{
  "best_global_step": 24000,
  "best_metric": 0.6216655969619751,
  "best_model_checkpoint": "/content/drive/MyDrive/fyp-2025/ModelFinetuniningData/SmolLM3-3B-instruct-customerservice/outputs/checkpoint-24000",
  "epoch": 3.0,
  "eval_steps": 2000,
  "global_step": 24063,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006233636703652911,
      "grad_norm": 3.068476438522339,
      "learning_rate": 8.139534883720931e-07,
      "loss": 2.6369,
      "step": 50
    },
    {
      "epoch": 0.012467273407305822,
      "grad_norm": 1.983383059501648,
      "learning_rate": 1.6445182724252492e-06,
      "loss": 2.516,
      "step": 100
    },
    {
      "epoch": 0.018700910110958733,
      "grad_norm": 1.4683409929275513,
      "learning_rate": 2.4750830564784057e-06,
      "loss": 2.2605,
      "step": 150
    },
    {
      "epoch": 0.024934546814611644,
      "grad_norm": 1.5052752494812012,
      "learning_rate": 3.305647840531562e-06,
      "loss": 1.8336,
      "step": 200
    },
    {
      "epoch": 0.031168183518264555,
      "grad_norm": 0.6434653997421265,
      "learning_rate": 4.136212624584718e-06,
      "loss": 1.2924,
      "step": 250
    },
    {
      "epoch": 0.037401820221917466,
      "grad_norm": 0.5635990500450134,
      "learning_rate": 4.966777408637874e-06,
      "loss": 1.1154,
      "step": 300
    },
    {
      "epoch": 0.04363545692557038,
      "grad_norm": 0.8984834551811218,
      "learning_rate": 5.79734219269103e-06,
      "loss": 1.0431,
      "step": 350
    },
    {
      "epoch": 0.04986909362922329,
      "grad_norm": 0.5374437570571899,
      "learning_rate": 6.627906976744186e-06,
      "loss": 0.9776,
      "step": 400
    },
    {
      "epoch": 0.0561027303328762,
      "grad_norm": 0.6425241231918335,
      "learning_rate": 7.4584717607973425e-06,
      "loss": 0.9427,
      "step": 450
    },
    {
      "epoch": 0.06233636703652911,
      "grad_norm": 0.7180758714675903,
      "learning_rate": 8.2890365448505e-06,
      "loss": 0.9115,
      "step": 500
    },
    {
      "epoch": 0.06857000374018202,
      "grad_norm": 0.7414788603782654,
      "learning_rate": 9.119601328903655e-06,
      "loss": 0.8956,
      "step": 550
    },
    {
      "epoch": 0.07480364044383493,
      "grad_norm": 0.7637382745742798,
      "learning_rate": 9.950166112956811e-06,
      "loss": 0.8766,
      "step": 600
    },
    {
      "epoch": 0.08103727714748785,
      "grad_norm": 0.9042564630508423,
      "learning_rate": 1.0780730897009968e-05,
      "loss": 0.8627,
      "step": 650
    },
    {
      "epoch": 0.08727091385114076,
      "grad_norm": 0.7853376865386963,
      "learning_rate": 1.1611295681063124e-05,
      "loss": 0.842,
      "step": 700
    },
    {
      "epoch": 0.09350455055479366,
      "grad_norm": 0.9337324500083923,
      "learning_rate": 1.244186046511628e-05,
      "loss": 0.822,
      "step": 750
    },
    {
      "epoch": 0.09973818725844658,
      "grad_norm": 0.8988317251205444,
      "learning_rate": 1.3272425249169436e-05,
      "loss": 0.8269,
      "step": 800
    },
    {
      "epoch": 0.10597182396209949,
      "grad_norm": 0.9164941906929016,
      "learning_rate": 1.4102990033222592e-05,
      "loss": 0.8222,
      "step": 850
    },
    {
      "epoch": 0.1122054606657524,
      "grad_norm": 0.9890440106391907,
      "learning_rate": 1.4933554817275748e-05,
      "loss": 0.812,
      "step": 900
    },
    {
      "epoch": 0.1184390973694053,
      "grad_norm": 0.9969121813774109,
      "learning_rate": 1.5764119601328905e-05,
      "loss": 0.8018,
      "step": 950
    },
    {
      "epoch": 0.12467273407305822,
      "grad_norm": 1.0014499425888062,
      "learning_rate": 1.659468438538206e-05,
      "loss": 0.7974,
      "step": 1000
    },
    {
      "epoch": 0.13090637077671113,
      "grad_norm": 0.992202877998352,
      "learning_rate": 1.7425249169435217e-05,
      "loss": 0.7925,
      "step": 1050
    },
    {
      "epoch": 0.13714000748036403,
      "grad_norm": 0.9398910999298096,
      "learning_rate": 1.825581395348837e-05,
      "loss": 0.7835,
      "step": 1100
    },
    {
      "epoch": 0.14337364418401696,
      "grad_norm": 1.0303184986114502,
      "learning_rate": 1.908637873754153e-05,
      "loss": 0.7845,
      "step": 1150
    },
    {
      "epoch": 0.14960728088766986,
      "grad_norm": 1.0107524394989014,
      "learning_rate": 1.9916943521594686e-05,
      "loss": 0.7602,
      "step": 1200
    },
    {
      "epoch": 0.15584091759132276,
      "grad_norm": 1.0360383987426758,
      "learning_rate": 1.999980875991073e-05,
      "loss": 0.7654,
      "step": 1250
    },
    {
      "epoch": 0.1620745542949757,
      "grad_norm": 1.1466953754425049,
      "learning_rate": 1.9999147692449697e-05,
      "loss": 0.7591,
      "step": 1300
    },
    {
      "epoch": 0.1683081909986286,
      "grad_norm": 1.0271602869033813,
      "learning_rate": 1.999801446783615e-05,
      "loss": 0.749,
      "step": 1350
    },
    {
      "epoch": 0.17454182770228152,
      "grad_norm": 1.1700372695922852,
      "learning_rate": 1.9996409139580664e-05,
      "loss": 0.7487,
      "step": 1400
    },
    {
      "epoch": 0.18077546440593442,
      "grad_norm": 1.1871458292007446,
      "learning_rate": 1.9994331783486415e-05,
      "loss": 0.7481,
      "step": 1450
    },
    {
      "epoch": 0.18700910110958732,
      "grad_norm": 1.1252415180206299,
      "learning_rate": 1.9991782497645624e-05,
      "loss": 0.7398,
      "step": 1500
    },
    {
      "epoch": 0.19324273781324025,
      "grad_norm": 1.1741167306900024,
      "learning_rate": 1.99887614024349e-05,
      "loss": 0.7394,
      "step": 1550
    },
    {
      "epoch": 0.19947637451689315,
      "grad_norm": 1.27838134765625,
      "learning_rate": 1.9985268640509576e-05,
      "loss": 0.7375,
      "step": 1600
    },
    {
      "epoch": 0.20571001122054608,
      "grad_norm": 1.1667146682739258,
      "learning_rate": 1.998130437679696e-05,
      "loss": 0.7393,
      "step": 1650
    },
    {
      "epoch": 0.21194364792419898,
      "grad_norm": 1.224082112312317,
      "learning_rate": 1.997686879848855e-05,
      "loss": 0.7334,
      "step": 1700
    },
    {
      "epoch": 0.21817728462785188,
      "grad_norm": 1.2695999145507812,
      "learning_rate": 1.997196211503121e-05,
      "loss": 0.7305,
      "step": 1750
    },
    {
      "epoch": 0.2244109213315048,
      "grad_norm": 1.1981199979782104,
      "learning_rate": 1.9966584558117242e-05,
      "loss": 0.7203,
      "step": 1800
    },
    {
      "epoch": 0.2306445580351577,
      "grad_norm": 1.3349239826202393,
      "learning_rate": 1.9960736381673492e-05,
      "loss": 0.7202,
      "step": 1850
    },
    {
      "epoch": 0.2368781947388106,
      "grad_norm": 1.354668378829956,
      "learning_rate": 1.9954417861849332e-05,
      "loss": 0.719,
      "step": 1900
    },
    {
      "epoch": 0.24311183144246354,
      "grad_norm": 1.3404412269592285,
      "learning_rate": 1.994762929700362e-05,
      "loss": 0.7157,
      "step": 1950
    },
    {
      "epoch": 0.24934546814611644,
      "grad_norm": 1.3769536018371582,
      "learning_rate": 1.9940371007690626e-05,
      "loss": 0.7105,
      "step": 2000
    },
    {
      "epoch": 0.24934546814611644,
      "eval_loss": 0.7147436141967773,
      "eval_runtime": 430.9417,
      "eval_samples_per_second": 42.542,
      "eval_steps_per_second": 5.319,
      "step": 2000
    },
    {
      "epoch": 0.25557910484976937,
      "grad_norm": 1.3940078020095825,
      "learning_rate": 1.9932643336644877e-05,
      "loss": 0.7156,
      "step": 2050
    },
    {
      "epoch": 0.26181274155342227,
      "grad_norm": 1.3650559186935425,
      "learning_rate": 1.9924446648764995e-05,
      "loss": 0.7161,
      "step": 2100
    },
    {
      "epoch": 0.26804637825707517,
      "grad_norm": 1.4301294088363647,
      "learning_rate": 1.991578133109645e-05,
      "loss": 0.7125,
      "step": 2150
    },
    {
      "epoch": 0.27428001496072807,
      "grad_norm": 1.36927330493927,
      "learning_rate": 1.990664779281328e-05,
      "loss": 0.7031,
      "step": 2200
    },
    {
      "epoch": 0.280513651664381,
      "grad_norm": 1.332739233970642,
      "learning_rate": 1.9897046465198794e-05,
      "loss": 0.7078,
      "step": 2250
    },
    {
      "epoch": 0.2867472883680339,
      "grad_norm": 1.414450764656067,
      "learning_rate": 1.9886977801625176e-05,
      "loss": 0.7096,
      "step": 2300
    },
    {
      "epoch": 0.2929809250716868,
      "grad_norm": 1.4470826387405396,
      "learning_rate": 1.987644227753211e-05,
      "loss": 0.7097,
      "step": 2350
    },
    {
      "epoch": 0.2992145617753397,
      "grad_norm": 1.5148861408233643,
      "learning_rate": 1.98654403904043e-05,
      "loss": 0.6961,
      "step": 2400
    },
    {
      "epoch": 0.3054481984789926,
      "grad_norm": 1.5179797410964966,
      "learning_rate": 1.9853972659747986e-05,
      "loss": 0.6998,
      "step": 2450
    },
    {
      "epoch": 0.31168183518264553,
      "grad_norm": 1.4796333312988281,
      "learning_rate": 1.9842039627066433e-05,
      "loss": 0.7118,
