1.012, "step": 12407 }, { "epoch": 0.9496402877697842, "grad_norm": 0.6802069802304124, "learning_rate": 2.053248483158564e-06, "loss": 1.5922, "step": 12408 }, { "epoch": 0.9497168222868514, "grad_norm": 0.6364254725276395, "learning_rate": 2.0530873600227434e-06, "loss": 1.4651, "step": 12409 }, { "epoch": 0.9497933568039185, "grad_norm": 0.5470410056263476, "learning_rate": 2.052926479397034e-06, "loss": 1.1149, "step": 12410 }, { "epoch": 0.9498698913209858, "grad_norm": 0.6915498587665828, "learning_rate": 2.0527658412913214e-06, "loss": 1.5447, "step": 12411 }, { "epoch": 0.949946425838053, "grad_norm": 0.6478280162683726, "learning_rate": 2.0526054457154754e-06, "loss": 1.3657, "step": 12412 }, { "epoch": 0.9500229603551201, "grad_norm": 0.5781263702294429, "learning_rate": 2.0524452926793513e-06, "loss": 1.4961, "step": 12413 }, { "epoch": 0.9500994948721874, "grad_norm": 0.8717438928780719, "learning_rate": 2.0522853821927893e-06, "loss": 1.4517, "step": 12414 }, { "epoch": 0.9501760293892546, "grad_norm": 0.5265432482835495, "learning_rate": 2.0521257142656152e-06, "loss": 1.1371, "step": 12415 }, { "epoch": 0.9502525639063217, "grad_norm": 0.5322827049074554, "learning_rate": 2.0519662889076387e-06, "loss": 1.2424, "step": 12416 }, { "epoch": 0.950329098423389, "grad_norm": 0.6385189135861503, "learning_rate": 2.0518071061286564e-06, "loss": 1.4891, "step": 12417 }, { "epoch": 0.9504056329404561, "grad_norm": 0.8744599136584376, "learning_rate": 2.051648165938448e-06, "loss": 1.3194, "step": 12418 }, { "epoch": 0.9504821674575233, "grad_norm": 0.6314953054664064, "learning_rate": 2.0514894683467788e-06, "loss": 1.5996, "step": 12419 }, { "epoch": 0.9505587019745906, "grad_norm": 0.5839885318408004, "learning_rate": 2.051331013363401e-06, "loss": 1.2438, "step": 12420 }, { "epoch": 0.9506352364916577, "grad_norm": 0.5650199303109333, "learning_rate": 2.051172800998051e-06, "loss": 1.3363, "step": 12421 }, { "epoch": 0.950711771008725, "grad_norm": 0.8160590380766781, "learning_rate": 2.051014831260448e-06, "loss": 1.7458, "step": 12422 }, { "epoch": 0.9507883055257921, "grad_norm": 0.6069548544486065, "learning_rate": 2.0508571041602984e-06, "loss": 1.4614, "step": 12423 }, { "epoch": 0.9508648400428593, "grad_norm": 0.7132591273087339, "learning_rate": 2.0506996197072946e-06, "loss": 1.7327, "step": 12424 }, { "epoch": 0.9509413745599266, "grad_norm": 0.7164544474218278, "learning_rate": 2.0505423779111127e-06, "loss": 1.415, "step": 12425 }, { "epoch": 0.9510179090769937, "grad_norm": 0.7165313044527059, "learning_rate": 2.0503853787814133e-06, "loss": 1.5767, "step": 12426 }, { "epoch": 0.9510944435940609, "grad_norm": 0.8160812850212585, "learning_rate": 2.050228622327844e-06, "loss": 1.627, "step": 12427 }, { "epoch": 0.9511709781111282, "grad_norm": 0.6808051813883904, "learning_rate": 2.0500721085600346e-06, "loss": 1.3832, "step": 12428 }, { "epoch": 0.9512475126281953, "grad_norm": 0.6038052444690188, "learning_rate": 2.0499158374876036e-06, "loss": 1.4724, "step": 12429 }, { "epoch": 0.9513240471452625, "grad_norm": 0.5590761876673165, "learning_rate": 2.0497598091201517e-06, "loss": 1.385, "step": 12430 }, { "epoch": 0.9514005816623297, "grad_norm": 0.6043082801884901, "learning_rate": 2.0496040234672663e-06, "loss": 1.3587, "step": 12431 }, { "epoch": 0.9514771161793969, "grad_norm": 0.7580366275740773, "learning_rate": 2.0494484805385193e-06, "loss": 1.5224, "step": 12432 }, { "epoch": 0.9515536506964641, "grad_norm": 0.7053100151228971, "learning_rate": 2.0492931803434675e-06, "loss": 1.4454, "step": 12433 }, { "epoch": 0.9516301852135313, "grad_norm": 0.56981748104379, "learning_rate": 2.0491381228916525e-06, "loss": 1.2382, "step": 12434 }, { "epoch": 0.9517067197305985, "grad_norm": 0.579129091426273, "learning_rate": 2.048983308192603e-06, "loss": 1.3456, "step": 12435 }, { "epoch": 0.9517832542476657, "grad_norm": 0.584287735850797, "learning_rate": 2.0488287362558296e-06, "loss": 1.3104, "step": 12436 }, { "epoch": 0.9518597887647329, "grad_norm": 0.5572659803290616, "learning_rate": 2.0486744070908303e-06, "loss": 1.2429, "step": 12437 }, { "epoch": 0.9519363232818001, "grad_norm": 0.6012636855847646, "learning_rate": 2.0485203207070883e-06, "loss": 1.4896, "step": 12438 }, { "epoch": 0.9520128577988672, "grad_norm": 0.5217977305866791, "learning_rate": 2.0483664771140703e-06, "loss": 1.1642, "step": 12439 }, { "epoch": 0.9520893923159345, "grad_norm": 0.6246544228166447, "learning_rate": 2.048212876321228e-06, "loss": 1.4749, "step": 12440 }, { "epoch": 0.9521659268330017, "grad_norm": 0.48938627626028397, "learning_rate": 2.0480595183380015e-06, "loss": 1.1016, "step": 12441 }, { "epoch": 0.9522424613500688, "grad_norm": 0.5868824226076417, "learning_rate": 2.047906403173812e-06, "loss": 1.3115, "step": 12442 }, { "epoch": 0.9523189958671361, "grad_norm": 0.5652845703285613, "learning_rate": 2.0477535308380674e-06, "loss": 1.1991, "step": 12443 }, { "epoch": 0.9523955303842033, "grad_norm": 0.526559181524524, "learning_rate": 2.0476009013401607e-06, "loss": 1.1773, "step": 12444 }, { "epoch": 0.9524720649012705, "grad_norm": 0.5901038645455388, "learning_rate": 2.0474485146894705e-06, "loss": 1.2786, "step": 12445 }, { "epoch": 0.9525485994183377, "grad_norm": 0.5991079538063374, "learning_rate": 2.047296370895359e-06, "loss": 1.4272, "step": 12446 }, { "epoch": 0.9526251339354048, "grad_norm": 0.6293109011463355, "learning_rate": 2.0471444699671745e-06, "loss": 1.4726, "step": 12447 }, { "epoch": 0.9527016684524721, "grad_norm": 0.6061379754623365, "learning_rate": 2.0469928119142513e-06, "loss": 1.3681, "step": 12448 }, { "epoch": 0.9527782029695393, "grad_norm": 0.6247730110083407, "learning_rate": 2.046841396745907e-06, "loss": 1.5022, "step": 12449 }, { "epoch": 0.9528547374866064, "grad_norm": 0.5547604696619521, "learning_rate": 2.0466902244714445e-06, "loss": 1.2953, "step": 12450 }, { "epoch": 0.9529312720036737, "grad_norm": 1.5176937556055405, "learning_rate": 2.046539295100153e-06, "loss": 1.5742, "step": 12451 }, { "epoch": 0.9530078065207409, "grad_norm": 0.5057504740659413, "learning_rate": 2.0463886086413058e-06, "loss": 0.9973, "step": 12452 }, { "epoch": 0.953084341037808, "grad_norm": 0.6739893220818564, "learning_rate": 2.0462381651041616e-06, "loss": 1.4454, "step": 12453 }, { "epoch": 0.9531608755548753, "grad_norm": 0.7227059465906968, "learning_rate": 2.0460879644979636e-06, "loss": 1.6278, "step": 12454 }, { "epoch": 0.9532374100719424, "grad_norm": 0.5199286575781775, "learning_rate": 2.045938006831942e-06, "loss": 1.0988, "step": 12455 }, { "epoch": 0.9533139445890096, "grad_norm": 0.5903641698534512, "learning_rate": 2.045788292115308e-06, "loss": 1.297, "step": 12456 }, { "epoch": 0.9533904791060769, "grad_norm": 0.7216969413049122, "learning_rate": 2.0456388203572634e-06, "loss": 1.5271, "step": 12457 }, { "epoch": 0.953467013623144, "grad_norm": 0.6440248852261432, "learning_rate": 2.045489591566991e-06, "loss": 1.7165, "step": 12458 }, { "epoch": 0.9535435481402113, "grad_norm": 0.5710494237659797, "learning_rate": 2.0453406057536594e-06, "loss": 1.2559, "step": 12459 }, { "epoch": 0.9536200826572784, "grad_norm": 0.5549398719865792, "learning_rate": 2.045191862926423e-06, "loss": 1.3092, "step": 12460 }, { "epoch": 0.9536966171743456, "grad_norm": 0.7078280540022122, "learning_rate": 2.045043363094422e-06, "loss": 1.6253, "step": 12461 }, { "epoch": 0.9537731516914129, "grad_norm": 0.7939064147820835, "learning_rate": 2.044895106266779e-06, "loss": 1.8164, "step": 12462 }, { "epoch": 0.95384968620848, "grad_norm": 0.586365511800334, "learning_rate": 2.0447470924526042e-06, "loss": 1.3981, "step": 12463 }, { "epoch": 0.9539262207255472, "grad_norm": 0.6807617564162732, "learning_rate": 2.0445993216609926e-06, "loss": 1.4884, "step": 12464 }, { "epoch": 0.9540027552426145, "grad_norm": 0.528860468579306, "learning_rate": 2.044451793901023e-06, "loss": 1.1098, "step": 12465 }, { "epoch": 0.9540792897596816, "grad_norm": 0.6818399636474727, "learning_rate": 2.04430450918176e-06, "loss": 1.5824, "step": 12466 }, { "epoch": 0.9541558242767488, "grad_norm": 0.5292521188316119, "learning_rate": 2.0441574675122523e-06, "loss": 1.242, "step": 12467 }, { "epoch": 0.954232358793816, "grad_norm": 0.8064533577918958, "learning_rate": 2.0440106689015365e-06, "loss": 1.4545, "step": 12468 }, { "epoch": 0.9543088933108832, "grad_norm": 0.6553719758087702, "learning_rate": 2.043864113358631e-06, "loss": 1.7525, "step": 12469 }, { "epoch": 0.9543854278279504, "grad_norm": 0.8259863578218407, "learning_rate": 2.0437178008925413e-06, "loss": 1.4614, "step": 12470 }, { "epoch": 0.9544619623450176, "grad_norm": 0.5378593825774426, "learning_rate": 2.0435717315122566e-06, "loss": 1.1347, "step": 12471 }, { "epoch": 0.9545384968620848, "grad_norm": 0.6566987211941319, "learning_rate": 2.043425905226752e-06, "loss": 1.5805, "step": 12472 }, { "epoch": 0.954615031379152, "grad_norm": 0.8774348247962849, "learning_rate": 2.043280322044988e-06, "loss": 1.3983, "step": 12473 }, { "epoch": 0.9546915658962192, "grad_norm": 0.661601134359074, "learning_rate": 2.0431349819759096e-06, "loss": 1.149, "step": 12474 }, { "epoch": 0.9547681004132864, "grad_norm": 0.5713579890646407, "learning_rate": 2.0429898850284466e-06, "loss": 1.3254, "step": 12475 }, { "epoch": 0.9548446349303535, "grad_norm": 0.5611437589845801, "learning_rate": 2.042845031211514e-06, "loss": 1.4268, "step": 12476 }, { "epoch": 0.9549211694474208, "grad_norm": 0.6462663608170978, "learning_rate": 2.0427004205340125e-06, "loss": 1.3657, "step": 12477 }, { "epoch": 0.954997703964488, "grad_norm": 0.5388400640815417, "learning_rate": 2.042556053004827e-06, "loss": 1.2711, "step": 12478 }, { "epoch": 0.9550742384815551, "grad_norm": 0.6878707787180994, "learning_rate": 2.042411928632828e-06, "loss": 1.3245, "step": 12479 }, { "epoch": 0.9551507729986224, "grad_norm": 0.6070880177125718, "learning_rate": 2.0422680474268713e-06, "loss": 1.3465, "step": 12480 }, { "epoch": 0.9552273075156896, "grad_norm": 0.5641437035143144, "learning_rate": 2.042124409395797e-06, "loss": 1.1442, "step": 12481 }, { "epoch": 0.9553038420327568, "grad_norm": 0.583240746746143, "learning_rate": 2.041981014548431e-06, "loss": 1.2245, "step": 12482 }, { "epoch": 0.955380376549824, "grad_norm": 0.624456159068127, "learning_rate": 2.0418378628935838e-06, "loss": 1.1956, "step": 12483 }, { "epoch": 0.9554569110668911, "grad_norm": 0.5973418627253583, "learning_rate": 2.041694954440051e-06, "loss": 1.1884, "step": 12484 }, { "epoch": 0.9555334455839584, "grad_norm": 0.5691445391348035, "learning_rate": 2.0415522891966126e-06, "loss": 1.2936, "step": 12485 }, { "epoch": 0.9556099801010256, "grad_norm": 0.6101441926042257, "learning_rate": 2.0414098671720357e-06, "loss": 1.3772, "step": 12486 }, { "epoch": 0.9556865146180927, "grad_norm": 0.5957880183146543, "learning_rate": 2.04126768837507e-06, "loss": 1.3406, "step": 12487 }, { "epoch": 0.95576304913516, "grad_norm": 0.5708908573400702, "learning_rate": 2.0411257528144527e-06, "loss": 1.4672, "step": 12488 }, { "epoch": 0.9558395836522272, "grad_norm": 1.6484586555767216, "learning_rate": 2.0409840604989026e-06, "loss": 1.3452, "step": 12489 }, { "epoch": 0.9559161181692943, "grad_norm": 0.5899555698348221, "learning_rate": 