      "step": 2500
    },
    {
      "epoch": 0.3179154718862985,
      "grad_norm": 1.5036157369613647,
      "learning_rate": 1.982964185583434e-05,
      "loss": 0.6978,
      "step": 2550
    },
    {
      "epoch": 0.3241491085899514,
      "grad_norm": 1.5439958572387695,
      "learning_rate": 1.9816779931471238e-05,
      "loss": 0.7074,
      "step": 2600
    },
    {
      "epoch": 0.3303827452936043,
      "grad_norm": 1.4821096658706665,
      "learning_rate": 1.980345446131385e-05,
      "loss": 0.6938,
      "step": 2650
    },
    {
      "epoch": 0.3366163819972572,
      "grad_norm": 1.5255374908447266,
      "learning_rate": 1.9789666074587405e-05,
      "loss": 0.6934,
      "step": 2700
    },
    {
      "epoch": 0.3428500187009101,
      "grad_norm": 1.5652164220809937,
      "learning_rate": 1.9775415422375942e-05,
      "loss": 0.693,
      "step": 2750
    },
    {
      "epoch": 0.34908365540456304,
      "grad_norm": 1.5904299020767212,
      "learning_rate": 1.9760703177591547e-05,
      "loss": 0.6934,
      "step": 2800
    },
    {
      "epoch": 0.35531729210821594,
      "grad_norm": 1.6197664737701416,
      "learning_rate": 1.9745530034942594e-05,
      "loss": 0.7105,
      "step": 2850
    },
    {
      "epoch": 0.36155092881186884,
      "grad_norm": 1.5920610427856445,
      "learning_rate": 1.9729896710900927e-05,
      "loss": 0.6872,
      "step": 2900
    },
    {
      "epoch": 0.36778456551552174,
      "grad_norm": 1.6224204301834106,
      "learning_rate": 1.9713803943668045e-05,
      "loss": 0.6894,
      "step": 2950
    },
    {
      "epoch": 0.37401820221917464,
      "grad_norm": 1.5544761419296265,
      "learning_rate": 1.9697252493140228e-05,
      "loss": 0.6878,
      "step": 3000
    },
    {
      "epoch": 0.3802518389228276,
      "grad_norm": 1.6158560514450073,
      "learning_rate": 1.9680243140872664e-05,
      "loss": 0.7063,
      "step": 3050
    },
    {
      "epoch": 0.3864854756264805,
      "grad_norm": 1.6046901941299438,
      "learning_rate": 1.966277669004254e-05,
      "loss": 0.6894,
      "step": 3100
    },
    {
      "epoch": 0.3927191123301334,
      "grad_norm": 1.5848560333251953,
      "learning_rate": 1.9644853965411125e-05,
      "loss": 0.687,
      "step": 3150
    },
    {
      "epoch": 0.3989527490337863,
      "grad_norm": 1.6193879842758179,
      "learning_rate": 1.96264758132848e-05,
      "loss": 0.6942,
      "step": 3200
    },
    {
      "epoch": 0.4051863857374392,
      "grad_norm": 1.5715765953063965,
      "learning_rate": 1.9607643101475146e-05,
      "loss": 0.6892,
      "step": 3250
    },
    {
      "epoch": 0.41142002244109216,
      "grad_norm": 1.5581854581832886,
      "learning_rate": 1.95883567192579e-05,
      "loss": 0.6946,
      "step": 3300
    },
    {
      "epoch": 0.41765365914474506,
      "grad_norm": 1.6775258779525757,
      "learning_rate": 1.9568617577331014e-05,
      "loss": 0.6852,
      "step": 3350
    },
    {
      "epoch": 0.42388729584839796,
      "grad_norm": 1.6307775974273682,
      "learning_rate": 1.954842660777164e-05,
      "loss": 0.6715,
      "step": 3400
    },
    {
      "epoch": 0.43012093255205086,
      "grad_norm": 1.626529574394226,
      "learning_rate": 1.9527784763992106e-05,
      "loss": 0.68,
      "step": 3450
    },
    {
      "epoch": 0.43635456925570376,
      "grad_norm": 1.6327893733978271,
      "learning_rate": 1.9506693020694904e-05,
      "loss": 0.6834,
      "step": 3500
    },
    {
      "epoch": 0.44258820595935666,
      "grad_norm": 1.606713056564331,
      "learning_rate": 1.948515237382666e-05,
      "loss": 0.677,
      "step": 3550
    },
    {
      "epoch": 0.4488218426630096,
      "grad_norm": 1.5914915800094604,
      "learning_rate": 1.9463163840531125e-05,
      "loss": 0.6861,
      "step": 3600
    },
    {
      "epoch": 0.4550554793666625,
      "grad_norm": 1.6924699544906616,
      "learning_rate": 1.9440728459101112e-05,
      "loss": 0.6887,
      "step": 3650
    },
    {
      "epoch": 0.4612891160703154,
      "grad_norm": 1.7773985862731934,
      "learning_rate": 1.9417847288929495e-05,
      "loss": 0.6786,
      "step": 3700
    },
    {
      "epoch": 0.4675227527739683,
      "grad_norm": 1.673498511314392,
      "learning_rate": 1.9394521410459182e-05,
      "loss": 0.6849,
      "step": 3750
    },
    {
      "epoch": 0.4737563894776212,
      "grad_norm": 1.6292023658752441,
      "learning_rate": 1.9370751925132082e-05,
      "loss": 0.6783,
      "step": 3800
    },
    {
      "epoch": 0.4799900261812742,
      "grad_norm": 1.738358736038208,
      "learning_rate": 1.9346539955337113e-05,
      "loss": 0.669,
      "step": 3850
    },
    {
      "epoch": 0.4862236628849271,
      "grad_norm": 1.7295713424682617,
      "learning_rate": 1.9321886644357178e-05,
      "loss": 0.6827,
      "step": 3900
    },
    {
      "epoch": 0.49245729958858,
      "grad_norm": 1.6828645467758179,
      "learning_rate": 1.9296793156315216e-05,
      "loss": 0.6696,
      "step": 3950
    },
    {
      "epoch": 0.4986909362922329,
      "grad_norm": 1.5834518671035767,
      "learning_rate": 1.9271260676119205e-05,
      "loss": 0.6824,
      "step": 4000
    },
    {
      "epoch": 0.4986909362922329,
      "eval_loss": 0.6768468022346497,
      "eval_runtime": 427.7007,
      "eval_samples_per_second": 42.864,
      "eval_steps_per_second": 5.359,
      "step": 4000
    },
    {
      "epoch": 0.5049245729958858,
      "grad_norm": 1.7124220132827759,
      "learning_rate": 1.924529040940621e-05,
      "loss": 0.6834,
      "step": 4050
    },
    {
      "epoch": 0.5111582096995387,
      "grad_norm": 1.6820108890533447,
      "learning_rate": 1.9218883582485476e-05,
      "loss": 0.6763,
      "step": 4100
    },
    {
      "epoch": 0.5173918464031916,
      "grad_norm": 1.6797947883605957,
      "learning_rate": 1.9192041442280494e-05,
      "loss": 0.6659,
      "step": 4150
    },
    {
      "epoch": 0.5236254831068445,
      "grad_norm": 1.738492488861084,
      "learning_rate": 1.916476525627014e-05,
      "loss": 0.671,
      "step": 4200
    },
    {
      "epoch": 0.5298591198104975,
      "grad_norm": 1.7402714490890503,
      "learning_rate": 1.9137056312428827e-05,
      "loss": 0.6716,
      "step": 4250
    },
    {
      "epoch": 0.5360927565141503,
      "grad_norm": 1.5923080444335938,
      "learning_rate": 1.910891591916567e-05,
      "loss": 0.672,
      "step": 4300
    },
    {
      "epoch": 0.5423263932178033,
      "grad_norm": 1.8019920587539673,
      "learning_rate": 1.908034540526272e-05,
      "loss": 0.687,
      "step": 4350
    },
    {
      "epoch": 0.5485600299214561,
      "grad_norm": 1.7456427812576294,
      "learning_rate": 1.9051346119812208e-05,
      "loss": 0.6745,
      "step": 4400
    },
    {
      "epoch": 0.5547936666251091,
      "grad_norm": 1.677939534187317,
      "learning_rate": 1.902191943215285e-05,
      "loss": 0.6731,
      "step": 4450
    },
    {
      "epoch": 0.561027303328762,
      "grad_norm": 1.7911790609359741,
      "learning_rate": 1.8992066731805175e-05,
      "loss": 0.6745,
      "step": 4500
    },
    {
      "epoch": 0.5672609400324149,
      "grad_norm": 1.8749769926071167,
      "learning_rate": 1.8961789428405933e-05,
      "loss": 0.6705,
      "step": 4550
    },
    {
      "epoch": 0.5734945767360679,
      "grad_norm": 1.7382309436798096,
      "learning_rate": 1.8931088951641512e-05,
      "loss": 0.6752,
      "step": 4600
    },
    {
      "epoch": 0.5797282134397207,
      "grad_norm": 1.7381843328475952,
      "learning_rate": 1.8899966751180435e-05,
      "loss": 0.6707,
      "step": 4650
    },
    {
      "epoch": 0.5859618501433737,
      "grad_norm": 1.6726250648498535,
      "learning_rate": 1.8868424296604913e-05,
      "loss": 0.6719,
      "step": 4700
    },
    {
      "epoch": 0.5921954868470266,
      "grad_norm": 1.7997413873672485,
      "learning_rate": 1.8836463077341447e-05,
      "loss": 0.663,
      "step": 4750
    },
    {
      "epoch": 0.5984291235506795,
      "grad_norm": 1.7714271545410156,
      "learning_rate": 1.880408460259049e-05,
      "loss": 0.6715,
      "step": 4800
    },
    {
      "epoch": 0.6046627602543324,
      "grad_norm": 1.7211424112319946,
      "learning_rate": 1.8771290401255194e-05,
      "loss": 0.6639,
      "step": 4850
    },
    {
      "epoch": 0.6108963969579853,
      "grad_norm": 1.7623661756515503,
      "learning_rate": 1.873808202186922e-05,
      "loss": 0.6671,
      "step": 4900
    },
    {
      "epoch": 0.6171300336616382,
      "grad_norm": 1.8073209524154663,
      "learning_rate": 1.87044610325236e-05,
      "loss": 0.6696,
      "step": 4950
    },
    {
      "epoch": 0.6233636703652911,
      "grad_norm": 1.7280222177505493,
      "learning_rate": 1.8670429020792703e-05,
      "loss": 0.6653,
      "step": 5000
    },
    {
      "epoch": 0.629597307068944,
      "grad_norm": 1.845685601234436,
      "learning_rate": 1.8635987593659274e-05,
      "loss": 0.6644,
      "step": 5050
    },
    {
      "epoch": 0.635830943772597,
      "grad_norm": 1.8520327806472778,