2.040842611437128e-06, "loss": 1.4354, "step": 12490 }, { "epoch": 0.9559926526863616, "grad_norm": 0.587915194993435, "learning_rate": 2.040701405637819e-06, "loss": 1.3391, "step": 12491 }, { "epoch": 0.9560691872034287, "grad_norm": 0.5848630241259992, "learning_rate": 2.040560443109651e-06, "loss": 1.3709, "step": 12492 }, { "epoch": 0.956145721720496, "grad_norm": 0.6218264216053665, "learning_rate": 2.040419723861287e-06, "loss": 1.3501, "step": 12493 }, { "epoch": 0.9562222562375632, "grad_norm": 0.943003204998483, "learning_rate": 2.0402792479013714e-06, "loss": 1.729, "step": 12494 }, { "epoch": 0.9562987907546303, "grad_norm": 0.6227190754335262, "learning_rate": 2.0401390152385363e-06, "loss": 1.403, "step": 12495 }, { "epoch": 0.9563753252716976, "grad_norm": 0.5593295749210455, "learning_rate": 2.039999025881397e-06, "loss": 1.2605, "step": 12496 }, { "epoch": 0.9564518597887647, "grad_norm": 0.5910330378037315, "learning_rate": 2.0398592798385566e-06, "loss": 1.43, "step": 12497 }, { "epoch": 0.9565283943058319, "grad_norm": 0.6134861799694693, "learning_rate": 2.0397197771186005e-06, "loss": 1.4637, "step": 12498 }, { "epoch": 0.9566049288228992, "grad_norm": 0.5284449134727172, "learning_rate": 2.0395805177301008e-06, "loss": 1.2119, "step": 12499 }, { "epoch": 0.9566814633399663, "grad_norm": 0.6618462010530628, "learning_rate": 2.0394415016816125e-06, "loss": 1.4645, "step": 12500 }, { "epoch": 0.9567579978570335, "grad_norm": 0.6175551836908841, "learning_rate": 2.0393027289816784e-06, "loss": 1.2008, "step": 12501 }, { "epoch": 0.9568345323741008, "grad_norm": 0.593541712328326, "learning_rate": 2.0391641996388243e-06, "loss": 1.4632, "step": 12502 }, { "epoch": 0.9569110668911679, "grad_norm": 0.6110593546288227, "learning_rate": 2.0390259136615634e-06, "loss": 1.3894, "step": 12503 }, { "epoch": 0.9569876014082351, "grad_norm": 0.5983254950841135, "learning_rate": 2.0388878710583913e-06, "loss": 1.2766, "step": 12504 }, { "epoch": 0.9570641359253023, "grad_norm": 0.5032608431394129, "learning_rate": 2.03875007183779e-06, "loss": 1.1007, "step": 12505 }, { "epoch": 0.9571406704423695, "grad_norm": 0.6957759133597666, "learning_rate": 2.038612516008225e-06, "loss": 1.8475, "step": 12506 }, { "epoch": 0.9572172049594367, "grad_norm": 0.5478063492749932, "learning_rate": 2.0384752035781504e-06, "loss": 1.1831, "step": 12507 }, { "epoch": 0.9572937394765039, "grad_norm": 0.594027449899326, "learning_rate": 2.038338134556002e-06, "loss": 1.5989, "step": 12508 }, { "epoch": 0.9573702739935711, "grad_norm": 0.623330809475538, "learning_rate": 2.0382013089502015e-06, "loss": 1.2275, "step": 12509 }, { "epoch": 0.9574468085106383, "grad_norm": 0.5763477150257391, "learning_rate": 2.0380647267691556e-06, "loss": 1.4511, "step": 12510 }, { "epoch": 0.9575233430277055, "grad_norm": 0.5084346419571425, "learning_rate": 2.037928388021257e-06, "loss": 1.0371, "step": 12511 }, { "epoch": 0.9575998775447727, "grad_norm": 0.6958560113867654, "learning_rate": 2.0377922927148837e-06, "loss": 1.7926, "step": 12512 }, { "epoch": 0.9576764120618398, "grad_norm": 0.910370472571557, "learning_rate": 2.037656440858395e-06, "loss": 1.5389, "step": 12513 }, { "epoch": 0.9577529465789071, "grad_norm": 0.6004082572537482, "learning_rate": 2.037520832460141e-06, "loss": 1.4246, "step": 12514 }, { "epoch": 0.9578294810959743, "grad_norm": 0.6112346926820484, "learning_rate": 2.0373854675284525e-06, "loss": 1.4846, "step": 12515 }, { "epoch": 0.9579060156130415, "grad_norm": 0.6702817161273891, "learning_rate": 2.0372503460716463e-06, "loss": 1.6415, "step": 12516 }, { "epoch": 0.9579825501301087, "grad_norm": 0.5745648384927967, "learning_rate": 2.0371154680980256e-06, "loss": 1.1574, "step": 12517 }, { "epoch": 0.9580590846471759, "grad_norm": 0.7363717760505311, "learning_rate": 2.0369808336158768e-06, "loss": 1.362, "step": 12518 }, { "epoch": 0.9581356191642431, "grad_norm": 0.5296623839613203, "learning_rate": 2.0368464426334737e-06, "loss": 1.139, "step": 12519 }, { "epoch": 0.9582121536813103, "grad_norm": 0.4773316417850967, "learning_rate": 2.036712295159072e-06, "loss": 1.0039, "step": 12520 }, { "epoch": 0.9582886881983774, "grad_norm": 0.5500748540828412, "learning_rate": 2.0365783912009147e-06, "loss": 1.2489, "step": 12521 }, { "epoch": 0.9583652227154447, "grad_norm": 0.7494455815821407, "learning_rate": 2.0364447307672297e-06, "loss": 1.5108, "step": 12522 }, { "epoch": 0.9584417572325119, "grad_norm": 0.59308896584969, "learning_rate": 2.0363113138662294e-06, "loss": 1.2778, "step": 12523 }, { "epoch": 0.958518291749579, "grad_norm": 0.5733154252773369, "learning_rate": 2.0361781405061106e-06, "loss": 1.1897, "step": 12524 }, { "epoch": 0.9585948262666463, "grad_norm": 0.6716394790051798, "learning_rate": 2.036045210695057e-06, "loss": 1.73, "step": 12525 }, { "epoch": 0.9586713607837134, "grad_norm": 0.9293363874756952, "learning_rate": 2.0359125244412355e-06, "loss": 1.5983, "step": 12526 }, { "epoch": 0.9587478953007806, "grad_norm": 0.5157825117027427, "learning_rate": 2.0357800817527985e-06, "loss": 0.9611, "step": 12527 }, { "epoch": 0.9588244298178479, "grad_norm": 0.5626577586344331, "learning_rate": 2.0356478826378843e-06, "loss": 1.2071, "step": 12528 }, { "epoch": 0.958900964334915, "grad_norm": 0.5411438372931349, "learning_rate": 2.0355159271046155e-06, "loss": 1.1931, "step": 12529 }, { "epoch": 0.9589774988519822, "grad_norm": 0.6765168142929237, "learning_rate": 2.0353842151610998e-06, "loss": 1.4783, "step": 12530 }, { "epoch": 0.9590540333690495, "grad_norm": 0.5158634606768323, "learning_rate": 2.0352527468154294e-06, "loss": 1.1494, "step": 12531 }, { "epoch": 0.9591305678861166, "grad_norm": 0.4994381065515891, "learning_rate": 2.035121522075683e-06, "loss": 1.0692, "step": 12532 }, { "epoch": 0.9592071024031839, "grad_norm": 0.6527653278663161, "learning_rate": 2.0349905409499224e-06, "loss": 1.4973, "step": 12533 }, { "epoch": 0.959283636920251, "grad_norm": 0.7306274296542187, "learning_rate": 2.034859803446197e-06, "loss": 1.6379, "step": 12534 }, { "epoch": 0.9593601714373182, "grad_norm": 0.6994293068000936, "learning_rate": 2.034729309572538e-06, "loss": 1.37, "step": 12535 }, { "epoch": 0.9594367059543855, "grad_norm": 0.5385120781294565, "learning_rate": 2.0345990593369646e-06, "loss": 1.338, "step": 12536 }, { "epoch": 0.9595132404714526, "grad_norm": 0.7316372850751236, "learning_rate": 2.0344690527474788e-06, "loss": 1.4698, "step": 12537 }, { "epoch": 0.9595897749885198, "grad_norm": 1.0970166561865988, "learning_rate": 2.0343392898120694e-06, "loss": 1.3133, "step": 12538 }, { "epoch": 0.9596663095055871, "grad_norm": 0.5726049019191473, "learning_rate": 2.0342097705387096e-06, "loss": 1.3411, "step": 12539 }, { "epoch": 0.9597428440226542, "grad_norm": 0.49713990287009524, "learning_rate": 2.0340804949353564e-06, "loss": 0.9156, "step": 12540 }, { "epoch": 0.9598193785397214, "grad_norm": 0.5518659150322466, "learning_rate": 2.033951463009953e-06, "loss": 1.1677, "step": 12541 }, { "epoch": 0.9598959130567886, "grad_norm": 0.6885983600636159, "learning_rate": 2.0338226747704288e-06, "loss": 1.6419, "step": 12542 }, { "epoch": 0.9599724475738558, "grad_norm": 0.6548254638735024, "learning_rate": 2.0336941302246956e-06, "loss": 1.5284, "step": 12543 }, { "epoch": 0.960048982090923, "grad_norm": 0.5465026513920223, "learning_rate": 2.033565829380652e-06, "loss": 1.2455, "step": 12544 }, { "epoch": 0.9601255166079902, "grad_norm": 0.5471689854611728, "learning_rate": 2.0334377722461813e-06, "loss": 1.1567, "step": 12545 }, { "epoch": 0.9602020511250574, "grad_norm": 0.7799950023494427, "learning_rate": 2.0333099588291517e-06, "loss": 1.3804, "step": 12546 }, { "epoch": 0.9602785856421246, "grad_norm": 0.8970455955911841, "learning_rate": 2.0331823891374165e-06, "loss": 1.275, "step": 12547 }, { "epoch": 0.9603551201591918, "grad_norm": 0.6949603399279779, "learning_rate": 2.0330550631788135e-06, "loss": 1.5776, "step": 12548 }, { "epoch": 0.960431654676259, "grad_norm": 0.4872271196104132, "learning_rate": 2.0329279809611667e-06, "loss": 1.0716, "step": 12549 }, { "epoch": 0.9605081891933261, "grad_norm": 0.7592016224462363, "learning_rate": 2.032801142492284e-06, "loss": 1.6765, "step": 12550 }, { "epoch": 0.9605847237103934, "grad_norm": 0.7083400603055813, "learning_rate": 2.0326745477799582e-06, "loss": 1.6713, "step": 12551 }, { "epoch": 0.9606612582274606, "grad_norm": 0.5901566939222682, "learning_rate": 2.032548196831969e-06, "loss": 1.3447, "step": 12552 }, { "epoch": 0.9607377927445278, "grad_norm": 0.6170893708368447, "learning_rate": 2.0324220896560785e-06, "loss": 1.3541, "step": 12553 }, { "epoch": 0.960814327261595, "grad_norm": 0.4953344958506863, "learning_rate": 2.032296226260036e-06, "loss": 0.9963, "step": 12554 }, { "epoch": 0.9608908617786622, "grad_norm": 0.6130893411623299, "learning_rate": 2.0321706066515743e-06, "loss": 1.5225, "step": 12555 }, { "epoch": 0.9609673962957294, "grad_norm": 0.522977942148194, "learning_rate": 2.0320452308384116e-06, "loss": 1.1291, "step": 12556 }, { "epoch": 0.9610439308127966, "grad_norm": 0.5967716711807887, "learning_rate": 2.0319200988282524e-06, "loss": 1.4233, "step": 12557 }, { "epoch": 0.9611204653298637, "grad_norm": 0.5827773804641941, "learning_rate": 2.031795210628785e-06, "loss": 1.4673, "step": 12558 }, { "epoch": 0.961196999846931, "grad_norm": 0.5803532145284214, "learning_rate": 2.031670566247682e-06, "loss": 1.3987, "step": 12559 }, { "epoch": 0.9612735343639982, "grad_norm": 0.8341387030993952, "learning_rate": 2.0315461656926023e-06, "loss": 1.4176, "step": 12560 }, { "epoch": 0.9613500688810653, "grad_norm": 0.7814967880575964, "learning_rate": 2.03142200897119e-06, "loss": 1.669, "step": 12561 }, { "epoch": 0.9614266033981326, "grad_norm": 0.59267052446662, "learning_rate": 2.031298096091073e-06, "loss": 1.3373, "step": 12562 }, { "epoch": 0.9615031379151997, "grad_norm": 0.5759560203558132, "learning_rate": 2.031174427059865e-06, "loss": 1.4513, "step": 12563 }, { "epoch": 0.9615796724322669, "grad_norm": 0.5017283716946357, "learning_rate": 2.0310510018851655e-06, "loss": 1.0497, "step": 12564 }, { "epoch": 0.9616562069493342, "grad_norm": 0.7554760316483972, "learning_rate": 2.030927820574557e-06, "loss": 1.5669, "step": 12565 }, { "epoch": 0.9617327414664013, "grad_norm": 0.6419675166058803, "learning_rate": 2.030804883135608e-06, "loss": 1.5612, "step": 12566 }, { "epoch": 0.9618092759834685, "grad_norm": 0.7346818147377874, "learning_rate": 2.030682189575873e-06, "loss": 1.258, "step": 12567 }, { "epoch": 0.9618858105005358, "grad_norm": 0.6584615503086638, "learning_rate": 2.0305597399028904e-06, "loss": 1.5152, "step": 12568 }, { "epoch": 0.9619623450176029, "grad_norm": 0.5806103230104719, "learning_rate": 2.030437534124183e-06, "loss": 1.126, "step": 12569 }, { "epoch": 0.9620388795346702, "grad_norm": 0.73684965706359, "learning_rate": 2.030315572247261e-06, "loss": 1.6605, "step": 12570 }, { "epoch": 0.9621154140517373, "grad_norm": 0.6356699847370669, "learning_rate": 2.030193854279617e-06, "loss": 1.398, "step": 12571 }, { "epoch": 0.9621919485688045, "grad_norm": 0.5907102298848725, "learning_rate": 2.0300723802287303e-06, "loss": 1.3548, "step": 12572 }, { "epoch": 0.9622684830858718, "grad_norm": 0.5579743298739478, "learning_rate": 2.029951150102064e-06, "loss": 1.2441, "step": 12573 }, { "epoch": 0.9623450176029389, "grad_norm": 0.6825578612415032, "learning_rate": 2.0298301639070675e-06, "loss": 1.673, "step": 12574 }, { "epoch": 0.9624215521200061, "grad_norm": 0.5470282612211448, "learning_rate": 2.0297094216511744e-06, "loss": 1.2412, "step": 12575 }, { "epoch": 0.9624980866370734, "grad_norm": 0.8663906564696594, "learning_rate": 2.0295889233418034e-06, "loss": 1.6384, "step": 12576 }, { "epoch": 0.9625746211541405, "grad_norm": 0.6738970006138636, "learning_rate": 2.029468668986358e-06, "loss": 1.3441, "step": 12577 }, { "epoch": 0.9626511556712077, "grad_norm": 1.642067953421314, "learning_rate": 2.029348658592227e-06, "loss": 1.6889, "step": 12578 }, { "epoch": 0.9627276901882749, "grad_norm": 0.7725798500496982, "learning_rate": 2.0292288921667848e-06, "loss": 1.2012, "step": 12579 }, { "epoch": 0.9628042247053421, "grad_norm": 0.6159540263532012, "learning_rate": 2.0291093697173897e-06, "loss": 1.4788, "step": 12580 }, { "epoch": 0.9628807592224093, "grad_norm": 0.6272085209009398, "learning_rate": 2.0289900912513855e-06, "loss": 1.5246, "step": 12581 }, { "epoch": 0.9629572937394765, "grad_norm": 0.5807819345695983, "learning_rate": 2.0288710567761014e-06, "loss": 1.4383, "step": 12582 }, { "epoch": 0.9630338282565437, "grad_norm": 0.5785386495616764, "learning_rate": 2.0287522662988504e-06, "loss": 1.3363, "step": 12583 }, { "epoch": 0.963110362773611, "grad_norm": 0.6766481455420502, "learning_rate": 2.0286337198269324e-06, "loss": 1.6632, "step": 12584 }, { "epoch": 0.9631868972906781, "grad_norm": 0.6113852321549694, "learning_rate": 2.028515417367631e-06, "loss": 1.5395, "step": 12585 }, { "epoch": 0.9632634318077453, "grad_norm": 0.7942384508983621, "learning_rate": 2.0283973589282146e-06, "loss": 1.7515, "step": 12586 }, { "epoch": 0.9633399663248124, "grad_norm": 0.819106163983628, "learning_rate": 2.028279544515937e-06, "loss": 1.6356, "step": 12587 }, { "epoch": 0.9634165008418797, "grad_norm": 0.5313975862125939, "learning_rate": 2.028161974138038e-06, "loss": 0.9961, "step": 12588 }, { "epoch": 0.9634930353589469, "grad_norm": 0.6224779099195636, "learning_rate": 2.02804464780174e-06, "loss": 1.5454, "step": 12589 }, { "epoch": 0.963569569876014, "grad_norm": 0.5712940702829938, "learning_rate": 2.0279275655142532e-06, "loss": 1.4304, "step": 12590 }, { "epoch": 0.9636461043930813, "grad_norm": 0.5285505168513364, "learning_rate": 2.027810727282771e-06, "loss": 1.2699, "step": 12591 }, { "epoch": 0.9637226389101485, "grad_norm": 0.5542594042285752, "learning_rate": 2.027694133114472e-06, "loss": 1.3122, "step": 12592 }, { "epoch": 0.9637991734272157, "grad_norm": 0.9086846821872873, "learning_rate": 2.027577783016521e-06, "loss": 1.4793, "step": 12593 }, { "epoch": 0.9638757079442829, "grad_norm": 0.6233733922225988, "learning_rate": 2.0274616769960662e-06, "loss": 1.5295, "step": 12594 }, { "epoch": 0.96395224246135, "grad_norm": 0.6772992894402061, "learning_rate": 2.027345815060241e-06, "loss": 1.4952, "step": 12595 }, { "epoch": 0.9640287769784173, "grad_norm": 0.5842753145630744, "learning_rate": 2.0272301972161654e-06, "loss": 1.3624, "step": 12596 }, { "epoch": 0.9641053114954845, "grad_norm": 0.6601920574486035, "learning_rate": 2.0271148234709424e-06, "loss": 1.5754, "step": 12597 }, { "epoch": 0.9641818460125516, "grad_norm": 0.5777323735546731, "learning_rate": 2.0269996938316617e-06, "loss": 1.2533, "step": 12598 }, { "epoch": 0.9642583805296189, "grad_norm": 0.5471790148312217, "learning_rate": 2.026884808305397e-06, "loss": 1.3085, "step": 12599 }, { "epoch": 0.964334915046686, "grad_norm": 0.6203829943765715, "learning_rate": 2.0267701668992066e-06, "loss": 1.5971, "step": 12600 }, { "epoch": 0.9644114495637532, "grad_norm": 0.6110643851525519, "learning_rate": 2.026655769620135e-06, "loss": 1.3532, "step": 12601 }, { "epoch": 0.9644879840808205, "grad_norm": 0.6113298666115056, "learning_rate": 2.0265416164752107e-06, "loss": 1.3353, "step": 12602 }, { "epoch": 0.9645645185978876, "grad_norm": 0.5933205793461814, "learning_rate": 2.026427707471449e-06, "loss": 1.4309, "step": 12603 }, { "epoch": 0.9646410531149548, "grad_norm": 0.5857370187720654, "learning_rate": 2.026314042615846e-06, "loss": 1.258, "step": 12604 }, { "epoch": 0.9647175876320221, "grad_norm": 0.6231431682173854, "learning_rate": 2.0262006219153883e-06, "loss": 1.6155, "step": 12605 }, { "epoch": 0.9647941221490892, "grad_norm": 0.9913205063846405, "learning_rate": 2.0260874453770436e-06, "loss": 1.5541, "step": 12606 }, { "epoch": 0.9648706566661565, "grad_norm": 0.5989393717720021, "learning_rate": 2.0259745130077655e-06, "loss": 1.252, "step": 12607 }, { "epoch": 0.9649471911832236, "grad_norm": 0.6126892536854319, "learning_rate": 2.025861824814494e-06, "loss": 1.5302, "step": 12608 }, { "epoch": 0.9650237257002908, "grad_norm": 0.5787774653749391, "learning_rate": 2.0257493808041527e-06, "loss": 1.3788, "step": 12609 }, { "epoch": 0.9651002602173581, "grad_norm": 0.5776235781919207, "learning_rate": 2.0256371809836495e-06, "loss": 1.2712, "step": 12610 }, { "epoch": 0.9651767947344252, "grad_norm": 0.5180187457244794, "learning_rate": 2.025525225359879e-06, "loss": 1.1352, "step": 12611 }, { "epoch": 0.9652533292514924, "grad_norm": 0.716789172201739, "learning_rate": 2.0254135139397203e-06, "loss": 1.4296, "step": 12612 }, { "epoch": 0.9653298637685597, "grad_norm": 0.7149201331107015, "learning_rate": 2.0253020467300373e-06, "loss": 1.4054, "step": 12613 }, { "epoch": 0.9654063982856268, "grad_norm": 0.8899788674638932, "learning_rate": 2.0251908237376784e-06, "loss": 1.6265, "step": 12614 }, { "epoch": 0.965482932802694, "grad_norm": 0.7936709364570095, "learning_rate": 2.025079844969477e-06, "loss": 1.5449, "step": 12615 }, { "epoch": 0.9655594673197612, "grad_norm": 0.5229439592734639, "learning_rate": 2.0249691104322537e-06, "loss": 1.2735, "step": 12616 }, { "epoch": 0.9656360018368284, "grad_norm": 0.6543431330606234, "learning_rate": 2.024858620132811e-06, "loss": 1.3193, "step": 12617 }, { "epoch": 0.9657125363538956, "grad_norm": 0.59284335598009, "learning_rate": 2.0247483740779383e-06, "loss": 1.51, "step": 12618 }, { "epoch": 0.9657890708709628, "grad_norm": 0.5912635644492041, "learning_rate": 2.024638372274409e-06, "loss": 1.4228, "step": 12619 }, { "epoch": 0.96586560538803, "grad_norm": 0.5100465765690876, "learning_rate": 2.024528614728982e-06, "loss": 1.0276, "step": 12620 }, { "epoch": 0.9659421399050973, "grad_norm": 0.6939718646003379, "learning_rate": 2.0244191014484012e-06, "loss": 1.6009, "step": 12621 }, { "epoch": 0.9660186744221644, "grad_norm": 0.6762019281531224, "learning_rate": 2.0243098324393957e-06, "loss": 1.4113, "step": 12622 }, { "epoch": 0.9660952089392316, "grad_norm": 0.46355998221936456, "learning_rate": 2.0242008077086795e-06, "loss": 0.8886, "step": 12623 }, { "epoch": 0.9661717434562987, "grad_norm": 0.9070545949462645, "learning_rate": 2.024092027262951e-06, "loss": 1.5072, "step": 12624 }, { "epoch": 0.966248277973366, "grad_norm": 0.5873657463404146, "learning_rate": 2.0239834911088937e-06, "loss": 1.2801, "step": 12625 }, { "epoch": 0.9663248124904332, "grad_norm": 0.6203790565987203, "learning_rate": 2.023875199253177e-06, "loss": 1.0721, "step": 12626 }, { "epoch": 0.9664013470075004, "grad_norm": 0.656819362878654, "learning_rate": 2.023767151702454e-06, "loss": 1.715, "step": 12627 }, { "epoch": 0.9664778815245676, "grad_norm": 0.6921906428243418, "learning_rate": 2.0236593484633644e-06, "loss": 1.5304, "step": 12628 }, { "epoch": 0.9665544160416347, "grad_norm": 0.8138390771108401, "learning_rate": 2.0235517895425313e-06, "loss": 1.7919, "step": 12629 }, { "epoch": 0.966630950558702, "grad_norm": 0.5586844300906779, "learning_rate": 2.0234444749465633e-06, "loss": 1.4508, "step": 12630 }, { "epoch": 0.9667074850757692, "grad_norm": 0.6139846293241888, "learning_rate": 2.0233374046820546e-06, "loss": 1.3803, "step": 12631 }, { "epoch": 0.9667840195928363, "grad_norm": 0.5916192245927047, "learning_rate": 2.0232305787555836e-06, "loss": 1.3261, "step": 12632 }, { "epoch": 0.9668605541099036, "grad_norm": 0.5075778265217696, "learning_rate": 2.0231239971737145e-06, "loss": 1.0617, "step": 12633 }, { "epoch": 0.9669370886269708, "grad_norm": 0.5966629698914888, "learning_rate": 2.023017659942995e-06, "loss": 1.3644, "step": 12634 }, { "epoch": 0.9670136231440379, "grad_norm": 1.64484098643405, "learning_rate": 2.02291156706996e-06, "loss": 1.2037, "step": 12635 }, { "epoch": 0.9670901576611052, "grad_norm": 0.556434491477474, "learning_rate": 2.0228057185611273e-06, "loss": 1.4208, "step": 12636 }, { "epoch": 0.9671666921781723, "grad_norm": 0.9978016766890082, "learning_rate": 2.0227001144230005e-06, "loss": 1.4509, "step": 12637 }, { "epoch": 0.9672432266952395, "grad_norm": 0.545980226446758, "learning_rate": 2.022594754662069e-06, "loss": 1.1691, "step": 12638 }, { "epoch": 0.9673197612123068, "grad_norm": 0.581354700404191, "learning_rate": 2.0224896392848054e-06, "loss": 1.2974, "step": 12639 }, { "epoch": 0.9673962957293739, "grad_norm": 0.6281695320227758, "learning_rate": 2.0223847682976695e-06, "loss": 1.4462, "step": 12640 }, { "epoch": 0.9674728302464412, "grad_norm": 0.5813269188888023, "learning_rate": 2.0222801417071043e-06, "loss": 1.2861, "step": 12641 }, { "epoch": 0.9675493647635084, "grad_norm": 0.6060943726909276, "learning_rate": 2.0221757595195375e-06, "loss": 1.4188, "step": 12642 }, { "epoch": 0.9676258992805755, "grad_norm": 0.6248811152070712, "learning_rate": 2.0220716217413837e-06, "loss": 1.5379, "step": 12643 }, { "epoch": 0.9677024337976428, "grad_norm": 0.5211648706124732, "learning_rate": 2.0219677283790414e-06, "loss": 1.1048, "step": 12644 }, { "epoch": 0.9677789683147099, "grad_norm": 0.6683184423211099, "learning_rate": 2.0218640794388943e-06, "loss": 1.5739, "step": 12645 }, { "epoch": 0.9678555028317771, "grad_norm": 0.5460430785678041, "learning_rate": 2.0217606749273098e-06, "loss": 1.1825, "step": 12646 }, { "epoch": 0.9679320373488444, "grad_norm": 0.6600913785443799, "learning_rate": 2.021657514850643e-06, "loss": 1.5348, "step": 12647 }, { "epoch": 0.9680085718659115, "grad_norm": 0.5468591710953098, "learning_rate": 2.0215545992152306e-06, "loss": 1.1137, "step": 12648 }, { "epoch": 0.9680851063829787, "grad_norm": 0.6203463270136902, "learning_rate": 2.021451928027397e-06, "loss": 1.3642, "step": 12649 }, { "epoch": 0.968161640900046, "grad_norm": 0.6626489485058219, "learning_rate": 2.021349501293451e-06, "loss": 1.4895, "step": 12650 }, { "epoch": 0.9682381754171131, "grad_norm": 0.6286940160293969, "learning_rate": 2.021247319019685e-06, "loss": 1.1373, "step": 12651 }, { "epoch": 0.9683147099341803, "grad_norm": 0.46107507696038497, "learning_rate": 2.021145381212379e-06, "loss": 0.9926, "step": 12652 }, { "epoch": 0.9683912444512475, "grad_norm": 1.0567692853190422, "learning_rate": 2.0210436878777938e-06, "loss": 1.7551, "step": 12653 }, { "epoch": 0.9684677789683147, "grad_norm": 0.6309002310610599, "learning_rate": 2.0209422390221804e-06, "loss": 1.2736, "step": 12654 }, { "epoch": 0.968544313485382, "grad_norm": 0.5654638384063263, "learning_rate": 2.020841034651771e-06, "loss": 1.0088, "step": 12655 }, { "epoch": 