      "learning_rate": 1.860113837743853e-05,
      "loss": 0.6755,
      "step": 5100
    },
    {
      "epoch": 0.6420645804762498,
      "grad_norm": 1.7361048460006714,
      "learning_rate": 1.8565883017701404e-05,
      "loss": 0.6703,
      "step": 5150
    },
    {
      "epoch": 0.6482982171799028,
      "grad_norm": 1.8101972341537476,
      "learning_rate": 1.8530223179196807e-05,
      "loss": 0.6635,
      "step": 5200
    },
    {
      "epoch": 0.6545318538835556,
      "grad_norm": 1.7235043048858643,
      "learning_rate": 1.8494160545773036e-05,
      "loss": 0.6634,
      "step": 5250
    },
    {
      "epoch": 0.6607654905872086,
      "grad_norm": 1.8479424715042114,
      "learning_rate": 1.8457696820298253e-05,
      "loss": 0.6604,
      "step": 5300
    },
    {
      "epoch": 0.6669991272908615,
      "grad_norm": 1.6182621717453003,
      "learning_rate": 1.84208337245801e-05,
      "loss": 0.6704,
      "step": 5350
    },
    {
      "epoch": 0.6732327639945144,
      "grad_norm": 1.7111644744873047,
      "learning_rate": 1.8383572999284353e-05,
      "loss": 0.6715,
      "step": 5400
    },
    {
      "epoch": 0.6794664006981673,
      "grad_norm": 1.7877291440963745,
      "learning_rate": 1.8345916403852777e-05,
      "loss": 0.6666,
      "step": 5450
    },
    {
      "epoch": 0.6857000374018202,
      "grad_norm": 1.8633828163146973,
      "learning_rate": 1.8307865716420005e-05,
      "loss": 0.6643,
      "step": 5500
    },
    {
      "epoch": 0.6919336741054731,
      "grad_norm": 1.7590093612670898,
      "learning_rate": 1.8269422733729597e-05,
      "loss": 0.6639,
      "step": 5550
    },
    {
      "epoch": 0.6981673108091261,
      "grad_norm": 1.7710891962051392,
      "learning_rate": 1.8230589271049196e-05,
      "loss": 0.6666,
      "step": 5600
    },
    {
      "epoch": 0.7044009475127789,
      "grad_norm": 1.7103811502456665,
      "learning_rate": 1.819136716208481e-05,
      "loss": 0.6625,
      "step": 5650
    },
    {
      "epoch": 0.7106345842164319,
      "grad_norm": 1.7894842624664307,
      "learning_rate": 1.815175825889421e-05,
      "loss": 0.6655,
      "step": 5700
    },
    {
      "epoch": 0.7168682209200847,
      "grad_norm": 1.823889136314392,
      "learning_rate": 1.811176443179951e-05,
      "loss": 0.6536,
      "step": 5750
    },
    {
      "epoch": 0.7231018576237377,
      "grad_norm": 1.8381397724151611,
      "learning_rate": 1.807138756929881e-05,
      "loss": 0.6653,
      "step": 5800
    },
    {
      "epoch": 0.7293354943273906,
      "grad_norm": 1.8083523511886597,
      "learning_rate": 1.8030629577977064e-05,
      "loss": 0.6691,
      "step": 5850
    },
    {
      "epoch": 0.7355691310310435,
      "grad_norm": 1.8512042760849,
      "learning_rate": 1.798949238241601e-05,
      "loss": 0.6617,
      "step": 5900
    },
    {
      "epoch": 0.7418027677346964,
      "grad_norm": 1.8072071075439453,
      "learning_rate": 1.7947977925103315e-05,
      "loss": 0.6673,
      "step": 5950
    },
    {
      "epoch": 0.7480364044383493,
      "grad_norm": 1.8521157503128052,
      "learning_rate": 1.7906088166340864e-05,
      "loss": 0.6629,
      "step": 6000
    },
    {
      "epoch": 0.7480364044383493,
      "eval_loss": 0.6585834622383118,
      "eval_runtime": 427.2182,
      "eval_samples_per_second": 42.912,
      "eval_steps_per_second": 5.365,
      "step": 6000
    },
    {
      "epoch": 0.7542700411420022,
      "grad_norm": 1.8681087493896484,
      "learning_rate": 1.786382508415216e-05,
      "loss": 0.6597,
      "step": 6050
    },
    {
      "epoch": 0.7605036778456552,
      "grad_norm": 1.895345687866211,
      "learning_rate": 1.7821190674188953e-05,
      "loss": 0.6618,
      "step": 6100
    },
    {
      "epoch": 0.766737314549308,
      "grad_norm": 1.8438936471939087,
      "learning_rate": 1.7778186949636983e-05,
      "loss": 0.6594,
      "step": 6150
    },
    {
      "epoch": 0.772970951252961,
      "grad_norm": 1.7014251947402954,
      "learning_rate": 1.7734815941120933e-05,
      "loss": 0.6619,
      "step": 6200
    },
    {
      "epoch": 0.7792045879566138,
      "grad_norm": 1.8361589908599854,
      "learning_rate": 1.769107969660855e-05,
      "loss": 0.6627,
      "step": 6250
    },
    {
      "epoch": 0.7854382246602668,
      "grad_norm": 1.772194266319275,
      "learning_rate": 1.7646980281313917e-05,
      "loss": 0.6564,
      "step": 6300
    },
    {
      "epoch": 0.7916718613639198,
      "grad_norm": 1.7633798122406006,
      "learning_rate": 1.760251977759995e-05,
      "loss": 0.6483,
      "step": 6350
    },
    {
      "epoch": 0.7979054980675726,
      "grad_norm": 1.8437433242797852,
      "learning_rate": 1.7557700284880063e-05,
      "loss": 0.6632,
      "step": 6400
    },
    {
      "epoch": 0.8041391347712256,
      "grad_norm": 1.8506488800048828,
      "learning_rate": 1.751252391951905e-05,
      "loss": 0.6474,
      "step": 6450
    },
    {
      "epoch": 0.8103727714748784,
      "grad_norm": 1.7333770990371704,
      "learning_rate": 1.7466992814733123e-05,
      "loss": 0.654,
      "step": 6500
    },
    {
      "epoch": 0.8166064081785314,
      "grad_norm": 1.7919718027114868,
      "learning_rate": 1.7421109120489206e-05,
      "loss": 0.6607,
      "step": 6550
    },
    {
      "epoch": 0.8228400448821843,
      "grad_norm": 1.9186142683029175,
      "learning_rate": 1.7374875003403402e-05,
      "loss": 0.6554,
      "step": 6600
    },
    {
      "epoch": 0.8290736815858372,
      "grad_norm": 1.850548267364502,
      "learning_rate": 1.7328292646638694e-05,
      "loss": 0.6519,
      "step": 6650
    },
    {
      "epoch": 0.8353073182894901,
      "grad_norm": 1.8770530223846436,
      "learning_rate": 1.7281364249801846e-05,
      "loss": 0.6512,
      "step": 6700
    },
    {
      "epoch": 0.841540954993143,
      "grad_norm": 1.9304884672164917,
      "learning_rate": 1.723409202883955e-05,
      "loss": 0.6584,
      "step": 6750
    },
    {
      "epoch": 0.8477745916967959,
      "grad_norm": 1.901544451713562,
      "learning_rate": 1.7186478215933776e-05,
      "loss": 0.662,
      "step": 6800
    },
    {
      "epoch": 0.8540082284004489,
      "grad_norm": 1.9472225904464722,
      "learning_rate": 1.713852505939639e-05,
      "loss": 0.6518,
      "step": 6850
    },
    {
      "epoch": 0.8602418651041017,
      "grad_norm": 1.8834494352340698,
      "learning_rate": 1.7090234823562956e-05,
      "loss": 0.6575,
      "step": 6900
    },
    {
      "epoch": 0.8664755018077547,
      "grad_norm": 1.960532307624817,
      "learning_rate": 1.7041609788685853e-05,
      "loss": 0.6498,
      "step": 6950
    },
    {
      "epoch": 0.8727091385114075,
      "grad_norm": 1.9708137512207031,
      "learning_rate": 1.699265225082658e-05,
      "loss": 0.6584,
      "step": 7000
    },
    {
      "epoch": 0.8789427752150605,
      "grad_norm": 1.8422749042510986,
      "learning_rate": 1.694336452174733e-05,
      "loss": 0.6558,
      "step": 7050
    },
    {
      "epoch": 0.8851764119187133,
      "grad_norm": 1.9792207479476929,
      "learning_rate": 1.689374892880185e-05,
      "loss": 0.6475,
      "step": 7100
    },
    {
      "epoch": 0.8914100486223663,
      "grad_norm": 1.7789386510849,
      "learning_rate": 1.684380781482553e-05,
      "loss": 0.6478,
      "step": 7150
    },
    {
      "epoch": 0.8976436853260192,
      "grad_norm": 1.996501088142395,
      "learning_rate": 1.679354353802478e-05,
      "loss": 0.6567,
      "step": 7200
    },
    {
      "epoch": 0.9038773220296721,
      "grad_norm": 1.904082179069519,
      "learning_rate": 1.674295847186567e-05,
      "loss": 0.6555,
      "step": 7250
    },
    {
      "epoch": 0.910110958733325,
      "grad_norm": 1.7943248748779297,
      "learning_rate": 1.6692055004961867e-05,
      "loss": 0.6515,
      "step": 7300
    },
    {
      "epoch": 0.9163445954369779,
      "grad_norm": 1.813570261001587,
      "learning_rate": 1.664083554096183e-05,
      "loss": 0.6539,
      "step": 7350
    },
    {
      "epoch": 0.9225782321406308,
      "grad_norm": 1.762556552886963,
      "learning_rate": 1.6589302498435324e-05,
      "loss": 0.6457,
      "step": 7400
    },
    {
      "epoch": 0.9288118688442838,
      "grad_norm": 1.8273394107818604,
      "learning_rate": 1.6537458310759215e-05,
      "loss": 0.6536,
      "step": 7450
    },
    {
      "epoch": 0.9350455055479366,
      "grad_norm": 1.7868757247924805,
      "learning_rate": 1.648530542600255e-05,
      "loss": 0.6426,
      "step": 7500
    },
    {
      "epoch": 0.9412791422515896,
      "grad_norm": 1.8329124450683594,
      "learning_rate": 1.6432846306810982e-05,
      "loss": 0.6443,
      "step": 7550
    },
    {
      "epoch": 0.9475127789552424,
      "grad_norm": 1.9399909973144531,
      "learning_rate": 1.6380083430290467e-05,
      "loss": 0.6523,
      "step": 7600
    },
    {
      "epoch": 0.9537464156588954,
      "grad_norm": 1.9386730194091797,
      "learning_rate": 1.632701928789031e-05,
      "loss": 0.6483,
      "step": 7650
    },
    {
      "epoch": 0.9599800523625484,
      "grad_norm": 1.9241693019866943,