0.9686208480024491, "grad_norm": 0.5345265578763919, "learning_rate": 2.020740074772783e-06, "loss": 1.2094, "step": 12656 }, { "epoch": 0.9686973825195163, "grad_norm": 0.695146322091357, "learning_rate": 2.0206393593914206e-06, "loss": 1.6029, "step": 12657 }, { "epoch": 0.9687739170365836, "grad_norm": 0.4474013968712227, "learning_rate": 2.0205388885138724e-06, "loss": 0.8255, "step": 12658 }, { "epoch": 0.9688504515536507, "grad_norm": 0.9353398676647581, "learning_rate": 2.020438662146311e-06, "loss": 1.4515, "step": 12659 }, { "epoch": 0.9689269860707179, "grad_norm": 0.7770405471591006, "learning_rate": 2.020338680294895e-06, "loss": 1.2054, "step": 12660 }, { "epoch": 0.969003520587785, "grad_norm": 0.5670159722975221, "learning_rate": 2.0202389429657672e-06, "loss": 1.3951, "step": 12661 }, { "epoch": 0.9690800551048523, "grad_norm": 0.5803687138097727, "learning_rate": 2.0201394501650563e-06, "loss": 1.2022, "step": 12662 }, { "epoch": 0.9691565896219195, "grad_norm": 0.6741693162387311, "learning_rate": 2.020040201898875e-06, "loss": 1.4955, "step": 12663 }, { "epoch": 0.9692331241389867, "grad_norm": 0.4976416782651571, "learning_rate": 2.0199411981733212e-06, "loss": 1.0468, "step": 12664 }, { "epoch": 0.9693096586560539, "grad_norm": 0.5935186554554558, "learning_rate": 2.0198424389944782e-06, "loss": 1.3157, "step": 12665 }, { "epoch": 0.969386193173121, "grad_norm": 0.6149249622891474, "learning_rate": 2.019743924368415e-06, "loss": 1.5157, "step": 12666 }, { "epoch": 0.9694627276901883, "grad_norm": 0.550736267592258, "learning_rate": 2.0196456543011832e-06, "loss": 1.3238, "step": 12667 }, { "epoch": 0.9695392622072555, "grad_norm": 0.5357914406745619, "learning_rate": 2.019547628798821e-06, "loss": 1.2809, "step": 12668 }, { "epoch": 0.9696157967243226, "grad_norm": 0.625854131252981, "learning_rate": 2.0194498478673526e-06, "loss": 1.4008, "step": 12669 }, { "epoch": 0.9696923312413899, "grad_norm": 0.8044571684999682, "learning_rate": 2.0193523115127844e-06, "loss": 1.4502, "step": 12670 }, { "epoch": 0.9697688657584571, "grad_norm": 0.5874734479325402, "learning_rate": 2.0192550197411105e-06, "loss": 1.3703, "step": 12671 }, { "epoch": 0.9698454002755242, "grad_norm": 0.5937195909081936, "learning_rate": 2.019157972558308e-06, "loss": 1.45, "step": 12672 }, { "epoch": 0.9699219347925915, "grad_norm": 0.5716771532134466, "learning_rate": 2.019061169970341e-06, "loss": 1.2722, "step": 12673 }, { "epoch": 0.9699984693096586, "grad_norm": 0.687263084927521, "learning_rate": 2.018964611983155e-06, "loss": 1.2955, "step": 12674 }, { "epoch": 0.9700750038267258, "grad_norm": 0.4783792621994175, "learning_rate": 2.0188682986026854e-06, "loss": 0.9469, "step": 12675 }, { "epoch": 0.9701515383437931, "grad_norm": 1.0562914121100573, "learning_rate": 2.018772229834849e-06, "loss": 1.4319, "step": 12676 }, { "epoch": 0.9702280728608602, "grad_norm": 0.5931664264721547, "learning_rate": 2.0186764056855475e-06, "loss": 1.5425, "step": 12677 }, { "epoch": 0.9703046073779275, "grad_norm": 0.8666445016119863, "learning_rate": 2.01858082616067e-06, "loss": 1.4865, "step": 12678 }, { "epoch": 0.9703811418949947, "grad_norm": 0.6140281888740875, "learning_rate": 2.018485491266089e-06, "loss": 1.378, "step": 12679 }, { "epoch": 0.9704576764120618, "grad_norm": 0.9840067020180392, "learning_rate": 2.0183904010076612e-06, "loss": 2.0441, "step": 12680 }, { "epoch": 0.9705342109291291, "grad_norm": 0.6001775023784707, "learning_rate": 2.01829555539123e-06, "loss": 1.468, "step": 12681 }, { "epoch": 0.9706107454461962, "grad_norm": 0.765040709996283, "learning_rate": 2.018200954422623e-06, "loss": 1.881, "step": 12682 }, { "epoch": 0.9706872799632634, "grad_norm": 0.550596664667728, "learning_rate": 2.018106598107653e-06, "loss": 1.1426, "step": 12683 }, { "epoch": 0.9707638144803307, "grad_norm": 0.5600945407211506, "learning_rate": 2.0180124864521176e-06, "loss": 1.3764, "step": 12684 }, { "epoch": 0.9708403489973978, "grad_norm": 0.6722636401389314, "learning_rate": 2.017918619461798e-06, "loss": 1.4173, "step": 12685 }, { "epoch": 0.970916883514465, "grad_norm": 0.7359781443361445, "learning_rate": 2.0178249971424632e-06, "loss": 1.5725, "step": 12686 }, { "epoch": 0.9709934180315323, "grad_norm": 0.5291303455904031, "learning_rate": 2.017731619499865e-06, "loss": 1.2308, "step": 12687 }, { "epoch": 0.9710699525485994, "grad_norm": 0.5650542017139786, "learning_rate": 2.0176384865397404e-06, "loss": 1.331, "step": 12688 }, { "epoch": 0.9711464870656666, "grad_norm": 0.6192989932392869, "learning_rate": 2.0175455982678125e-06, "loss": 1.3555, "step": 12689 }, { "epoch": 0.9712230215827338, "grad_norm": 0.5114335428102721, "learning_rate": 2.0174529546897876e-06, "loss": 1.0475, "step": 12690 }, { "epoch": 0.971299556099801, "grad_norm": 0.6982796261214581, "learning_rate": 2.0173605558113594e-06, "loss": 1.6331, "step": 12691 }, { "epoch": 0.9713760906168682, "grad_norm": 0.6439766406620502, "learning_rate": 2.017268401638205e-06, "loss": 1.7578, "step": 12692 }, { "epoch": 0.9714526251339354, "grad_norm": 0.6271268462087047, "learning_rate": 2.017176492175985e-06, "loss": 1.5107, "step": 12693 }, { "epoch": 0.9715291596510026, "grad_norm": 0.6131076448813992, "learning_rate": 2.0170848274303476e-06, "loss": 1.5277, "step": 12694 }, { "epoch": 0.9716056941680697, "grad_norm": 0.5803660215184503, "learning_rate": 2.0169934074069263e-06, "loss": 1.3643, "step": 12695 }, { "epoch": 0.971682228685137, "grad_norm": 0.7318182957480669, "learning_rate": 2.016902232111336e-06, "loss": 1.3936, "step": 12696 }, { "epoch": 0.9717587632022042, "grad_norm": 0.7665848443114899, "learning_rate": 2.01681130154918e-06, "loss": 1.7753, "step": 12697 }, { "epoch": 0.9718352977192714, "grad_norm": 0.5425109711477031, "learning_rate": 2.0167206157260453e-06, "loss": 1.2999, "step": 12698 }, { "epoch": 0.9719118322363386, "grad_norm": 0.693639056417406, "learning_rate": 2.0166301746475036e-06, "loss": 1.3279, "step": 12699 }, { "epoch": 0.9719883667534058, "grad_norm": 0.5651905738837837, "learning_rate": 2.0165399783191117e-06, "loss": 1.3519, "step": 12700 }, { "epoch": 0.972064901270473, "grad_norm": 0.6405167008547767, "learning_rate": 2.016450026746412e-06, "loss": 1.5627, "step": 12701 }, { "epoch": 0.9721414357875402, "grad_norm": 0.7903063991305396, "learning_rate": 2.0163603199349317e-06, "loss": 1.5638, "step": 12702 }, { "epoch": 0.9722179703046073, "grad_norm": 0.6311137100892326, "learning_rate": 2.016270857890181e-06, "loss": 1.5254, "step": 12703 }, { "epoch": 0.9722945048216746, "grad_norm": 0.5476774402700787, "learning_rate": 2.016181640617659e-06, "loss": 1.3934, "step": 12704 }, { "epoch": 0.9723710393387418, "grad_norm": 0.6366953848126552, "learning_rate": 2.016092668122846e-06, "loss": 1.4669, "step": 12705 }, { "epoch": 0.9724475738558089, "grad_norm": 0.6129158519761218, "learning_rate": 2.016003940411209e-06, "loss": 1.4467, "step": 12706 }, { "epoch": 0.9725241083728762, "grad_norm": 0.5492754225892837, "learning_rate": 2.0159154574882e-06, "loss": 1.3861, "step": 12707 }, { "epoch": 0.9726006428899434, "grad_norm": 0.7060048452351775, "learning_rate": 2.0158272193592546e-06, "loss": 1.5455, "step": 12708 }, { "epoch": 0.9726771774070105, "grad_norm": 0.7065611968551766, "learning_rate": 2.0157392260297963e-06, "loss": 1.7667, "step": 12709 }, { "epoch": 0.9727537119240778, "grad_norm": 0.5931620864298229, "learning_rate": 2.0156514775052303e-06, "loss": 1.4248, "step": 12710 }, { "epoch": 0.9728302464411449, "grad_norm": 0.5375961391824945, "learning_rate": 2.015563973790948e-06, "loss": 1.1616, "step": 12711 }, { "epoch": 0.9729067809582121, "grad_norm": 0.5456980881487957, "learning_rate": 2.015476714892327e-06, "loss": 1.1401, "step": 12712 }, { "epoch": 0.9729833154752794, "grad_norm": 0.5584633488210133, "learning_rate": 2.0153897008147277e-06, "loss": 1.2902, "step": 12713 }, { "epoch": 0.9730598499923465, "grad_norm": 0.5620402004190155, "learning_rate": 2.015302931563497e-06, "loss": 1.2252, "step": 12714 }, { "epoch": 0.9731363845094138, "grad_norm": 0.9701131077679246, "learning_rate": 2.015216407143966e-06, "loss": 1.7543, "step": 12715 }, { "epoch": 0.973212919026481, "grad_norm": 0.5669243852558272, "learning_rate": 2.0151301275614512e-06, "loss": 1.361, "step": 12716 }, { "epoch": 0.9732894535435481, "grad_norm": 0.44387937623507284, "learning_rate": 2.0150440928212544e-06, "loss": 0.8953, "step": 12717 }, { "epoch": 0.9733659880606154, "grad_norm": 0.5921196760790921, "learning_rate": 2.014958302928661e-06, "loss": 1.3292, "step": 12718 }, { "epoch": 0.9734425225776825, "grad_norm": 0.6419423865235815, "learning_rate": 2.0148727578889426e-06, "loss": 1.7126, "step": 12719 }, { "epoch": 0.9735190570947497, "grad_norm": 0.5129254207382056, "learning_rate": 2.014787457707354e-06, "loss": 1.0179, "step": 12720 }, { "epoch": 0.973595591611817, "grad_norm": 0.486757092219783, "learning_rate": 2.014702402389139e-06, "loss": 1.0033, "step": 12721 }, { "epoch": 0.9736721261288841, "grad_norm": 0.6245386010018, "learning_rate": 2.014617591939522e-06, "loss": 1.2655, "step": 12722 }, { "epoch": 0.9737486606459513, "grad_norm": 0.5893840890503176, "learning_rate": 2.0145330263637135e-06, "loss": 1.4215, "step": 12723 }, { "epoch": 0.9738251951630186, "grad_norm": 0.6528640747211251, "learning_rate": 2.0144487056669102e-06, "loss": 1.5581, "step": 12724 }, { "epoch": 0.9739017296800857, "grad_norm": 0.5973006908424259, "learning_rate": 2.0143646298542933e-06, "loss": 1.4364, "step": 12725 }, { "epoch": 0.9739782641971529, "grad_norm": 0.5109491119616176, "learning_rate": 2.0142807989310285e-06, "loss": 1.1739, "step": 12726 }, { "epoch": 0.9740547987142201, "grad_norm": 0.7601607682565568, "learning_rate": 2.0141972129022665e-06, "loss": 1.489, "step": 12727 }, { "epoch": 0.9741313332312873, "grad_norm": 0.5606371461459928, "learning_rate": 2.014113871773142e-06, "loss": 1.1989, "step": 12728 }, { "epoch": 0.9742078677483546, "grad_norm": 0.5651884633809977, "learning_rate": 2.0140307755487774e-06, "loss": 1.2878, "step": 12729 }, { "epoch": 0.9742844022654217, "grad_norm": 0.8143905790531258, "learning_rate": 2.013947924234278e-06, "loss": 1.3878, "step": 12730 }, { "epoch": 0.9743609367824889, "grad_norm": 1.0019411313688928, "learning_rate": 2.0138653178347342e-06, "loss": 1.9276, "step": 12731 }, { "epoch": 0.974437471299556, "grad_norm": 0.5455745897622685, "learning_rate": 2.013782956355221e-06, "loss": 1.129, "step": 12732 }, { "epoch": 0.9745140058166233, "grad_norm": 0.5711335765745514, "learning_rate": 2.0137008398008003e-06, "loss": 1.4433, "step": 12733 }, { "epoch": 0.9745905403336905, "grad_norm": 0.6190316941086579, "learning_rate": 2.013618968176516e-06, "loss": 1.2844, "step": 12734 }, { "epoch": 0.9746670748507577, "grad_norm": 0.6121741112748055, "learning_rate": 2.0135373414874e-06, "loss": 1.5701, "step": 12735 }, { "epoch": 0.9747436093678249, "grad_norm": 0.5956596275120006, "learning_rate": 2.013455959738466e-06, "loss": 1.4072, "step": 12736 }, { "epoch": 0.9748201438848921, "grad_norm": 0.5572598816427277, "learning_rate": 2.013374822934716e-06, "loss": 1.2701, "step": 12737 }, { "epoch": 0.9748966784019593, "grad_norm": 0.5943883600913117, "learning_rate": 2.0132939310811343e-06, "loss": 1.4447, "step": 12738 }, { "epoch": 0.9749732129190265, "grad_norm": 0.9970639837709627, "learning_rate": 2.0132132841826923e-06, "loss": 1.4658, "step": 12739 }, { "epoch": 0.9750497474360936, "grad_norm": 0.641315505971041, "learning_rate": 2.0131328822443435e-06, "loss": 1.482, "step": 12740 }, { "epoch": 0.9751262819531609, "grad_norm": 0.5529531775792694, "learning_rate": 2.013052725271029e-06, "loss": 1.2237, "step": 12741 }, { "epoch": 0.9752028164702281, "grad_norm": 0.6300458005953556, "learning_rate": 2.012972813267674e-06, "loss": 1.4698, "step": 12742 }, { "epoch": 0.9752793509872952, "grad_norm": 0.5726322351376061, "learning_rate": 2.012893146239188e-06, "loss": 1.4037, "step": 12743 }, { "epoch": 0.9753558855043625, "grad_norm": 0.5986253087251795, "learning_rate": 2.012813724190466e-06, "loss": 1.16, "step": 12744 }, { "epoch": 0.9754324200214297, "grad_norm": 0.6421580620005376, "learning_rate": 2.012734547126389e-06, "loss": 1.5724, "step": 12745 }, { "epoch": 0.9755089545384968, "grad_norm": 0.6766911670717521, "learning_rate": 2.0126556150518205e-06, "loss": 1.6515, "step": 12746 }, { "epoch": 0.9755854890555641, "grad_norm": 0.48943067105715693, "learning_rate": 2.012576927971611e-06, "loss": 0.9478, "step": 12747 }, { "epoch": 0.9756620235726312, "grad_norm": 0.5786495535247197, "learning_rate": 2.012498485890596e-06, "loss": 1.2301, "step": 12748 }, { "epoch": 0.9757385580896984, "grad_norm": 0.6193683227592057, "learning_rate": 2.0124202888135933e-06, "loss": 1.5276, "step": 12749 }, { "epoch": 0.9758150926067657, "grad_norm": 0.8906596469710879, "learning_rate": 2.0123423367454094e-06, "loss": 1.5888, "step": 12750 }, { "epoch": 0.9758916271238328, "grad_norm": 0.5093570320562814, "learning_rate": 2.0122646296908326e-06, "loss": 1.2068, "step": 12751 }, { "epoch": 0.9759681616409001, "grad_norm": 0.5766566450886513, "learning_rate": 2.0121871676546376e-06, "loss": 1.2772, "step": 12752 }, { "epoch": 0.9760446961579673, "grad_norm": 0.5371962983633722, "learning_rate": 2.012109950641585e-06, "loss": 1.3016, "step": 12753 }, { "epoch": 0.9761212306750344, "grad_norm": 0.7316364454440586, "learning_rate": 2.0120329786564187e-06, "loss": 1.36, "step": 12754 }, { "epoch": 0.9761977651921017, "grad_norm": 0.522681839491357, "learning_rate": 2.0119562517038676e-06, "loss": 1.1124, "step": 12755 }, { "epoch": 0.9762742997091688, "grad_norm": 0.5463578780922873, "learning_rate": 2.0118797697886465e-06, "loss": 1.3014, "step": 12756 }, { "epoch": 0.976350834226236, "grad_norm": 1.120380910771984, "learning_rate": 2.0118035329154546e-06, "loss": 1.9206, "step": 12757 }, { "epoch": 0.9764273687433033, "grad_norm": 0.5896196350869969, "learning_rate": 2.0117275410889757e-06, "loss": 1.3171, "step": 12758 }, { "epoch": 0.9765039032603704, "grad_norm": 0.7108161357584198, "learning_rate": 2.0116517943138795e-06, "loss": 1.8344, "step": 12759 }, { "epoch": 0.9765804377774376, "grad_norm": 0.7011136318336519, "learning_rate": 2.0115762925948207e-06, "loss": 1.5387, "step": 12760 }, { "epoch": 0.9766569722945049, "grad_norm": 0.5686646953858726, "learning_rate": 2.011501035936437e-06, "loss": 1.3221, "step": 12761 }, { "epoch": 0.976733506811572, "grad_norm": 0.5537066056570545, "learning_rate": 2.0114260243433527e-06, "loss": 1.2636, "step": 12762 }, { "epoch": 0.9768100413286392, "grad_norm": 0.6724447053624169, "learning_rate": 2.0113512578201774e-06, "loss": 1.6311, "step": 12763 }, { "epoch": 0.9768865758457064, "grad_norm": 0.5294989181091281, "learning_rate": 2.011276736371505e-06, "loss": 1.2331, "step": 12764 }, { "epoch": 0.9769631103627736, "grad_norm": 0.5609337693023323, "learning_rate": 2.011202460001913e-06, "loss": 1.2683, "step": 12765 }, { "epoch": 0.9770396448798409, "grad_norm": 0.5666478920063635, "learning_rate": 2.0111284287159668e-06, "loss": 1.1043, "step": 12766 }, { "epoch": 0.977116179396908, "grad_norm": 0.5470110833740155, "learning_rate": 2.011054642518214e-06, "loss": 1.2357, "step": 12767 }, { "epoch": 0.9771927139139752, "grad_norm": 0.5592499424851226, "learning_rate": 2.0109811014131884e-06, "loss": 1.2629, "step": 12768 }, { "epoch": 0.9772692484310423, "grad_norm": 0.6198502133236853, "learning_rate": 2.0109078054054094e-06, "loss": 1.503, "step": 12769 }, { "epoch": 0.9773457829481096, "grad_norm": 0.8414245896078093, "learning_rate": 2.0108347544993797e-06, "loss": 1.8176, "step": 12770 }, { "epoch": 0.9774223174651768, "grad_norm": 1.4579199010484054, "learning_rate": 2.010761948699588e-06, "loss": 1.9611, "step": 12771 }, { "epoch": 0.977498851982244, "grad_norm": 0.6237568118035556, "learning_rate": 2.010689388010508e-06, "loss": 1.425, "step": 12772 }, { "epoch": 0.9775753864993112, "grad_norm": 0.9166328591795662, "learning_rate": 2.0106170724365976e-06, "loss": 1.3593, "step": 12773 }, { "epoch": 0.9776519210163784, "grad_norm": 0.6731949320065966, "learning_rate": 2.0105450019823e-06, "loss": 1.1886, "step": 12774 }, { "epoch": 0.9777284555334456, "grad_norm": 0.6713073084448073, "learning_rate": 2.010473176652044e-06, "loss": 1.4937, "step": 12775 }, { "epoch": 0.9778049900505128, "grad_norm": 0.5708738990339386, "learning_rate": 2.0104015964502426e-06, "loss": 1.2532, "step": 12776 }, { "epoch": 0.9778815245675799, "grad_norm": 0.6334471495510766, "learning_rate": 2.010330261381293e-06, "loss": 1.2286, "step": 12777 }, { "epoch": 0.9779580590846472, "grad_norm": 0.6175108769130332, "learning_rate": 2.01025917144958e-06, "loss": 1.3388, "step": 12778 }, { "epoch": 0.9780345936017144, "grad_norm": 0.5725108456520259, "learning_rate": 2.01018832665947e-06, "loss": 1.3465, "step": 12779 }, { "epoch": 0.9781111281187815, "grad_norm": 0.5385263055528285, "learning_rate": 2.010117727015316e-06, "loss": 1.1632, "step": 12780 }, { "epoch": 0.9781876626358488, "grad_norm": 0.6045103301447879, "learning_rate": 2.010047372521457e-06, "loss": 1.3098, "step": 12781 }, { "epoch": 0.978264197152916, "grad_norm": 0.5662077336357008, "learning_rate": 2.0099772631822154e-06, "loss": 1.1904, "step": 12782 }, { "epoch": 0.9783407316699831, "grad_norm": 0.6441520753221698, "learning_rate": 2.0099073990018976e-06, "loss": 1.3096, "step": 12783 }, { "epoch": 0.9784172661870504, "grad_norm": 0.6090976497667826, "learning_rate": 2.0098377799847983e-06, "loss": 1.4117, "step": 12784 }, { "epoch": 0.9784938007041175, "grad_norm": 0.8434317957006586, "learning_rate": 2.009768406135194e-06, "loss": 1.3854, "step": 12785 }, { "epoch": 0.9785703352211848, "grad_norm": 0.6934930349861788, "learning_rate": 2.0096992774573468e-06, "loss": 1.2834, "step": 12786 }, { "epoch": 0.978646869738252, "grad_norm": 0.6741191122339301, "learning_rate": 2.009630393955505e-06, "loss": 1.3733, "step": 12787 }, { "epoch": 0.9787234042553191, "grad_norm": 0.4871984740261329, "learning_rate": 2.0095617556339005e-06, "loss": 0.9303, "step": 12788 }, { "epoch": 0.9787999387723864, "grad_norm": 0.611731125816087, "learning_rate": 2.009493362496751e-06, "loss": 1.3718, "step": 12789 }, { "epoch": 0.9788764732894536, "grad_norm": 0.577372996171574, "learning_rate": 2.009425214548259e-06, "loss": 1.3891, "step": 12790 }, { "epoch": 0.9789530078065207, "grad_norm": 0.6056682902829671, "learning_rate": 2.009357311792611e-06, "loss": 1.3272, "step": 12791 }, { "epoch": 0.979029542323588, "grad_norm": 0.4712106756690492, "learning_rate": 2.00928965423398e-06, "loss": 0.9881, "step": 12792 }, { "epoch": 0.9791060768406551, "grad_norm": 0.5839697171222797, "learning_rate": 2.0092222418765223e-06, "loss": 1.2917, "step": 12793 }, { "epoch": 0.9791826113577223, "grad_norm": 0.6257080695214687, "learning_rate": 2.0091550747243797e-06, "loss": 1.2479, "step": 12794 }, { "epoch": 0.9792591458747896, "grad_norm": 0.7984223036223994, "learning_rate": 2.0090881527816807e-06, "loss": 1.5476, "step": 12795 }, { "epoch": 0.9793356803918567, "grad_norm": 0.5394308192360823, "learning_rate": 2.009021476052535e-06, "loss": 1.1596, "step": 12796 }, { "epoch": 0.9794122149089239, "grad_norm": 0.6252509028721155, "learning_rate": 2.008955044541041e-06, "loss": 1.5736, "step": 12797 }, { "epoch": 0.9794887494259911, "grad_norm": 0.528460487860595, "learning_rate": 2.0088888582512803e-06, "loss": 1.1578, "step": 12798 }, { "epoch": 0.9795652839430583, "grad_norm": 0.5157234298723982, "learning_rate": 2.0088229171873194e-06, "loss": 1.0101, "step": 12799 }, { "epoch": 0.9796418184601255, "grad_norm": 0.5951457090202011, "learning_rate": 2.0087572213532092e-06, "loss": 1.3363, "step": 12800 }, { "epoch": 0.9797183529771927, "grad_norm": 0.5848833868597006, "learning_rate": 2.0086917707529872e-06, "loss": 1.513, "step": 12801 }, { "epoch": 0.9797948874942599, "grad_norm": 0.5996629435886762, "learning_rate": 2.0086265653906747e-06, "loss": 1.3961, "step": 12802 }, { "epoch": 0.9798714220113272, "grad_norm": 0.6115786038556003, "learning_rate": 2.0085616052702777e-06, "loss": 1.1938, "step": 12803 }, { "epoch": 0.9799479565283943, "grad_norm": 0.5672884506814158, "learning_rate": 2.008496890395788e-06, "loss": 1.3416, "step": 12804 }, { "epoch": 0.9800244910454615, "grad_norm": 0.5731311836257776, "learning_rate": 2.0084324207711816e-06, "loss": 1.4831, "step": 12805 }, { "epoch": 0.9801010255625286, "grad_norm": 0.768141885036815, "learning_rate": 2.00836819640042e-06, "loss": 1.4151, "step": 12806 }, { "epoch": 0.9801775600795959, "grad_norm": 0.5773831451903458, "learning_rate": 2.0083042172874485e-06, "loss": 1.497, "step": 12807 }, { "epoch": 0.9802540945966631, "grad_norm": 0.4955793859371158, "learning_rate": 2.008240483436199e-06, "loss": 1.0733, "step": 12808 }, { "epoch": 0.9803306291137303, "grad_norm": 0.5902799446255941, "learning_rate": 2.008176994850588e-06, "loss": 1.5084, "step": 12809 }, { "epoch": 0.9804071636307975, "grad_norm": 0.5457100144316986, "learning_rate": 2.0081137515345154e-06, "loss": 1.2196, "step": 12810 }, { "epoch": 0.9804836981478647, "grad_norm": 0.5553582859300915, "learning_rate": 2.008050753491867e-06, "loss": 1.2375, "step": 12811 }, { "epoch": 0.9805602326649319, "grad_norm": 0.6916694373485788, "learning_rate": 2.007988000726514e-06, "loss": 1.7229, "step": 12812 }, { "epoch": 0.9806367671819991, "grad_norm": 0.5810990695598984, "learning_rate": 2.007925493242312e-06, "loss": 1.3971, "step": 12813 }, { "epoch": 0.9807133016990662, "grad_norm": 0.6685092051210569, "learning_rate": 2.007863231043102e-06, "loss": 1.7145, "step": 12814 }, { "epoch": 0.9807898362161335, "grad_norm": 0.5300041296568401, "learning_rate": 2.0078012141327093e-06, "loss": 1.1963, "step": 12815 }, { "epoch": 0.9808663707332007, "grad_norm": 0.5592571650384801, "learning_rate": 2.0077394425149448e-06, "loss": 1.3651, "step": 12816 }, { "epoch": 0.9809429052502678, "grad_norm": 0.6481220092201441, "learning_rate": 2.007677916193603e-06, "loss": 1.6512, "step": 12817 }, { "epoch": 0.9810194397673351, "grad_norm": 0.5674723952040949, "learning_rate": 2.0076166351724643e-06, "loss": 1.2711, "step": 12818 }, { "epoch": 0.9810959742844023, "grad_norm": 0.5966768504358213, "learning_rate": 2.007555599455295e-06, "loss": 1.3833, "step": 12819 }, { "epoch": 0.9811725088014694, "grad_norm": 0.5976681213407189, "learning_rate": 2.0074948090458456e-06, "loss": 1.4446, "step": 12820 }, { "epoch": 0.9812490433185367, "grad_norm": 0.4925982823852052, "learning_rate": 