      "learning_rate": 1.627365638528551e-05,
      "loss": 0.6512,
      "step": 7700
    },
    {
      "epoch": 0.9662136890662012,
      "grad_norm": 1.8745570182800293,
      "learning_rate": 1.621999724225844e-05,
      "loss": 0.6486,
      "step": 7750
    },
    {
      "epoch": 0.9724473257698542,
      "grad_norm": 1.9266139268875122,
      "learning_rate": 1.6166044392579877e-05,
      "loss": 0.6486,
      "step": 7800
    },
    {
      "epoch": 0.978680962473507,
      "grad_norm": 2.0344784259796143,
      "learning_rate": 1.6111800383889345e-05,
      "loss": 0.6468,
      "step": 7850
    },
    {
      "epoch": 0.98491459917716,
      "grad_norm": 1.822317123413086,
      "learning_rate": 1.605726777757482e-05,
      "loss": 0.6516,
      "step": 7900
    },
    {
      "epoch": 0.9911482358808129,
      "grad_norm": 1.9456557035446167,
      "learning_rate": 1.6002449148651784e-05,
      "loss": 0.6602,
      "step": 7950
    },
    {
      "epoch": 0.9973818725844658,
      "grad_norm": 1.867618203163147,
      "learning_rate": 1.5947347085641632e-05,
      "loss": 0.6443,
      "step": 8000
    },
    {
      "epoch": 0.9973818725844658,
      "eval_loss": 0.6477607488632202,
      "eval_runtime": 427.1421,
      "eval_samples_per_second": 42.92,
      "eval_steps_per_second": 5.366,
      "step": 8000
    },
    {
      "epoch": 1.0036155092881187,
      "grad_norm": 1.7887365818023682,
      "learning_rate": 1.5891964190449447e-05,
      "loss": 0.6239,
      "step": 8050
    },
    {
      "epoch": 1.0098491459917716,
      "grad_norm": 2.0542945861816406,
      "learning_rate": 1.583630307824113e-05,
      "loss": 0.6369,
      "step": 8100
    },
    {
      "epoch": 1.0160827826954244,
      "grad_norm": 1.9550195932388306,
      "learning_rate": 1.5780366377319913e-05,
      "loss": 0.6403,
      "step": 8150
    },
    {
      "epoch": 1.0223164193990775,
      "grad_norm": 1.8278982639312744,
      "learning_rate": 1.572415672900226e-05,
      "loss": 0.6335,
      "step": 8200
    },
    {
      "epoch": 1.0285500561027303,
      "grad_norm": 1.96578049659729,
      "learning_rate": 1.5667676787493148e-05,
      "loss": 0.6448,
      "step": 8250
    },
    {
      "epoch": 1.0347836928063832,
      "grad_norm": 1.8349359035491943,
      "learning_rate": 1.5610929219760715e-05,
      "loss": 0.6375,
      "step": 8300
    },
    {
      "epoch": 1.0410173295100362,
      "grad_norm": 2.0022618770599365,
      "learning_rate": 1.5553916705410347e-05,
      "loss": 0.6385,
      "step": 8350
    },
    {
      "epoch": 1.047250966213689,
      "grad_norm": 2.030137062072754,
      "learning_rate": 1.5496641936558135e-05,
      "loss": 0.6417,
      "step": 8400
    },
    {
      "epoch": 1.053484602917342,
      "grad_norm": 2.004664182662964,
      "learning_rate": 1.543910761770377e-05,
      "loss": 0.6429,
      "step": 8450
    },
    {
      "epoch": 1.059718239620995,
      "grad_norm": 1.9750394821166992,
      "learning_rate": 1.5381316465602808e-05,
      "loss": 0.6401,
      "step": 8500
    },
    {
      "epoch": 1.0659518763246478,
      "grad_norm": 2.0573947429656982,
      "learning_rate": 1.532327120913843e-05,
      "loss": 0.6395,
      "step": 8550
    },
    {
      "epoch": 1.0721855130283007,
      "grad_norm": 2.017814874649048,
      "learning_rate": 1.526497458919253e-05,
      "loss": 0.6259,
      "step": 8600
    },
    {
      "epoch": 1.0784191497319535,
      "grad_norm": 2.02846622467041,
      "learning_rate": 1.5206429358516341e-05,
      "loss": 0.6349,
      "step": 8650
    },
    {
      "epoch": 1.0846527864356066,
      "grad_norm": 2.0966057777404785,
      "learning_rate": 1.5147638281600423e-05,
      "loss": 0.6324,
      "step": 8700
    },
    {
      "epoch": 1.0908864231392594,
      "grad_norm": 2.06284499168396,
      "learning_rate": 1.5088604134544135e-05,
      "loss": 0.6387,
      "step": 8750
    },
    {
      "epoch": 1.0971200598429123,
      "grad_norm": 2.2421064376831055,
      "learning_rate": 1.502932970492454e-05,
      "loss": 0.6457,
      "step": 8800
    },
    {
      "epoch": 1.1033536965465653,
      "grad_norm": 2.078819751739502,
      "learning_rate": 1.4969817791664779e-05,
      "loss": 0.6232,
      "step": 8850
    },
    {
      "epoch": 1.1095873332502182,
      "grad_norm": 2.146176338195801,
      "learning_rate": 1.4910071204901916e-05,
      "loss": 0.6296,
      "step": 8900
    },
    {
      "epoch": 1.115820969953871,
      "grad_norm": 2.0687825679779053,
      "learning_rate": 1.4850092765854233e-05,
      "loss": 0.6321,
      "step": 8950
    },
    {
      "epoch": 1.122054606657524,
      "grad_norm": 1.9350824356079102,
      "learning_rate": 1.4789885306688019e-05,
      "loss": 0.6372,
      "step": 9000
    },
    {
      "epoch": 1.128288243361177,
      "grad_norm": 2.09902286529541,
      "learning_rate": 1.4729451670383829e-05,
      "loss": 0.6345,
      "step": 9050
    },
    {
      "epoch": 1.1345218800648298,
      "grad_norm": 2.0889132022857666,
      "learning_rate": 1.4668794710602248e-05,
      "loss": 0.6488,
      "step": 9100
    },
    {
      "epoch": 1.1407555167684826,
      "grad_norm": 2.0531392097473145,
      "learning_rate": 1.4607917291549131e-05,
      "loss": 0.6333,
      "step": 9150
    },
    {
      "epoch": 1.1469891534721357,
      "grad_norm": 2.0206382274627686,
      "learning_rate": 1.4546822287840372e-05,
      "loss": 0.6364,
      "step": 9200
    },
    {
      "epoch": 1.1532227901757885,
      "grad_norm": 2.1208789348602295,
      "learning_rate": 1.4485512584366146e-05,
      "loss": 0.64,
      "step": 9250
    },
    {
      "epoch": 1.1594564268794414,
      "grad_norm": 1.9847346544265747,
      "learning_rate": 1.4423991076154704e-05,
      "loss": 0.614,
      "step": 9300
    },
    {
      "epoch": 1.1656900635830945,
      "grad_norm": 2.0352087020874023,
      "learning_rate": 1.436226066823566e-05,
      "loss": 0.6395,
      "step": 9350
    },
    {
      "epoch": 1.1719237002867473,
      "grad_norm": 2.1553215980529785,
      "learning_rate": 1.4300324275502806e-05,
      "loss": 0.63,
      "step": 9400
    },
    {
      "epoch": 1.1781573369904002,
      "grad_norm": 2.3195507526397705,
      "learning_rate": 1.4238184822576499e-05,
      "loss": 0.6287,
      "step": 9450
    },
    {
      "epoch": 1.184390973694053,
      "grad_norm": 2.0281853675842285,
      "learning_rate": 1.4175845243665536e-05,
      "loss": 0.6428,
      "step": 9500
    },
    {
      "epoch": 1.190624610397706,
      "grad_norm": 2.049612522125244,
      "learning_rate": 1.4113308482428617e-05,
      "loss": 0.6418,
      "step": 9550
    },
    {
      "epoch": 1.196858247101359,
      "grad_norm": 2.1924421787261963,
      "learning_rate": 1.4050577491835338e-05,
      "loss": 0.6374,
      "step": 9600
    },
    {
      "epoch": 1.2030918838050118,
      "grad_norm": 2.0243258476257324,
      "learning_rate": 1.3987655234026752e-05,
      "loss": 0.6312,
      "step": 9650
    },
    {
      "epoch": 1.2093255205086648,
      "grad_norm": 1.9988375902175903,
      "learning_rate": 1.39245446801755e-05,
      "loss": 0.638,
      "step": 9700
    },
    {
      "epoch": 1.2155591572123177,
      "grad_norm": 2.065078020095825,
      "learning_rate": 1.3861248810345516e-05,
      "loss": 0.638,
      "step": 9750
    },
    {
      "epoch": 1.2217927939159705,
      "grad_norm": 2.1999423503875732,
      "learning_rate": 1.3797770613351307e-05,
      "loss": 0.6343,
      "step": 9800
    },
    {
      "epoch": 1.2280264306196236,
      "grad_norm": 2.228478193283081,
      "learning_rate": 1.373411308661682e-05,
      "loss": 0.637,
      "step": 9850
    },
    {
      "epoch": 1.2342600673232764,
      "grad_norm": 2.0829901695251465,
      "learning_rate": 1.3670279236033902e-05,
      "loss": 0.6386,
      "step": 9900
    },
    {
      "epoch": 1.2404937040269293,
      "grad_norm": 2.0941994190216064,
      "learning_rate": 1.3606272075820374e-05,
      "loss": 0.6203,
      "step": 9950
    },
    {
      "epoch": 1.2467273407305823,
      "grad_norm": 2.094511032104492,
      "learning_rate": 1.3542094628377686e-05,
      "loss": 0.6347,
      "step": 10000
    },
    {
      "epoch": 1.2467273407305823,
      "eval_loss": 0.6406242251396179,
      "eval_runtime": 429.8845,
      "eval_samples_per_second": 42.646,
      "eval_steps_per_second": 5.332,
      "step": 10000
    },
    {
      "epoch": 1.2529609774342352,
      "grad_norm": 2.276191473007202,
      "learning_rate": 1.3477749924148206e-05,
      "loss": 0.6427,
      "step": 10050
    },
    {
      "epoch": 1.259194614137888,
      "grad_norm": 2.1274471282958984,
      "learning_rate": 1.3413241001472132e-05,
      "loss": 0.6292,
      "step": 10100
    },
    {
      "epoch": 1.2654282508415409,
      "grad_norm": 2.167529582977295,
      "learning_rate": 1.334857090644401e-05,
      "loss": 0.6347,
      "step": 10150
    },
    {
      "epoch": 1.271661887545194,
      "grad_norm": 1.9703667163848877,
      "learning_rate": 1.3283742692768892e-05,
      "loss": 0.6401,
      "step": 10200