2.0074342639478493e-06, "loss": 1.0109, "step": 12821 }, { "epoch": 0.9813255778356038, "grad_norm": 0.559126583602027, "learning_rate": 2.007373964165028e-06, "loss": 1.2395, "step": 12822 }, { "epoch": 0.981402112352671, "grad_norm": 0.5621994217262686, "learning_rate": 2.0073139097010856e-06, "loss": 1.3, "step": 12823 }, { "epoch": 0.9814786468697383, "grad_norm": 0.6097636044785765, "learning_rate": 2.0072541005597128e-06, "loss": 1.492, "step": 12824 }, { "epoch": 0.9815551813868054, "grad_norm": 0.6517184090721705, "learning_rate": 2.0071945367445836e-06, "loss": 1.7266, "step": 12825 }, { "epoch": 0.9816317159038727, "grad_norm": 0.6452017962692822, "learning_rate": 2.007135218259359e-06, "loss": 1.5365, "step": 12826 }, { "epoch": 0.9817082504209399, "grad_norm": 0.5706670423243378, "learning_rate": 2.0070761451076824e-06, "loss": 1.3651, "step": 12827 }, { "epoch": 0.981784784938007, "grad_norm": 0.6673016248355, "learning_rate": 2.007017317293184e-06, "loss": 1.7001, "step": 12828 }, { "epoch": 0.9818613194550743, "grad_norm": 0.6080672672909275, "learning_rate": 2.0069587348194786e-06, "loss": 1.3499, "step": 12829 }, { "epoch": 0.9819378539721414, "grad_norm": 0.8286211804150263, "learning_rate": 2.0069003976901657e-06, "loss": 1.5831, "step": 12830 }, { "epoch": 0.9820143884892086, "grad_norm": 0.6718123534487274, "learning_rate": 2.006842305908829e-06, "loss": 1.4785, "step": 12831 }, { "epoch": 0.9820909230062759, "grad_norm": 0.5942236011540515, "learning_rate": 2.0067844594790383e-06, "loss": 1.2464, "step": 12832 }, { "epoch": 0.982167457523343, "grad_norm": 0.5722939220135045, "learning_rate": 2.0067268584043485e-06, "loss": 1.3458, "step": 12833 }, { "epoch": 0.9822439920404102, "grad_norm": 0.9101821195398886, "learning_rate": 2.0066695026882976e-06, "loss": 1.5459, "step": 12834 }, { "epoch": 0.9823205265574774, "grad_norm": 0.5778343960358845, "learning_rate": 2.00661239233441e-06, "loss": 1.3354, "step": 12835 }, { "epoch": 0.9823970610745446, "grad_norm": 0.5623166789058012, "learning_rate": 2.006555527346195e-06, "loss": 1.3326, "step": 12836 }, { "epoch": 0.9824735955916118, "grad_norm": 0.5772641706969432, "learning_rate": 2.0064989077271474e-06, "loss": 1.4411, "step": 12837 }, { "epoch": 0.982550130108679, "grad_norm": 0.6385442286393599, "learning_rate": 2.0064425334807445e-06, "loss": 1.3908, "step": 12838 }, { "epoch": 0.9826266646257462, "grad_norm": 0.5040994969452601, "learning_rate": 2.006386404610451e-06, "loss": 1.0394, "step": 12839 }, { "epoch": 0.9827031991428135, "grad_norm": 0.8004091148851903, "learning_rate": 2.0063305211197152e-06, "loss": 1.7707, "step": 12840 }, { "epoch": 0.9827797336598806, "grad_norm": 0.5141056485921808, "learning_rate": 2.006274883011971e-06, "loss": 1.2628, "step": 12841 }, { "epoch": 0.9828562681769478, "grad_norm": 0.6699219494043622, "learning_rate": 2.006219490290637e-06, "loss": 1.2192, "step": 12842 }, { "epoch": 0.982932802694015, "grad_norm": 0.51363391081007, "learning_rate": 2.006164342959117e-06, "loss": 1.127, "step": 12843 }, { "epoch": 0.9830093372110822, "grad_norm": 0.6219503405480319, "learning_rate": 2.006109441020799e-06, "loss": 1.3634, "step": 12844 }, { "epoch": 0.9830858717281494, "grad_norm": 0.5995195784866213, "learning_rate": 2.0060547844790556e-06, "loss": 1.2834, "step": 12845 }, { "epoch": 0.9831624062452166, "grad_norm": 0.6231680830423497, "learning_rate": 2.0060003733372465e-06, "loss": 1.5255, "step": 12846 }, { "epoch": 0.9832389407622838, "grad_norm": 0.5423738531410948, "learning_rate": 2.0059462075987144e-06, "loss": 1.3091, "step": 12847 }, { "epoch": 0.983315475279351, "grad_norm": 0.6711884363448423, "learning_rate": 2.0058922872667872e-06, "loss": 1.1275, "step": 12848 }, { "epoch": 0.9833920097964182, "grad_norm": 0.714391635179475, "learning_rate": 2.0058386123447778e-06, "loss": 1.4578, "step": 12849 }, { "epoch": 0.9834685443134854, "grad_norm": 0.5983104381285768, "learning_rate": 2.0057851828359844e-06, "loss": 1.3327, "step": 12850 }, { "epoch": 0.9835450788305525, "grad_norm": 0.5976375993115199, "learning_rate": 2.0057319987436894e-06, "loss": 1.3198, "step": 12851 }, { "epoch": 0.9836216133476198, "grad_norm": 0.600307959783823, "learning_rate": 2.0056790600711613e-06, "loss": 1.2682, "step": 12852 }, { "epoch": 0.983698147864687, "grad_norm": 0.4606607263262271, "learning_rate": 2.005626366821653e-06, "loss": 0.9077, "step": 12853 }, { "epoch": 0.9837746823817541, "grad_norm": 0.6823075551036633, "learning_rate": 2.005573918998401e-06, "loss": 1.5156, "step": 12854 }, { "epoch": 0.9838512168988214, "grad_norm": 0.5268979684420186, "learning_rate": 2.005521716604629e-06, "loss": 1.1858, "step": 12855 }, { "epoch": 0.9839277514158886, "grad_norm": 0.5618530498065, "learning_rate": 2.0054697596435432e-06, "loss": 1.1062, "step": 12856 }, { "epoch": 0.9840042859329557, "grad_norm": 0.5741173647061589, "learning_rate": 2.0054180481183374e-06, "loss": 1.1901, "step": 12857 }, { "epoch": 0.984080820450023, "grad_norm": 0.6266619260824011, "learning_rate": 2.0053665820321877e-06, "loss": 1.6603, "step": 12858 }, { "epoch": 0.9841573549670901, "grad_norm": 0.6509116701319567, "learning_rate": 2.0053153613882577e-06, "loss": 1.3793, "step": 12859 }, { "epoch": 0.9842338894841574, "grad_norm": 0.5563292982459035, "learning_rate": 2.005264386189693e-06, "loss": 1.2122, "step": 12860 }, { "epoch": 0.9843104240012246, "grad_norm": 0.5431900382231442, "learning_rate": 2.0052136564396265e-06, "loss": 1.1135, "step": 12861 }, { "epoch": 0.9843869585182917, "grad_norm": 0.5748096193258354, "learning_rate": 2.005163172141175e-06, "loss": 1.2116, "step": 12862 }, { "epoch": 0.984463493035359, "grad_norm": 0.623286647977024, "learning_rate": 2.0051129332974406e-06, "loss": 1.49, "step": 12863 }, { "epoch": 0.9845400275524261, "grad_norm": 0.5133928629882603, "learning_rate": 2.00506293991151e-06, "loss": 1.2235, "step": 12864 }, { "epoch": 0.9846165620694933, "grad_norm": 0.6196182265172655, "learning_rate": 2.0050131919864554e-06, "loss": 1.3945, "step": 12865 }, { "epoch": 0.9846930965865606, "grad_norm": 0.5871589333469133, "learning_rate": 2.0049636895253325e-06, "loss": 1.3121, "step": 12866 }, { "epoch": 0.9847696311036277, "grad_norm": 0.6794866994450874, "learning_rate": 2.0049144325311836e-06, "loss": 1.4895, "step": 12867 }, { "epoch": 0.9848461656206949, "grad_norm": 0.6231932494441415, "learning_rate": 2.004865421007035e-06, "loss": 1.26, "step": 12868 }, { "epoch": 0.9849227001377622, "grad_norm": 0.6001120847039604, "learning_rate": 2.004816654955898e-06, "loss": 1.2964, "step": 12869 }, { "epoch": 0.9849992346548293, "grad_norm": 0.6158677641153437, "learning_rate": 2.004768134380769e-06, "loss": 1.4223, "step": 12870 }, { "epoch": 0.9850757691718965, "grad_norm": 0.6829253082906915, "learning_rate": 2.00471985928463e-06, "loss": 1.7251, "step": 12871 }, { "epoch": 0.9851523036889637, "grad_norm": 0.6440561219310915, "learning_rate": 2.0046718296704455e-06, "loss": 1.6796, "step": 12872 }, { "epoch": 0.9852288382060309, "grad_norm": 1.4754217569973933, "learning_rate": 2.0046240455411678e-06, "loss": 1.5393, "step": 12873 }, { "epoch": 0.9853053727230981, "grad_norm": 0.575409550156121, "learning_rate": 2.004576506899733e-06, "loss": 1.3648, "step": 12874 }, { "epoch": 0.9853819072401653, "grad_norm": 0.6332210033618608, "learning_rate": 2.0045292137490613e-06, "loss": 1.433, "step": 12875 }, { "epoch": 0.9854584417572325, "grad_norm": 0.6014886335799557, "learning_rate": 2.0044821660920593e-06, "loss": 1.4578, "step": 12876 }, { "epoch": 0.9855349762742998, "grad_norm": 0.5618529334987871, "learning_rate": 2.004435363931617e-06, "loss": 1.3196, "step": 12877 }, { "epoch": 0.9856115107913669, "grad_norm": 0.6825534662125436, "learning_rate": 2.0043888072706105e-06, "loss": 1.3091, "step": 12878 }, { "epoch": 0.9856880453084341, "grad_norm": 0.5943134315136818, "learning_rate": 2.0043424961119003e-06, "loss": 1.3868, "step": 12879 }, { "epoch": 0.9857645798255013, "grad_norm": 0.5602628085635682, "learning_rate": 2.0042964304583316e-06, "loss": 1.2484, "step": 12880 }, { "epoch": 0.9858411143425685, "grad_norm": 0.6440555217088629, "learning_rate": 2.0042506103127354e-06, "loss": 1.5686, "step": 12881 }, { "epoch": 0.9859176488596357, "grad_norm": 0.9597871736397374, "learning_rate": 2.0042050356779266e-06, "loss": 1.7027, "step": 12882 }, { "epoch": 0.9859941833767029, "grad_norm": 0.6180885866152941, "learning_rate": 2.0041597065567056e-06, "loss": 1.2974, "step": 12883 }, { "epoch": 0.9860707178937701, "grad_norm": 0.5481317848639905, "learning_rate": 2.004114622951858e-06, "loss": 1.2023, "step": 12884 }, { "epoch": 0.9861472524108373, "grad_norm": 0.6427062728960972, "learning_rate": 2.0040697848661527e-06, "loss": 1.4658, "step": 12885 }, { "epoch": 0.9862237869279045, "grad_norm": 0.6071272125600197, "learning_rate": 2.0040251923023456e-06, "loss": 1.4826, "step": 12886 }, { "epoch": 0.9863003214449717, "grad_norm": 0.6203560000194823, "learning_rate": 2.003980845263176e-06, "loss": 1.3639, "step": 12887 }, { "epoch": 0.9863768559620388, "grad_norm": 0.7968301398561867, "learning_rate": 2.00393674375137e-06, "loss": 1.6489, "step": 12888 }, { "epoch": 0.9864533904791061, "grad_norm": 0.5967155106250451, "learning_rate": 2.003892887769635e-06, "loss": 1.3876, "step": 12889 }, { "epoch": 0.9865299249961733, "grad_norm": 0.6210822279184204, "learning_rate": 2.003849277320668e-06, "loss": 1.4926, "step": 12890 }, { "epoch": 0.9866064595132404, "grad_norm": 0.7218558103166355, "learning_rate": 2.003805912407148e-06, "loss": 1.592, "step": 12891 }, { "epoch": 0.9866829940303077, "grad_norm": 0.6007614947396186, "learning_rate": 2.0037627930317385e-06, "loss": 1.3643, "step": 12892 }, { "epoch": 0.9867595285473749, "grad_norm": 0.6153913865601355, "learning_rate": 2.00371991919709e-06, "loss": 1.4138, "step": 12893 }, { "epoch": 0.986836063064442, "grad_norm": 0.6221382050165924, "learning_rate": 2.0036772909058356e-06, "loss": 1.4778, "step": 12894 }, { "epoch": 0.9869125975815093, "grad_norm": 0.6378902642599417, "learning_rate": 2.0036349081605955e-06, "loss": 1.3043, "step": 12895 }, { "epoch": 0.9869891320985764, "grad_norm": 0.6110303708749313, "learning_rate": 2.003592770963973e-06, "loss": 1.3804, "step": 12896 }, { "epoch": 0.9870656666156437, "grad_norm": 0.6583866443050268, "learning_rate": 2.0035508793185585e-06, "loss": 1.5543, "step": 12897 }, { "epoch": 0.9871422011327109, "grad_norm": 0.6328009219492228, "learning_rate": 2.0035092332269247e-06, "loss": 1.3418, "step": 12898 }, { "epoch": 0.987218735649778, "grad_norm": 0.6484687388746055, "learning_rate": 2.003467832691631e-06, "loss": 1.4133, "step": 12899 }, { "epoch": 0.9872952701668453, "grad_norm": 0.5579763424669943, "learning_rate": 2.0034266777152205e-06, "loss": 1.0757, "step": 12900 }, { "epoch": 0.9873718046839124, "grad_norm": 0.9757975857330834, "learning_rate": 2.003385768300223e-06, "loss": 1.3238, "step": 12901 }, { "epoch": 0.9874483392009796, "grad_norm": 0.550178478548795, "learning_rate": 2.003345104449151e-06, "loss": 1.2885, "step": 12902 }, { "epoch": 0.9875248737180469, "grad_norm": 0.5998230720405251, "learning_rate": 2.0033046861645038e-06, "loss": 1.4103, "step": 12903 }, { "epoch": 0.987601408235114, "grad_norm": 0.6793404259439142, "learning_rate": 