    },
    {
      "epoch": 1.2778955242488468,
      "grad_norm": 2.1070733070373535,
      "learning_rate": 1.3218759421618172e-05,
      "loss": 0.639,
      "step": 10250
    },
    {
      "epoch": 1.2841291609524996,
      "grad_norm": 3.016781806945801,
      "learning_rate": 1.3153624161485001e-05,
      "loss": 0.6307,
      "step": 10300
    },
    {
      "epoch": 1.2903627976561527,
      "grad_norm": 2.169923782348633,
      "learning_rate": 1.308833998803942e-05,
      "loss": 0.6343,
      "step": 10350
    },
    {
      "epoch": 1.2965964343598055,
      "grad_norm": 2.0714709758758545,
      "learning_rate": 1.302290998398311e-05,
      "loss": 0.6438,
      "step": 10400
    },
    {
      "epoch": 1.3028300710634584,
      "grad_norm": 2.3308637142181396,
      "learning_rate": 1.295733723890384e-05,
      "loss": 0.6305,
      "step": 10450
    },
    {
      "epoch": 1.3090637077671112,
      "grad_norm": 2.1425728797912598,
      "learning_rate": 1.2891624849129572e-05,
      "loss": 0.6269,
      "step": 10500
    },
    {
      "epoch": 1.3152973444707643,
      "grad_norm": 2.1231236457824707,
      "learning_rate": 1.2825775917582257e-05,
      "loss": 0.6255,
      "step": 10550
    },
    {
      "epoch": 1.3215309811744171,
      "grad_norm": 2.166088581085205,
      "learning_rate": 1.2759793553631307e-05,
      "loss": 0.6231,
      "step": 10600
    },
    {
      "epoch": 1.32776461787807,
      "grad_norm": 2.2344117164611816,
      "learning_rate": 1.2693680872946787e-05,
      "loss": 0.6371,
      "step": 10650
    },
    {
      "epoch": 1.333998254581723,
      "grad_norm": 2.150299549102783,
      "learning_rate": 1.2627440997352269e-05,
      "loss": 0.6306,
      "step": 10700
    },
    {
      "epoch": 1.340231891285376,
      "grad_norm": 2.1095736026763916,
      "learning_rate": 1.256107705467745e-05,
      "loss": 0.6377,
      "step": 10750
    },
    {
      "epoch": 1.3464655279890287,
      "grad_norm": 2.1861791610717773,
      "learning_rate": 1.2494592178610438e-05,
      "loss": 0.6322,
      "step": 10800
    },
    {
      "epoch": 1.3526991646926816,
      "grad_norm": 2.164217710494995,
      "learning_rate": 1.2427989508549781e-05,
      "loss": 0.6362,
      "step": 10850
    },
    {
      "epoch": 1.3589328013963347,
      "grad_norm": 2.4076497554779053,
      "learning_rate": 1.236127218945623e-05,
      "loss": 0.6293,
      "step": 10900
    },
    {
      "epoch": 1.3651664380999875,
      "grad_norm": 2.2930872440338135,
      "learning_rate": 1.2294443371704237e-05,
      "loss": 0.6395,
      "step": 10950
    },
    {
      "epoch": 1.3714000748036406,
      "grad_norm": 2.1939048767089844,
      "learning_rate": 1.2227506210933187e-05,
      "loss": 0.6285,
      "step": 11000
    },
    {
      "epoch": 1.3776337115072934,
      "grad_norm": 2.297088623046875,
      "learning_rate": 1.2160463867898398e-05,
      "loss": 0.6319,
      "step": 11050
    },
    {
      "epoch": 1.3838673482109463,
      "grad_norm": 2.3358867168426514,
      "learning_rate": 1.2093319508321863e-05,
      "loss": 0.6304,
      "step": 11100
    },
    {
      "epoch": 1.390100984914599,
      "grad_norm": 2.369279146194458,
      "learning_rate": 1.2026076302742778e-05,
      "loss": 0.6282,
      "step": 11150
    },
    {
      "epoch": 1.3963346216182522,
      "grad_norm": 1.9688912630081177,
      "learning_rate": 1.1958737426367806e-05,
      "loss": 0.6295,
      "step": 11200
    },
    {
      "epoch": 1.402568258321905,
      "grad_norm": 2.2516233921051025,
      "learning_rate": 1.1891306058921178e-05,
      "loss": 0.6198,
      "step": 11250
    },
    {
      "epoch": 1.4088018950255579,
      "grad_norm": 2.360363721847534,
      "learning_rate": 1.1823785384494515e-05,
      "loss": 0.624,
      "step": 11300
    },
    {
      "epoch": 1.415035531729211,
      "grad_norm": 2.1417112350463867,
      "learning_rate": 1.1756178591396499e-05,
      "loss": 0.6359,
      "step": 11350
    },
    {
      "epoch": 1.4212691684328638,
      "grad_norm": 2.332770824432373,
      "learning_rate": 1.168848887200231e-05,
      "loss": 0.633,
      "step": 11400
    },
    {
      "epoch": 1.4275028051365166,
      "grad_norm": 2.2573702335357666,
      "learning_rate": 1.162071942260289e-05,
      "loss": 0.6308,
      "step": 11450
    },
    {
      "epoch": 1.4337364418401695,
      "grad_norm": 2.218397617340088,
      "learning_rate": 1.1552873443254002e-05,
      "loss": 0.6335,
      "step": 11500
    },
    {
      "epoch": 1.4399700785438225,
      "grad_norm": 2.308323860168457,
      "learning_rate": 1.1484954137625141e-05,
      "loss": 0.6326,
      "step": 11550
    },
    {
      "epoch": 1.4462037152474754,
      "grad_norm": 2.1636953353881836,
      "learning_rate": 1.1416964712848248e-05,
      "loss": 0.6315,
      "step": 11600
    },
    {
      "epoch": 1.4524373519511282,
      "grad_norm": 2.222553014755249,
      "learning_rate": 1.1348908379366275e-05,
      "loss": 0.6199,
      "step": 11650
    },
    {
      "epoch": 1.4586709886547813,
      "grad_norm": 2.2691802978515625,
      "learning_rate": 1.1280788350781583e-05,
      "loss": 0.6254,
      "step": 11700
    },
    {
      "epoch": 1.4649046253584341,
      "grad_norm": 2.147172689437866,
      "learning_rate": 1.121260784370419e-05,
      "loss": 0.6204,
      "step": 11750
    },
    {
      "epoch": 1.471138262062087,
      "grad_norm": 2.095039129257202,
      "learning_rate": 1.1144370077599908e-05,
      "loss": 0.6276,
      "step": 11800
    },
    {
      "epoch": 1.4773718987657398,
      "grad_norm": 2.2065765857696533,
      "learning_rate": 1.10760782746383e-05,
      "loss": 0.6232,
      "step": 11850
    },
    {
      "epoch": 1.4836055354693929,
      "grad_norm": 2.236257791519165,
      "learning_rate": 1.1007735659540531e-05,
      "loss": 0.6222,
      "step": 11900
    },
    {
      "epoch": 1.4898391721730457,
      "grad_norm": 2.223972797393799,
      "learning_rate": 1.0939345459427106e-05,
      "loss": 0.6277,
      "step": 11950
    },
    {
      "epoch": 1.4960728088766988,
      "grad_norm": 2.3945045471191406,
      "learning_rate": 1.0870910903665479e-05,
      "loss": 0.6338,
      "step": 12000
    },
    {
      "epoch": 1.4960728088766988,
      "eval_loss": 0.6342150568962097,
      "eval_runtime": 428.7976,
      "eval_samples_per_second": 42.754,
      "eval_steps_per_second": 5.345,
      "step": 12000
    },
    {
      "epoch": 1.5023064455803516,
      "grad_norm": 2.4395430088043213,
      "learning_rate": 1.080243522371757e-05,
      "loss": 0.6228,
      "step": 12050
    },
    {
      "epoch": 1.5085400822840045,
      "grad_norm": 2.184892177581787,
      "learning_rate": 1.0733921652987172e-05,
      "loss": 0.6322,
      "step": 12100
    },
    {
      "epoch": 1.5147737189876573,
      "grad_norm": 2.2245593070983887,
      "learning_rate": 1.0665373426667264e-05,
      "loss": 0.6284,
      "step": 12150
    },
    {
      "epoch": 1.5210073556913102,
      "grad_norm": 2.3077852725982666,
      "learning_rate": 1.0596793781587264e-05,
      "loss": 0.6196,
      "step": 12200
    },
    {
      "epoch": 1.5272409923949632,
      "grad_norm": 2.247385025024414,
      "learning_rate": 1.0528185956060173e-05,
      "loss": 0.6268,
      "step": 12250
    },
    {
      "epoch": 1.533474629098616,
      "grad_norm": 2.339890718460083,
      "learning_rate": 1.045955318972965e-05,
      "loss": 0.6333,
      "step": 12300
    },
    {
      "epoch": 1.5397082658022692,
      "grad_norm": 2.2456533908843994,
      "learning_rate": 1.0390898723417073e-05,
      "loss": 0.63,
      "step": 12350
    },
    {
      "epoch": 1.545941902505922,
      "grad_norm": 2.3100922107696533,
      "learning_rate": 1.0322225798968482e-05,
      "loss": 0.6285,
      "step": 12400
    },
    {
      "epoch": 1.5521755392095749,
      "grad_norm": 2.221842050552368,
      "learning_rate": 1.0253537659101495e-05,
      "loss": 0.6189,
      "step": 12450
    },
    {
      "epoch": 1.5584091759132277,
      "grad_norm": 2.4616003036499023,
      "learning_rate": 1.0184837547252213e-05,
      "loss": 0.6195,
      "step": 12500
    },
    {
      "epoch": 1.5646428126168805,
      "grad_norm": 2.494037628173828,
      "learning_rate": 1.0116128707422039e-05,
      "loss": 0.6185,
      "step": 12550
    },
    {
      "epoch": 1.5708764493205336,
      "grad_norm": 2.238983631134033,
      "learning_rate": 1.0047414384024513e-05,
      "loss": 0.6213,
      "step": 12600
    },
    {
      "epoch": 1.5771100860241867,
      "grad_norm": 2.3886332511901855,
      "learning_rate": 9.978697821732109e-06,
      "loss": 0.6222,
      "step": 12650
    },
    {
      "epoch": 1.5833437227278395,
      "grad_norm": 2.3459067344665527,
      "learning_rate": 9.90998226532302e-06,
      "loss": 0.6287,
      "step": 12700
    },
    {
      "epoch": 1.5895773594314924,
      "grad_norm": 2.3928728103637695,
      "learning_rate": 9.841270959527945e-06,
      "loss": 0.6222,
      "step": 12750
    },
    {
      "epoch": 1.5958109961351452,
      "grad_norm": 2.266845941543579,
      "learning_rate": 9.772567148876859e-06,
      "loss": 0.6212,
      "step": 12800
    },
    {
      "epoch": 1.602044632838798,
      "grad_norm": 2.4075541496276855,
      "learning_rate": 9.703874077545837e-06,
      "loss": 0.6283,
      "step": 12850
    },
    {
      "epoch": 1.608278269542451,
      "grad_norm": 2.3668038845062256,
      "learning_rate": 9.635194989203822e-06,
      "loss": 0.6182,
      "step": 12900
    },
    {
      "epoch": 1.614511906246104,
      "grad_norm": 2.2269511222839355,
      "learning_rate": 9.566533126859509e-06,
      "loss": 0.6247,
      "step": 12950
    },
    {
      "epoch": 1.620745542949757,
      "grad_norm": 2.3069663047790527,
      "learning_rate": 9.497891732708168e-06,
      "loss": 0.6274,
      "step": 13000
    },
    {
      "epoch": 1.6269791796534099,
      "grad_norm": 2.493015766143799,
      "learning_rate": 9.429274047978574e-06,
      "loss": 0.6359,
      "step": 13050
    },
    {
      "epoch": 1.6332128163570627,
      "grad_norm": 2.506463050842285,
      "learning_rate": 9.360683312779942e-06,
      "loss": 0.6304,
      "step": 13100
    },
    {
      "epoch": 1.6394464530607156,
      "grad_norm": 2.2598812580108643,
      "learning_rate": 9.29212276594895e-06,
      "loss": 0.6285,
      "step": 13150
    },
    {
      "epoch": 1.6456800897643684,
      "grad_norm": 2.741260290145874,
      "learning_rate": 9.223595644896773e-06,
      "loss": 0.6278,
      "step": 13200
    },
    {
      "epoch": 1.6519137264680215,
      "grad_norm": 2.3762969970703125,
      "learning_rate": 9.15510518545625e-06,
      "loss": 0.622,
      "step": 13250
    },
    {
      "epoch": 1.6581473631716743,
      "grad_norm": 2.4215378761291504,
      "learning_rate": 9.086654621729046e-06,
      "loss": 0.6281,
      "step": 13300
    },
    {
      "epoch": 1.6643809998753274,
      "grad_norm": 2.4529707431793213,
      "learning_rate": 9.018247185932973e-06,
      "loss": 0.6276,
      "step": 13350
    },
    {
      "epoch": 1.6706146365789802,
      "grad_norm": 2.2953782081604004,
      "learning_rate": 8.949886108249358e-06,
      "loss": 0.6194,
      "step": 13400
    },
    {
      "epoch": 1.676848273282633,
      "grad_norm": 2.3801350593566895,
      "learning_rate": 8.881574616670493e-06,
      "loss": 0.6258,
      "step": 13450
    },
    {
      "epoch": 1.683081909986286,
      "grad_norm": 2.409956216812134,
      "learning_rate": 8.813315936847247e-06,
      "loss": 0.6229,
      "step": 13500
    },
    {
      "epoch": 1.6893155466899388,
      "grad_norm": 2.444596529006958,
      "learning_rate": 8.745113291936718e-06,
      "loss": 0.6194,
      "step": 13550
    },
    {
      "epoch": 1.6955491833935918,
      "grad_norm": 2.469391345977783,
      "learning_rate": 8.676969902450054e-06,
      "loss": 0.6291,
      "step": 13600
    },
    {
      "epoch": 1.7017828200972447,
      "grad_norm": 2.2395176887512207,
      "learning_rate": 8.608888986100374e-06,
      "loss": 0.6185,
      "step": 13650
    },
    {
      "epoch": 1.7080164568008978,
      "grad_norm": 2.3262693881988525,
      "learning_rate": 8.540873757650845e-06,
      "loss": 0.6261,
      "step": 13700
    },
    {
      "epoch": 1.7142500935045506,
      "grad_norm": 2.501307725906372,
      "learning_rate": 8.472927428762845e-06,
      "loss": 0.6272,
      "step": 13750
    },
    {
      "epoch": 1.7204837302082034,
      "grad_norm": 2.3408379554748535,
      "learning_rate": 8.405053207844358e-06,
      "loss": 0.6256,
      "step": 13800
    },
    {
      "epoch": 1.7267173669118563,
      "grad_norm": 2.381819009780884,
      "learning_rate": 8.337254299898432e-06,
      "loss": 0.6286,
      "step": 13850
    },
    {
      "epoch": 1.7329510036155091,
      "grad_norm": 2.5323784351348877,
      "learning_rate": 8.269533906371862e-06,
      "loss": 0.6326,
      "step": 13900
    },
    {
      "epoch": 1.7391846403191622,
      "grad_norm": 2.52728271484375,
      "learning_rate": 8.201895225004004e-06,
      "loss": 0.6132,
      "step": 13950
    },
    {
      "epoch": 1.7454182770228153,
      "grad_norm": 2.3558783531188965,
      "learning_rate": 8.134341449675802e-06,
      "loss": 0.6128,
      "step": 14000
    },
    {
      "epoch": 1.7454182770228153,
      "eval_loss": 0.6293798089027405,
      "eval_runtime": 428.4311,
      "eval_samples_per_second": 42.791,
      "eval_steps_per_second": 5.35,
      "step": 14000
    },
    {
      "epoch": 1.7516519137264681,
      "grad_norm": 2.5308268070220947,
      "learning_rate": 8.066875770258952e-06,
      "loss": 0.6131,
      "step": 14050
    },
    {
      "epoch": 1.757885550430121,
      "grad_norm": 2.5065295696258545,
      "learning_rate": 7.99950137246528e-06,
      "loss": 0.6248,
      "step": 14100
    },
    {
      "epoch": 1.7641191871337738,
      "grad_norm": 2.4519286155700684,
      "learning_rate": 7.932221437696324e-06,
      "loss": 0.6243,
      "step": 14150
    },
    {
      "epoch": 1.7703528238374266,
      "grad_norm": 2.4246177673339844,
      "learning_rate": 7.865039142893091e-06,
      "loss": 0.6167,
      "step": 14200
    },
    {
      "epoch": 1.7765864605410797,
      "grad_norm": 2.379155397415161,
      "learning_rate": 7.797957660386072e-06,
      "loss": 0.6176,
      "step": 14250
    },
    {
      "epoch": 1.7828200972447326,
      "grad_norm": 2.446007490158081,
      "learning_rate": 7.730980157745414e-06,
      "loss": 0.6204,
      "step": 14300
    },
    {
      "epoch": 1.7890537339483856,
      "grad_norm": 2.490818500518799,
      "learning_rate": 7.664109797631365e-06,
      "loss": 0.6274,
      "step": 14350
    },
    {
      "epoch": 1.7952873706520385,
      "grad_norm": 2.4037506580352783,
      "learning_rate": 7.5973497376449304e-06,
      "loss": 0.6226,
      "step": 14400
    },
    {
      "epoch": 1.8015210073556913,
      "grad_norm": 2.4374518394470215,
      "learning_rate": 7.530703130178781e-06,
      "loss": 0.615,
      "step": 14450
    },
    {
      "epoch": 1.8077546440593442,
      "grad_norm": 2.554844379425049,
      "learning_rate": 7.46417312226837e-06,
      "loss": 0.6249,
      "step": 14500
    },
    {
      "epoch": 1.813988280762997,
      "grad_norm": 2.4678516387939453,
      "learning_rate": 7.397762855443374e-06,
      "loss": 0.6176,
      "step": 14550
    },
    {
      "epoch": 1.82022191746665,
      "grad_norm": 2.3095438480377197,
      "learning_rate": 7.331475465579303e-06,
      "loss": 0.6219,
      "step": 14600
    },
    {
      "epoch": 1.826455554170303,
      "grad_norm": 2.4066290855407715,
      "learning_rate": 7.265314082749471e-06,
      "loss": 0.6204,
      "step": 14650
    },
    {
      "epoch": 1.832689190873956,
      "grad_norm": 2.600909948348999,
      "learning_rate": 7.199281831077148e-06,
      "loss": 0.6232,
      "step": 14700
    },
    {
      "epoch": 1.8389228275776088,
      "grad_norm": 2.596193313598633,
      "learning_rate": 7.133381828588088e-06,
      "loss": 0.6239,
      "step": 14750
    },
    {
      "epoch": 1.8451564642812617,
      "grad_norm": 2.5088484287261963,
      "learning_rate": 7.0676171870632646e-06,
      "loss": 0.6209,
      "step": 14800
    },
    {
      "epoch": 1.8513901009849145,
      "grad_norm": 2.525998115539551,
      "learning_rate": 7.001991011891936e-06,
      "loss": 0.6182,
      "step": 14850
    },
    {
      "epoch": 1.8576237376885674,
      "grad_norm": 2.4280805587768555,
      "learning_rate": 6.93650640192502e-06,
      "loss": 0.6242,
      "step": 14900
    },
    {
      "epoch": 1.8638573743922204,
      "grad_norm": 2.3831937313079834,
      "learning_rate": 6.871166449328759e-06,
      "loss": 0.6278,
      "step": 14950
    },
    {
      "epoch": 1.8700910110958733,
      "grad_norm": 2.6162397861480713,
      "learning_rate": 6.8059742394387215e-06,
      "loss": 0.6106,
      "step": 15000
    },
    {
      "epoch": 1.8763246477995263,
      "grad_norm": 2.536106824874878,
      "learning_rate": 6.74093285061409e-06,
      "loss": 0.6217,
      "step": 15050
    },
    {
      "epoch": 1.8825582845031792,
      "grad_norm": 2.3776044845581055,
      "learning_rate": 6.67604535409233e-06,
      "loss": 0.6217,
      "step": 15100
    },
    {
      "epoch": 1.888791921206832,
      "grad_norm": 2.582655906677246,
      "learning_rate": 6.611314813844139e-06,
      "loss": 0.6217,
      "step": 15150
    },
    {
      "epoch": 1.8950255579104849,
      "grad_norm": 2.510911703109741,
      "learning_rate": 6.54674428642879e-06,
      "loss": 0.6204,
      "step": 15200
    },
    {
      "epoch": 1.9012591946141377,
      "grad_norm": 2.5079355239868164,
      "learning_rate": 6.482336820849784e-06,
      "loss": 0.6283,
      "step": 15250
    },
    {
      "epoch": 1.9074928313177908,
      "grad_norm": 2.548069953918457,
      "learning_rate": 6.418095458410894e-06,
      "loss": 0.6333,
      "step": 15300
    },
    {
      "epoch": 1.9137264680214439,
      "grad_norm": 2.4146578311920166,
      "learning_rate": 6.3540232325725325e-06,
      "loss": 0.6284,
      "step": 15350
    },
    {
      "epoch": 1.9199601047250967,