2.0032645134487647e-06, "loss": 1.6786, "step": 12904 }, { "epoch": 0.9876779427521812, "grad_norm": 0.606120721378309, "learning_rate": 2.0032245863044014e-06, "loss": 1.5147, "step": 12905 }, { "epoch": 0.9877544772692485, "grad_norm": 0.5798303654239988, "learning_rate": 2.0031849047338677e-06, "loss": 1.4747, "step": 12906 }, { "epoch": 0.9878310117863156, "grad_norm": 0.5425045660136804, "learning_rate": 2.0031454687396018e-06, "loss": 1.2066, "step": 12907 }, { "epoch": 0.9879075463033828, "grad_norm": 0.6832534010728192, "learning_rate": 2.003106278324027e-06, "loss": 1.4477, "step": 12908 }, { "epoch": 0.98798408082045, "grad_norm": 0.5657049543495118, "learning_rate": 2.00306733348955e-06, "loss": 1.3907, "step": 12909 }, { "epoch": 0.9880606153375172, "grad_norm": 0.6114566201938275, "learning_rate": 2.003028634238565e-06, "loss": 1.4301, "step": 12910 }, { "epoch": 0.9881371498545845, "grad_norm": 0.7748856692144541, "learning_rate": 2.0029901805734494e-06, "loss": 1.4373, "step": 12911 }, { "epoch": 0.9882136843716516, "grad_norm": 0.6665968851463621, "learning_rate": 2.0029519724965663e-06, "loss": 1.7295, "step": 12912 }, { "epoch": 0.9882902188887188, "grad_norm": 0.5686674343597797, "learning_rate": 2.0029140100102628e-06, "loss": 1.2616, "step": 12913 }, { "epoch": 0.9883667534057861, "grad_norm": 0.6949207635357988, "learning_rate": 2.002876293116871e-06, "loss": 1.7629, "step": 12914 }, { "epoch": 0.9884432879228532, "grad_norm": 0.6510676583259226, "learning_rate": 2.0028388218187092e-06, "loss": 1.6992, "step": 12915 }, { "epoch": 0.9885198224399204, "grad_norm": 0.6977131056956003, "learning_rate": 2.002801596118079e-06, "loss": 1.3628, "step": 12916 }, { "epoch": 0.9885963569569876, "grad_norm": 0.7290035954360218, "learning_rate": 2.0027646160172686e-06, "loss": 1.5763, "step": 12917 }, { "epoch": 0.9886728914740548, "grad_norm": 0.5643563770990814, "learning_rate": 2.00272788151855e-06, "loss": 1.104, "step": 12918 }, { "epoch": 0.988749425991122, "grad_norm": 0.6929228293966135, "learning_rate": 2.0026913926241794e-06, "loss": 1.724, "step": 12919 }, { "epoch": 0.9888259605081892, "grad_norm": 0.5630293127680679, "learning_rate": 2.0026551493363993e-06, "loss": 1.1972, "step": 12920 }, { "epoch": 0.9889024950252564, "grad_norm": 0.588490875975527, "learning_rate": 2.0026191516574363e-06, "loss": 1.4634, "step": 12921 }, { "epoch": 0.9889790295423236, "grad_norm": 0.7624878200422608, "learning_rate": 2.0025833995895033e-06, "loss": 1.4644, "step": 12922 }, { "epoch": 0.9890555640593908, "grad_norm": 0.5801080193246876, "learning_rate": 2.002547893134796e-06, "loss": 1.3181, "step": 12923 }, { "epoch": 0.989132098576458, "grad_norm": 0.8000878964104069, "learning_rate": 2.002512632295496e-06, "loss": 1.841, "step": 12924 }, { "epoch": 0.9892086330935251, "grad_norm": 0.5041592246662152, "learning_rate": 2.00247761707377e-06, "loss": 0.9868, "step": 12925 }, { "epoch": 0.9892851676105924, "grad_norm": 0.48599529504486555, "learning_rate": 2.00244284747177e-06, "loss": 1.0053, "step": 12926 }, { "epoch": 0.9893617021276596, "grad_norm": 0.5195792300324548, "learning_rate": 2.002408323491631e-06, "loss": 1.1865, "step": 12927 }, { "epoch": 0.9894382366447267, "grad_norm": 0.6353643240183906, "learning_rate": 2.002374045135476e-06, "loss": 1.3848, "step": 12928 }, { "epoch": 0.989514771161794, "grad_norm": 0.5831230591777042, "learning_rate": 2.00234001240541e-06, "loss": 1.264, "step": 12929 }, { "epoch": 0.9895913056788612, "grad_norm": 0.6045347852760444, "learning_rate": 2.002306225303524e-06, "loss": 1.3418, "step": 12930 }, { "epoch": 0.9896678401959283, "grad_norm": 0.5612177757479095, "learning_rate": 2.002272683831895e-06, "loss": 1.3346, "step": 12931 }, { "epoch": 0.9897443747129956, "grad_norm": 0.6250574672939232, "learning_rate": 2.0022393879925828e-06, "loss": 1.4152, "step": 12932 }, { "epoch": 0.9898209092300627, "grad_norm": 0.6572267195770075, "learning_rate": 2.002206337787634e-06, "loss": 1.427, "step": 12933 }, { "epoch": 0.98989744374713, "grad_norm": 0.5961036427482379, "learning_rate": 2.002173533219079e-06, "loss": 1.3501, "step": 12934 }, { "epoch": 0.9899739782641972, "grad_norm": 0.5635022540300328, "learning_rate": 2.0021409742889327e-06, "loss": 1.3305, "step": 12935 }, { "epoch": 0.9900505127812643, "grad_norm": 0.7026427843428079, "learning_rate": 2.0021086609991964e-06, "loss": 1.5589, "step": 12936 }, { "epoch": 0.9901270472983316, "grad_norm": 0.5647064416460915, "learning_rate": 2.0020765933518558e-06, "loss": 1.3336, "step": 12937 }, { "epoch": 0.9902035818153987, "grad_norm": 0.6719228835432961, "learning_rate": 2.0020447713488805e-06, "loss": 1.2365, "step": 12938 }, { "epoch": 0.9902801163324659, "grad_norm": 0.5078489588373307, "learning_rate": 2.002013194992226e-06, "loss": 1.1682, "step": 12939 }, { "epoch": 0.9903566508495332, "grad_norm": 0.6592993214350267, "learning_rate": 2.001981864283833e-06, "loss": 0.9853, "step": 12940 }, { "epoch": 0.9904331853666003, "grad_norm": 0.6510390699875501, "learning_rate": 2.0019507792256258e-06, "loss": 1.6122, "step": 12941 }, { "epoch": 0.9905097198836675, "grad_norm": 0.5437338235353509, "learning_rate": 2.001919939819515e-06, "loss": 1.1861, "step": 12942 }, { "epoch": 0.9905862544007348, "grad_norm": 0.6361386785695528, "learning_rate": 2.001889346067394e-06, "loss": 1.6135, "step": 12943 }, { "epoch": 0.9906627889178019, "grad_norm": 0.5482127918718506, "learning_rate": 2.001858997971144e-06, "loss": 1.2553, "step": 12944 }, { "epoch": 0.9907393234348691, "grad_norm": 1.4628299192986265, "learning_rate": 2.00182889553263e-06, "loss": 1.627, "step": 12945 }, { "epoch": 0.9908158579519363, "grad_norm": 0.5377838649605204, "learning_rate": 2.001799038753701e-06, "loss": 1.1726, "step": 12946 }, { "epoch": 0.9908923924690035, "grad_norm": 0.6388184674971775, "learning_rate": 2.0017694276361906e-06, "loss": 1.2779, "step": 12947 }, { "epoch": 0.9909689269860708, "grad_norm": 0.5892623106217463, "learning_rate": 2.0017400621819195e-06, "loss": 1.3901, "step": 12948 }, { "epoch": 0.9910454615031379, "grad_norm": 0.5478404072607487, "learning_rate": 2.001710942392691e-06, "loss": 1.2, "step": 12949 }, { "epoch": 0.9911219960202051, "grad_norm": 0.6037029992374066, "learning_rate": 2.0016820682702954e-06, "loss": 1.5477, "step": 12950 }, { "epoch": 0.9911985305372724, "grad_norm": 0.5335244847728262, "learning_rate": 2.001653439816506e-06, "loss": 1.1625, "step": 12951 }, { "epoch": 0.9912750650543395, "grad_norm": 0.6686997143070676, "learning_rate": 2.001625057033082e-06, "loss": 1.3954, "step": 12952 }, { "epoch": 0.9913515995714067, "grad_norm": 0.6066755682164484, "learning_rate": 2.0015969199217675e-06, "loss": 1.4399, "step": 12953 }, { "epoch": 0.9914281340884739, "grad_norm": 0.601640141824699, "learning_rate": 2.001569028484291e-06, "loss": 1.4789, "step": 12954 }, { "epoch": 0.9915046686055411, "grad_norm": 0.5468072998583415, "learning_rate": 2.001541382722366e-06, "loss": 1.2548, "step": 12955 }, { "epoch": 0.9915812031226083, "grad_norm": 0.5768007880373542, "learning_rate": 2.001513982637692e-06, "loss": 1.2078, "step": 12956 }, { "epoch": 0.9916577376396755, "grad_norm": 0.61524887592653, "learning_rate": 2.0014868282319527e-06, "loss": 1.1486, "step": 12957 }, { "epoch": 0.9917342721567427, "grad_norm": 0.7153387503575459, "learning_rate": 2.001459919506815e-06, "loss": 1.216, "step": 12958 }, { "epoch": 0.9918108066738099, "grad_norm": 0.6657516029769646, "learning_rate": 2.0014332564639334e-06, "loss": 1.6869, "step": 12959 }, { "epoch": 0.9918873411908771, "grad_norm": 0.6350288500639444, "learning_rate": 2.0014068391049465e-06, "loss": 1.4456, "step": 12960 }, { "epoch": 0.9919638757079443, "grad_norm": 0.5883656091840671, "learning_rate": 2.001380667431477e-06, "loss": 1.3887, "step": 12961 }, { "epoch": 0.9920404102250114, "grad_norm": 0.5075316593477702, "learning_rate": 2.001354741445132e-06, "loss": 1.1423, "step": 12962 }, { "epoch": 0.9921169447420787, "grad_norm": 0.5655956532647453, "learning_rate": 2.0013290611475058e-06, "loss": 1.4032, "step": 12963 }, { "epoch": 0.9921934792591459, "grad_norm": 0.5641460053745322, "learning_rate": 2.0013036265401762e-06, "loss": 1.2167, "step": 12964 }, { "epoch": 0.992270013776213, "grad_norm": 0.596601220074123, "learning_rate": 2.001278437624705e-06, "loss": 1.3547, "step": 12965 }, { "epoch": 0.9923465482932803, "grad_norm": 1.0167643179485004, "learning_rate": 2.0012534944026412e-06, "loss": 1.6547, "step": 12966 }, { "epoch": 0.9924230828103474, "grad_norm": 0.5530863737919655, "learning_rate": 2.001228796875516e-06, "loss": 1.1572, "step": 12967 }, { "epoch": 0.9924996173274147, "grad_norm": 0.6162451455257155, "learning_rate": 2.001204345044848e-06, "loss": 1.4665, "step": 12968 }, { "epoch": 0.9925761518444819, "grad_norm": 0.5453576338285594, "learning_rate": 2.0011801389121386e-06, "loss": 1.1247, "step": 12969 }, { "epoch": 0.992652686361549, "grad_norm": 0.8193141160001906, "learning_rate": 2.0011561784788765e-06, "loss": 1.3473, "step": 12970 }, { "epoch": 0.9927292208786163, "grad_norm": 0.6767878026698161, "learning_rate": 2.0011324637465324e-06, "loss": 1.5192, "step": 12971 }, { "epoch": 0.9928057553956835, "grad_norm": 0.5998916718903017, "learning_rate": 2.0011089947165645e-06, "loss": 1.4013, "step": 12972 }, { "epoch": 0.9928822899127506, "grad_norm": 0.5271434920565463, "learning_rate": 2.001085771390414e-06, "loss": 1.0021, "step": 12973 }, { "epoch": 0.9929588244298179, "grad_norm": 0.5949604128997958, "learning_rate": 2.0010627937695085e-06, "loss": 1.321, "step": 12974 }, { "epoch": 0.993035358946885, "grad_norm": 0.5253044198703639, "learning_rate": 2.0010400618552593e-06, "loss": 1.2548, "step": 12975 }, { "epoch": 0.9931118934639522, "grad_norm": 0.5253184156634917, "learning_rate": 2.001017575649063e-06, "loss": 1.261, "step": 12976 }, { "epoch": 0.9931884279810195, "grad_norm": 0.5843371908900523, "learning_rate": 2.000995335152302e-06, "loss": 1.4216, "step": 12977 }, { "epoch": 0.9932649624980866, "grad_norm": 0.6465281138623915, "learning_rate": 2.0009733403663424e-06, "loss": 1.4437, "step": 12978 }, { "epoch": 0.9933414970151538, "grad_norm": 1.0116860800380887, "learning_rate": 2.0009515912925358e-06, "loss": 1.2211, "step": 12979 }, { "epoch": 0.9934180315322211, "grad_norm": 0.5663261616783403, "learning_rate": 2.0009300879322175e-06, "loss": 1.2733, "step": 12980 }, { "epoch": 0.9934945660492882, "grad_norm": 0.8360513770394022, "learning_rate": 2.00090883028671e-06, "loss": 1.6645, "step": 12981 }, { "epoch": 0.9935711005663554, "grad_norm": 0.5557453602993356, "learning_rate": 2.000887818357319e-06, "loss": 1.3025, "step": 12982 }, { "epoch": 0.9936476350834226, "grad_norm": 0.7076568452259068, "learning_rate": 2.0008670521453357e-06, "loss": 1.7179, "step": 12983 }, { "epoch": 0.9937241696004898, "grad_norm": 0.5899487484415142, "learning_rate": 2.0008465316520353e-06, "loss": 1.2437, "step": 12984 }, { "epoch": 0.993800704117557, "grad_norm": 0.7985620467031248, "learning_rate": 2.0008262568786797e-06, "loss": 1.7464, "step": 12985 }, { "epoch": 0.9938772386346242, "grad_norm": 0.7019893459873681, "learning_rate": 2.000806227826514e-06, "loss": 1.6492, "step": 12986 }, { "epoch": 0.9939537731516914, "grad_norm": 