      "grad_norm": 2.556776523590088,
      "learning_rate": 6.2901231688085416e-06,
      "loss": 0.6209,
      "step": 15400
    },
    {
      "epoch": 1.9261937414287496,
      "grad_norm": 2.635026216506958,
      "learning_rate": 6.226398284463306e-06,
      "loss": 0.6221,
      "step": 15450
    },
    {
      "epoch": 1.9324273781324024,
      "grad_norm": 2.40042781829834,
      "learning_rate": 6.16285158860928e-06,
      "loss": 0.6247,
      "step": 15500
    },
    {
      "epoch": 1.9386610148360552,
      "grad_norm": 2.5498788356781006,
      "learning_rate": 6.099486081904914e-06,
      "loss": 0.6243,
      "step": 15550
    },
    {
      "epoch": 1.9448946515397083,
      "grad_norm": 2.656203031539917,
      "learning_rate": 6.036304756452942e-06,
      "loss": 0.6157,
      "step": 15600
    },
    {
      "epoch": 1.9511282882433612,
      "grad_norm": 2.546098470687866,
      "learning_rate": 5.973310595659123e-06,
      "loss": 0.6172,
      "step": 15650
    },
    {
      "epoch": 1.9573619249470142,
      "grad_norm": 2.7292096614837646,
      "learning_rate": 5.91050657409133e-06,
      "loss": 0.6175,
      "step": 15700
    },
    {
      "epoch": 1.963595561650667,
      "grad_norm": 2.2817304134368896,
      "learning_rate": 5.847895657339131e-06,
      "loss": 0.6225,
      "step": 15750
    },
    {
      "epoch": 1.96982919835432,
      "grad_norm": 2.702178955078125,
      "learning_rate": 5.785480801873717e-06,
      "loss": 0.6122,
      "step": 15800
    },
    {
      "epoch": 1.9760628350579728,
      "grad_norm": 2.6388230323791504,
      "learning_rate": 5.723264954908329e-06,
      "loss": 0.6253,
      "step": 15850
    },
    {
      "epoch": 1.9822964717616256,
      "grad_norm": 2.618553876876831,
      "learning_rate": 5.661251054259072e-06,
      "loss": 0.6176,
      "step": 15900
    },
    {
      "epoch": 1.9885301084652787,
      "grad_norm": 2.5211663246154785,
      "learning_rate": 5.5994420282062e-06,
      "loss": 0.6207,
      "step": 15950
    },
    {
      "epoch": 1.9947637451689315,
      "grad_norm": 2.5682690143585205,
      "learning_rate": 5.537840795355844e-06,
      "loss": 0.6241,
      "step": 16000
    },
    {
      "epoch": 1.9947637451689315,
      "eval_loss": 0.6254520416259766,
      "eval_runtime": 428.5136,
      "eval_samples_per_second": 42.783,
      "eval_steps_per_second": 5.349,
      "step": 16000
    },
    {
      "epoch": 2.0009973818725846,
      "grad_norm": 2.504669427871704,
      "learning_rate": 5.47645026450219e-06,
      "loss": 0.6126,
      "step": 16050
    },
    {
      "epoch": 2.0072310185762374,
      "grad_norm": 2.5481069087982178,
      "learning_rate": 5.4152733344901344e-06,
      "loss": 0.6096,
      "step": 16100
    },
    {
      "epoch": 2.0134646552798903,
      "grad_norm": 2.849485158920288,
      "learning_rate": 5.354312894078395e-06,
      "loss": 0.6096,
      "step": 16150
    },
    {
      "epoch": 2.019698291983543,
      "grad_norm": 2.6451196670532227,
      "learning_rate": 5.293571821803107e-06,
      "loss": 0.6091,
      "step": 16200
    },
    {
      "epoch": 2.025931928687196,
      "grad_norm": 2.602538824081421,
      "learning_rate": 5.2330529858419e-06,
      "loss": 0.6011,
      "step": 16250
    },
    {
      "epoch": 2.032165565390849,
      "grad_norm": 2.7579050064086914,
      "learning_rate": 5.172759243878465e-06,
      "loss": 0.6168,
      "step": 16300
    },
    {
      "epoch": 2.038399202094502,
      "grad_norm": 2.7441816329956055,
      "learning_rate": 5.112693442967606e-06,
      "loss": 0.6143,
      "step": 16350
    },
    {
      "epoch": 2.044632838798155,
      "grad_norm": 2.976849317550659,
      "learning_rate": 5.052858419400823e-06,
      "loss": 0.6057,
      "step": 16400
    },
    {
      "epoch": 2.050866475501808,
      "grad_norm": 2.8315844535827637,
      "learning_rate": 4.993256998572349e-06,
      "loss": 0.6135,
      "step": 16450
    },
    {
      "epoch": 2.0571001122054606,
      "grad_norm": 2.6977603435516357,
      "learning_rate": 4.933891994845781e-06,
      "loss": 0.6061,
      "step": 16500
    },
    {
      "epoch": 2.0633337489091135,
      "grad_norm": 2.6211798191070557,
      "learning_rate": 4.874766211421137e-06,
      "loss": 0.612,
      "step": 16550
    },
    {
      "epoch": 2.0695673856127663,
      "grad_norm": 2.8271079063415527,
      "learning_rate": 4.815882440202541e-06,
      "loss": 0.6003,
      "step": 16600
    },
    {
      "epoch": 2.0758010223164196,
      "grad_norm": 2.8569369316101074,
      "learning_rate": 4.757243461666341e-06,
      "loss": 0.615,
      "step": 16650
    },
    {
      "epoch": 2.0820346590200725,
      "grad_norm": 2.6585402488708496,
      "learning_rate": 4.698852044729848e-06,
      "loss": 0.6157,
      "step": 16700
    },
    {
      "epoch": 2.0882682957237253,
      "grad_norm": 2.931396722793579,
      "learning_rate": 4.640710946620579e-06,
      "loss": 0.6058,
      "step": 16750
    },
    {
      "epoch": 2.094501932427378,
      "grad_norm": 3.071336269378662,
      "learning_rate": 4.58282291274606e-06,
      "loss": 0.61,
      "step": 16800
    },
    {
      "epoch": 2.100735569131031,
      "grad_norm": 2.8899269104003906,
      "learning_rate": 4.525190676564189e-06,
      "loss": 0.6189,
      "step": 16850
    },
    {
      "epoch": 2.106969205834684,
      "grad_norm": 2.747736692428589,
      "learning_rate": 4.467816959454166e-06,
      "loss": 0.6087,
      "step": 16900
    },
    {
      "epoch": 2.1132028425383367,
      "grad_norm": 2.875563383102417,
      "learning_rate": 4.4107044705879835e-06,
      "loss": 0.6134,
      "step": 16950
    },
    {
      "epoch": 2.11943647924199,
      "grad_norm": 3.1397616863250732,
      "learning_rate": 4.353855906802508e-06,
      "loss": 0.6153,
      "step": 17000
    },
    {
      "epoch": 2.125670115945643,
      "grad_norm": 2.7929415702819824,
      "learning_rate": 4.297273952472128e-06,
      "loss": 0.6097,
      "step": 17050
    },
    {
      "epoch": 2.1319037526492957,
      "grad_norm": 2.733863592147827,
      "learning_rate": 4.24096127938201e-06,
      "loss": 0.6176,
      "step": 17100
    },
    {
      "epoch": 2.1381373893529485,
      "grad_norm": 3.0999526977539062,
      "learning_rate": 4.184920546601927e-06,
      "loss": 0.6186,
      "step": 17150
    },
    {
      "epoch": 2.1443710260566013,
      "grad_norm": 2.7846946716308594,
      "learning_rate": 4.129154400360691e-06,
      "loss": 0.6124,
      "step": 17200
    },
    {
      "epoch": 2.150604662760254,
      "grad_norm": 2.8064780235290527,
      "learning_rate": 4.073665473921232e-06,
      "loss": 0.5992,
      "step": 17250
    },
    {
      "epoch": 2.156838299463907,
      "grad_norm": 2.956892967224121,
      "learning_rate": 4.018456387456207e-06,
      "loss": 0.6029,
      "step": 17300
    },
    {
      "epoch": 2.1630719361675603,
      "grad_norm": 2.8882293701171875,
      "learning_rate": 3.963529747924326e-06,
      "loss": 0.6062,
      "step": 17350
    },
    {
      "epoch": 2.169305572871213,
      "grad_norm": 2.758193016052246,
      "learning_rate": 3.90888814894721e-06,
      "loss": 0.5965,
      "step": 17400
    },
    {
      "epoch": 2.175539209574866,
      "grad_norm": 2.9970481395721436,
      "learning_rate": 3.854534170686943e-06,
      "loss": 0.6085,
      "step": 17450
    },
    {
      "epoch": 2.181772846278519,
      "grad_norm": 2.677720546722412,
      "learning_rate": 3.8004703797242514e-06,
      "loss": 0.6141,
      "step": 17500
    },
    {
      "epoch": 2.1880064829821717,
      "grad_norm": 2.7552404403686523,
      "learning_rate": 3.746699328937261e-06,
      "loss": 0.6096,
      "step": 17550
    },
    {
      "epoch": 2.1942401196858246,
      "grad_norm": 2.9219369888305664,
      "learning_rate": 3.693223557381016e-06,
      "loss": 0.6074,
      "step": 17600
    },
    {
      "epoch": 2.200473756389478,
      "grad_norm": 2.8160207271575928,
      "learning_rate": 3.6400455901675248e-06,
      "loss": 0.6137,
      "step": 17650
    },
    {
      "epoch": 2.2067073930931307,
      "grad_norm": 2.738856792449951,
      "learning_rate": 3.5871679383465687e-06,
      "loss": 0.6057,
      "step": 17700
    },
    {
      "epoch": 2.2129410297967835,
      "grad_norm": 2.8775322437286377,
      "learning_rate": 3.534593098787107e-06,
      "loss": 0.6099,
      "step": 17750
    },
    {
      "epoch": 2.2191746665004364,
      "grad_norm": 2.7699577808380127,
      "learning_rate": 3.4823235540593857e-06,
      "loss": 0.6149,
      "step": 17800
    },
    {
      "epoch": 2.2254083032040892,
      "grad_norm": 2.9661619663238525,
      "learning_rate": 3.4303617723177085e-06,
      "loss": 0.6229,
      "step": 17850
    },
    {
      "epoch": 2.231641939907742,
      "grad_norm": 2.9372880458831787,
      "learning_rate": 3.3787102071838907e-06,
      "loss": 0.6093,
      "step": 17900
    },
    {
      "epoch": 2.237875576611395,
      "grad_norm": 2.9852278232574463,
      "learning_rate": 3.3273712976313966e-06,
      "loss": 0.6122,
      "step": 17950
    },
    {
      "epoch": 2.244109213315048,
      "grad_norm": 3.234891176223755,
      "learning_rate": 3.2763474678701847e-06,
      "loss": 0.615,
      "step": 18000
    },
    {
      "epoch": 2.244109213315048,
      "eval_loss": 0.6239316463470459,
      "eval_runtime": 429.4272,
      "eval_samples_per_second": 42.692,
      "eval_steps_per_second": 5.337,
      "step": 18000
    },
    {
      "epoch": 2.250342850018701,
      "grad_norm": 2.977307081222534,
      "learning_rate": 3.2256411272322097e-06,
      "loss": 0.6067,
      "step": 18050
    },
    {
      "epoch": 2.256576486722354,
      "grad_norm": 2.8029897212982178,
      "learning_rate": 3.175254670057698e-06,
      "loss": 0.6141,
      "step": 18100
    },
    {
      "epoch": 2.2628101234260067,
      "grad_norm": 2.885103702545166,
      "learning_rate": 3.125190475582034e-06,
      "loss": 0.6119,
      "step": 18150
    },
    {
      "epoch": 2.2690437601296596,
      "grad_norm": 2.944196939468384,
      "learning_rate": 3.0754509078234663e-06,
      "loss": 0.6102,
      "step": 18200
    },
    {
      "epoch": 2.2752773968333124,
      "grad_norm": 3.1355795860290527,
      "learning_rate": 3.0260383154714425e-06,
      "loss": 0.6101,
      "step": 18250
    },
    {
      "epoch": 2.2815110335369653,
      "grad_norm": 3.0765724182128906,
      "learning_rate": 2.9769550317757078e-06,
      "loss": 0.6127,
      "step": 18300
    },
    {
      "epoch": 2.2877446702406186,
      "grad_norm": 2.8139736652374268,
      "learning_rate": 2.9282033744361613e-06,
      "loss": 0.609,
      "step": 18350
    },
    {
      "epoch": 2.2939783069442714,
      "grad_norm": 3.026620626449585,
      "learning_rate": 2.8797856454933694e-06,
      "loss": 0.6092,
      "step": 18400
    },
    {
      "epoch": 2.3002119436479243,
      "grad_norm": 3.023926019668579,
      "learning_rate": 2.831704131219899e-06,
      "loss": 0.6079,
      "step": 18450
    },
    {
      "epoch": 2.306445580351577,
      "grad_norm": 2.937608480453491,
      "learning_rate": 2.7839611020123447e-06,
      "loss": 0.6134,
      "step": 18500
    },
    {
      "epoch": 2.31267921705523,
      "grad_norm": 3.0173606872558594,
      "learning_rate": 2.7365588122841227e-06,
      "loss": 0.6072,
      "step": 18550
    },
    {
      "epoch": 2.318912853758883,
      "grad_norm": 2.9301366806030273,
      "learning_rate": 2.689499500359022e-06,
      "loss": 0.6084,
      "step": 18600
    },
    {
      "epoch": 2.3251464904625356,
      "grad_norm": 3.1824910640716553,
      "learning_rate": 2.6427853883655085e-06,
      "loss": 0.6085,
      "step": 18650
    },
    {
      "epoch": 2.331380127166189,
      "grad_norm": 3.2777152061462402,
      "learning_rate": 2.5964186821317963e-06,
      "loss": 0.6161,
      "step": 18700
    },
    {
      "epoch": 2.3376137638698418,
      "grad_norm": 2.7468907833099365,
      "learning_rate": 2.550401571081692e-06,
      "loss": 0.606,
      "step": 18750
    },
    {
      "epoch": 2.3438474005734946,
      "grad_norm": 3.156879425048828,
      "learning_rate": 2.5047362281312004e-06,
      "loss": 0.6028,
      "step": 18800
    },
    {
      "epoch": 2.3500810372771475,
      "grad_norm": 2.995659112930298,
      "learning_rate": 2.459424809585943e-06,
      "loss": 0.6155,
      "step": 18850
    },
    {
      "epoch": 2.3563146739808003,
      "grad_norm": 3.141301393508911,
      "learning_rate": 2.41446945503931e-06,
      "loss": 0.5997,
      "step": 18900
    },
    {
      "epoch": 2.362548310684453,
      "grad_norm": 3.1456117630004883,
      "learning_rate": 2.3698722872714486e-06,
      "loss": 0.6206,
      "step": 18950
    },
    {
      "epoch": 2.368781947388106,
      "grad_norm": 3.0462348461151123,
      "learning_rate": 2.3256354121490197e-06,
      "loss": 0.6134,
      "step": 19000
    },
    {
      "epoch": 2.3750155840917593,
      "grad_norm": 3.090392827987671,
      "learning_rate": 2.2817609185257493e-06,
      "loss": 0.6028,
      "step": 19050
    },
    {
      "epoch": 2.381249220795412,
      "grad_norm": 2.762963056564331,
      "learning_rate": 2.2382508781438217e-06,
      "loss": 0.6119,
      "step": 19100
    },
    {
      "epoch": 2.387482857499065,
      "grad_norm": 2.848215341567993,
      "learning_rate": 2.195107345536013e-06,
      "loss": 0.6105,
      "step": 19150
    },
    {
      "epoch": 2.393716494202718,
      "grad_norm": 2.9583053588867188,
      "learning_rate": 2.152332357928719e-06,
      "loss": 0.5955,
      "step": 19200
    },
    {
      "epoch": 2.3999501309063707,
      "grad_norm": 2.798417568206787,
      "learning_rate": 2.109927935145718e-06,
      "loss": 0.6094,
      "step": 19250
    },
    {
      "epoch": 2.4061837676100235,
      "grad_norm": 3.254685878753662,
      "learning_rate": 2.0678960795128234e-06,
      "loss": 0.6026,
      "step": 19300
    },
    {
      "epoch": 2.4124174043136764,
      "grad_norm": 3.068819522857666,
      "learning_rate": 2.026238775763322e-06,
      "loss": 0.6101,
      "step": 19350
    },
    {
      "epoch": 2.4186510410173296,
      "grad_norm": 2.981771469116211,
      "learning_rate": 1.9849579909442595e-06,
      "loss": 0.6117,
      "step": 19400
    },
    {
      "epoch": 2.4248846777209825,
      "grad_norm": 3.050567626953125,
      "learning_rate": 1.944055674323554e-06,
      "loss": 0.6097,
      "step": 19450
    },
    {
      "epoch": 2.4311183144246353,
      "grad_norm": 2.9681475162506104,
      "learning_rate": 1.9035337572979561e-06,
      "loss": 0.6116,
      "step": 19500
    },
    {
      "epoch": 2.437351951128288,
      "grad_norm": 3.2052080631256104,
      "learning_rate": 1.8633941533018428e-06,
      "loss": 0.6112,
      "step": 19550
    },
    {
      "epoch": 2.443585587831941,
      "grad_norm": 3.2845706939697266,
      "learning_rate": 1.8236387577168735e-06,
      "loss": 0.6076,
      "step": 19600
    },
    {
      "epoch": 2.4498192245355943,
      "grad_norm": 3.319469928741455,
      "learning_rate": 1.784269447782484e-06,
      "loss": 0.6087,
      "step": 19650
    },
    {
      "epoch": 2.456052861239247,
      "grad_norm": 3.161952018737793,
      "learning_rate": 1.7452880825072448e-06,
      "loss": 0.6013,
      "step": 19700
    },
    {
      "epoch": 2.4622864979429,
      "grad_norm": 2.970977306365967,
      "learning_rate": 1.7066965025810844e-06,
      "loss": 0.6126,
      "step": 19750
    },
    {
      "epoch": 2.468520134646553,
      "grad_norm": 3.3727195262908936,
      "learning_rate": 1.668496530288366e-06,
      "loss": 0.6092,
      "step": 19800
    },
    {
      "epoch": 2.4747537713502057,
      "grad_norm": 3.2813334465026855,
      "learning_rate": 1.6306899694218436e-06,
      "loss": 0.6085,
      "step": 19850
    },
    {
      "epoch": 2.4809874080538585,
      "grad_norm": 3.0674593448638916,
      "learning_rate": 1.5932786051974792e-06,
      "loss": 0.6146,
      "step": 19900
    },
    {
      "epoch": 2.4872210447575114,
      "grad_norm": 2.9783222675323486,
      "learning_rate": 1.556264204170167e-06,
      "loss": 0.6098,
      "step": 19950
    },
    {
      "epoch": 2.4934546814611647,
      "grad_norm": 3.0052356719970703,
      "learning_rate": 1.519648514150286e-06,
      "loss": 0.6121,
      "step": 20000
    },
    {
      "epoch": 2.4934546814611647,
      "eval_loss": 0.622496485710144,
      "eval_runtime": 428.831,
      "eval_samples_per_second": 42.751,
      "eval_steps_per_second": 5.345,
      "step": 20000
    },
    {
      "epoch": 2.4996883181648175,
      "grad_norm": 3.3930184841156006,
      "learning_rate": 1.4834332641211956e-06,
      "loss": 0.6103,
      "step": 20050
    },
    {
      "epoch": 2.5059219548684704,
      "grad_norm": 3.2993013858795166,
      "learning_rate": 1.4476201641575793e-06,
      "loss": 0.6109,
      "step": 20100
    },
    {
      "epoch": 2.512155591572123,
      "grad_norm": 3.2028093338012695,
      "learning_rate": 1.4122109053446997e-06,
      "loss": 0.6199,
      "step": 20150
    },
    {
      "epoch": 2.518389228275776,
      "grad_norm": 3.160646915435791,
      "learning_rate": 1.3772071596985448e-06,
      "loss": 0.6069,
      "step": 20200
    },
    {
      "epoch": 2.524622864979429,
      "grad_norm": 3.4864635467529297,
      "learning_rate": 1.3426105800868782e-06,
      "loss": 0.6061,
      "step": 20250
    },
    {
      "epoch": 2.5308565016830817,
      "grad_norm": 3.162372589111328,
      "learning_rate": 1.3084228001511867e-06,
      "loss": 0.6087,
      "step": 20300
    },
    {
      "epoch": 2.537090138386735,
      "grad_norm": 3.0803229808807373,
      "learning_rate": 1.2746454342295456e-06,
      "loss": 0.6061,
      "step": 20350
    },
    {
      "epoch": 2.543323775090388,
      "grad_norm": 3.3600547313690186,
      "learning_rate": 1.2412800772803846e-06,
      "loss": 0.6129,
      "step": 20400
    },
    {
      "epoch": 2.5495574117940407,
      "grad_norm": 3.196950674057007,
      "learning_rate": 1.208328304807178e-06,
      "loss": 0.6032,
      "step": 20450
    },
    {
      "epoch": 2.5557910484976936,
      "grad_norm": 3.