0.6731516585037368, "learning_rate": 2.000786444496769e-06, "loss": 1.5152, "step": 12987 }, { "epoch": 0.9940303076687587, "grad_norm": 0.47630488267124016, "learning_rate": 2.00076690689066e-06, "loss": 0.9822, "step": 12988 }, { "epoch": 0.9941068421858258, "grad_norm": 0.5408306694857723, "learning_rate": 2.000747615009388e-06, "loss": 1.1314, "step": 12989 }, { "epoch": 0.994183376702893, "grad_norm": 0.6681734949185434, "learning_rate": 2.0007285688541375e-06, "loss": 1.5862, "step": 12990 }, { "epoch": 0.9942599112199602, "grad_norm": 0.5786720596320534, "learning_rate": 2.0007097684260796e-06, "loss": 1.3134, "step": 12991 }, { "epoch": 0.9943364457370274, "grad_norm": 0.5243478231983177, "learning_rate": 2.0006912137263692e-06, "loss": 1.22, "step": 12992 }, { "epoch": 0.9944129802540946, "grad_norm": 0.5683695364519806, "learning_rate": 2.0006729047561465e-06, "loss": 1.329, "step": 12993 }, { "epoch": 0.9944895147711618, "grad_norm": 0.9551139496540639, "learning_rate": 2.0006548415165357e-06, "loss": 1.6497, "step": 12994 }, { "epoch": 0.994566049288229, "grad_norm": 0.5082620274094084, "learning_rate": 2.0006370240086474e-06, "loss": 1.1525, "step": 12995 }, { "epoch": 0.9946425838052962, "grad_norm": 0.7066882347926456, "learning_rate": 2.0006194522335763e-06, "loss": 1.8106, "step": 12996 }, { "epoch": 0.9947191183223634, "grad_norm": 0.6820826533466966, "learning_rate": 2.0006021261924017e-06, "loss": 1.6229, "step": 12997 }, { "epoch": 0.9947956528394306, "grad_norm": 0.9063999562632229, "learning_rate": 2.0005850458861885e-06, "loss": 1.4713, "step": 12998 }, { "epoch": 0.9948721873564977, "grad_norm": 0.5714436694444994, "learning_rate": 2.0005682113159864e-06, "loss": 1.3635, "step": 12999 }, { "epoch": 0.994948721873565, "grad_norm": 0.5923732251316274, "learning_rate": 2.000551622482829e-06, "loss": 1.2191, "step": 13000 }, { "epoch": 0.9950252563906322, "grad_norm": 0.4890801327519138, "learning_rate": 2.000535279387736e-06, "loss": 0.958, "step": 13001 }, { "epoch": 0.9951017909076993, "grad_norm": 0.6097980075729084, "learning_rate": 2.0005191820317117e-06, "loss": 1.3567, "step": 13002 }, { "epoch": 0.9951783254247666, "grad_norm": 0.5631972921453298, "learning_rate": 2.000503330415745e-06, "loss": 1.148, "step": 13003 }, { "epoch": 0.9952548599418337, "grad_norm": 0.5640132682172709, "learning_rate": 2.00048772454081e-06, "loss": 1.3165, "step": 13004 }, { "epoch": 0.995331394458901, "grad_norm": 0.6063310470314272, "learning_rate": 2.0004723644078657e-06, "loss": 1.3852, "step": 13005 }, { "epoch": 0.9954079289759682, "grad_norm": 0.5990664856896989, "learning_rate": 2.000457250017855e-06, "loss": 1.2859, "step": 13006 }, { "epoch": 0.9954844634930353, "grad_norm": 0.5966199432719941, "learning_rate": 2.0004423813717075e-06, "loss": 1.3514, "step": 13007 }, { "epoch": 0.9955609980101026, "grad_norm": 0.8554454939268533, "learning_rate": 2.0004277584703367e-06, "loss": 1.5441, "step": 13008 }, { "epoch": 0.9956375325271698, "grad_norm": 0.5562475254932067, "learning_rate": 2.0004133813146402e-06, "loss": 1.2831, "step": 13009 }, { "epoch": 0.9957140670442369, "grad_norm": 0.6186008005751717, "learning_rate": 2.000399249905503e-06, "loss": 1.5586, "step": 13010 }, { "epoch": 0.9957906015613042, "grad_norm": 0.7147639831620815, "learning_rate": 2.000385364243792e-06, "loss": 1.6876, "step": 13011 }, { "epoch": 0.9958671360783713, "grad_norm": 0.6004708752175615, "learning_rate": 2.0003717243303605e-06, "loss": 1.2196, "step": 13012 }, { "epoch": 0.9959436705954385, "grad_norm": 0.5611291599386307, "learning_rate": 2.0003583301660466e-06, "loss": 1.1631, "step": 13013 }, { "epoch": 0.9960202051125058, "grad_norm": 0.6675837148869526, "learning_rate": 2.0003451817516744e-06, "loss": 1.5383, "step": 13014 }, { "epoch": 0.9960967396295729, "grad_norm": 0.6026217824281469, "learning_rate": 2.0003322790880504e-06, "loss": 1.3737, "step": 13015 }, { "epoch": 0.9961732741466401, "grad_norm": 0.6081291756624294, "learning_rate": 2.000319622175968e-06, "loss": 1.4586, "step": 13016 }, { "epoch": 0.9962498086637074, "grad_norm": 0.5486199660331457, "learning_rate": 2.000307211016205e-06, "loss": 1.2056, "step": 13017 }, { "epoch": 0.9963263431807745, "grad_norm": 0.7416947404171138, "learning_rate": 2.0002950456095235e-06, "loss": 1.1636, "step": 13018 }, { "epoch": 0.9964028776978417, "grad_norm": 0.5986646190911965, "learning_rate": 2.0002831259566714e-06, "loss": 1.203, "step": 13019 }, { "epoch": 0.9964794122149089, "grad_norm": 0.5702970058142802, "learning_rate": 2.000271452058381e-06, "loss": 1.3394, "step": 13020 }, { "epoch": 0.9965559467319761, "grad_norm": 0.8281869740748563, "learning_rate": 2.000260023915369e-06, "loss": 1.4024, "step": 13021 }, { "epoch": 0.9966324812490434, "grad_norm": 0.5206970726697996, "learning_rate": 2.0002488415283393e-06, "loss": 1.1883, "step": 13022 }, { "epoch": 0.9967090157661105, "grad_norm": 0.6506995591779388, "learning_rate": 2.0002379048979763e-06, "loss": 1.5551, "step": 13023 }, { "epoch": 0.9967855502831777, "grad_norm": 0.547913044524664, "learning_rate": 2.000227214024954e-06, "loss": 1.2865, "step": 13024 }, { "epoch": 0.996862084800245, "grad_norm": 0.5722468937587165, "learning_rate": 2.000216768909929e-06, "loss": 1.4032, "step": 13025 }, { "epoch": 0.9969386193173121, "grad_norm": 0.5714939446207787, "learning_rate": 2.0002065695535423e-06, "loss": 1.428, "step": 13026 }, { "epoch": 0.9970151538343793, "grad_norm": 0.5129720293713371, "learning_rate": 2.0001966159564214e-06, "loss": 1.0476, "step": 13027 }, { "epoch": 0.9970916883514465, "grad_norm": 0.6449051907958071, "learning_rate": 2.0001869081191774e-06, "loss": 1.5254, "step": 13028 }, { "epoch": 0.9971682228685137, "grad_norm": 0.5873010601283938, "learning_rate": 2.000177446042407e-06, "loss": 1.5867, "step": 13029 }, { "epoch": 0.9972447573855809, "grad_norm": 0.838360882078541, "learning_rate": 2.0001682297266918e-06, "loss": 1.7609, "step": 13030 }, { "epoch": 0.9973212919026481, "grad_norm": 1.0746765615647254, "learning_rate": 2.000159259172597e-06, "loss": 1.7631, "step": 13031 }, { "epoch": 0.9973978264197153, "grad_norm": 0.5941067402288294, "learning_rate": 2.0001505343806755e-06, "loss": 1.3446, "step": 13032 }, { "epoch": 0.9974743609367824, "grad_norm": 0.624359027218494, "learning_rate": 2.000142055351462e-06, "loss": 1.557, "step": 13033 }, { "epoch": 0.9975508954538497, "grad_norm": 0.683073878347367, "learning_rate": 2.000133822085478e-06, "loss": 1.6448, "step": 13034 }, { "epoch": 0.9976274299709169, "grad_norm": 0.6025164473791025, "learning_rate": 2.000125834583229e-06, "loss": 1.6021, "step": 13035 }, { "epoch": 0.997703964487984, "grad_norm": 0.5651239103523484, "learning_rate": 2.0001180928452064e-06, "loss": 1.3101, "step": 13036 }, { "epoch": 0.9977804990050513, "grad_norm": 0.6532989369560483, "learning_rate": 2.0001105968718858e-06, "loss": 1.7662, "step": 13037 }, { "epoch": 0.9978570335221185, "grad_norm": 0.5956310799835218, "learning_rate": 2.0001033466637267e-06, "loss": 1.3845, "step": 13038 }, { "epoch": 0.9979335680391856, "grad_norm": 0.5378157946296284, "learning_rate": 2.0000963422211757e-06, "loss": 1.1841, "step": 13039 }, { "epoch": 0.9980101025562529, "grad_norm": 0.49921974711512446, "learning_rate": 2.000089583544663e-06, "loss": 0.9712, "step": 13040 }, { "epoch": 0.99808663707332, "grad_norm": 0.8365927834523001, "learning_rate": 2.0000830706346038e-06, "loss": 1.6023, "step": 13041 }, { "epoch": 0.9981631715903873, "grad_norm": 0.7031414074711987, "learning_rate": 2.0000768034913985e-06, "loss": 1.442, "step": 13042 }, { "epoch": 0.9982397061074545, "grad_norm": 0.5907316598114836, "learning_rate": 2.0000707821154318e-06, "loss": 1.2648, "step": 13043 }, { "epoch": 0.9983162406245216, "grad_norm": 0.5963578300932091, "learning_rate": 2.0000650065070733e-06, "loss": 1.4159, "step": 13044 }, { "epoch": 0.9983927751415889, "grad_norm": 0.8052045636654689, "learning_rate": 2.0000594766666785e-06, "loss": 1.2897, "step": 13045 }, { "epoch": 0.9984693096586561, "grad_norm": 0.6339183171653208, "learning_rate": 2.000054192594587e-06, "loss": 1.4792, "step": 13046 }, { "epoch": 0.9985458441757232, "grad_norm": 0.5598725486102696, "learning_rate": 2.0000491542911237e-06, "loss": 1.3166, "step": 13047 }, { "epoch": 0.9986223786927905, "grad_norm": 0.7311813720046193, "learning_rate": 2.0000443617565976e-06, "loss": 1.4712, "step": 13048 }, { "epoch": 0.9986989132098576, "grad_norm": 0.8364376406762094, "learning_rate": 2.0000398149913035e-06, "loss": 1.5597, "step": 13049 }, { "epoch": 0.9987754477269248, "grad_norm": 0.700146736821876, "learning_rate": 2.0000355139955207e-06, "loss": 1.3969, "step": 13050 }, { "epoch": 0.9988519822439921, "grad_norm": 0.5439882237768264, "learning_rate": 2.0000314587695136e-06, "loss": 1.1795, "step": 13051 }, { "epoch": 0.9989285167610592, "grad_norm": 0.5189203680952374, "learning_rate": 2.000027649313532e-06, "loss": 1.1457, "step": 13052 }, { "epoch": 0.9990050512781264, "grad_norm": 0.576130071532637, "learning_rate": 2.0000240856278085e-06, "loss": 1.3755, "step": 13053 }, { "epoch": 0.9990815857951937, "grad_norm": 0.6666140435875639, "learning_rate": 2.0000207677125635e-06, "loss": 1.0529, "step": 13054 }, { "epoch": 0.9991581203122608, "grad_norm": 0.6314378092572882, "learning_rate": 2.0000176955679994e-06, "loss": 1.5929, "step": 13055 }, { "epoch": 0.999234654829328, "grad_norm": 0.8227603174692834, "learning_rate": 2.000014869194306e-06, "loss": 1.6969, "step": 13056 }, { "epoch": 0.9993111893463952, "grad_norm": 0.9867236389140297, "learning_rate": 2.000012288591657e-06, "loss": 1.9486, "step": 13057 }, { "epoch": 0.9993877238634624, "grad_norm": 0.5752934201708133, "learning_rate": 2.0000099537602107e-06, "loss": 1.2543, "step": 13058 }, { "epoch": 0.9994642583805297, "grad_norm": 0.5226959475694714, "learning_rate": 2.0000078647001105e-06, "loss": 1.135, "step": 13059 }, { "epoch": 0.9995407928975968, "grad_norm": 0.6772375885261784, "learning_rate": 2.0000060214114844e-06, "loss": 1.4702, "step": 13060 }, { "epoch": 0.999617327414664, "grad_norm": 0.586320572800883, "learning_rate": 2.0000044238944464e-06, "loss": 1.2433, "step": 13061 }, { "epoch": 0.9996938619317313, "grad_norm": 0.5917503552503686, "learning_rate": 2.0000030721490944e-06, "loss": 1.4423, "step": 13062 }, { "epoch": 0.9997703964487984, "grad_norm": 0.6223631563088833, "learning_rate": 2.0000019661755108e-06, "loss": 1.3462, "step": 13063 }, { "epoch": 0.9998469309658656, "grad_norm": 0.6660835953897518, "learning_rate": 2.0000011059737643e-06, "loss": 1.5074, "step": 13064 }, { "epoch": 0.9999234654829328, "grad_norm": 0.46425644119777787, "learning_rate": 2.000000491543908e-06, "loss": 0.9209, "step": 13065 }, { "epoch": 1.0, "grad_norm": 0.5101812190915097, "learning_rate": 2.000000122885979e-06, "loss": 1.0939, "step": 13066 }, { "epoch": 1.0, "step": 13066, "total_flos": 1583581017145344.0, "train_loss": 1.4104577668151463, "train_runtime": 25527.8837, "train_samples_per_second": 4.094, "train_steps_per_second": 0.512 } ], "logging_steps": 1, "max_steps": 13066, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 500000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 1583581017145344.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }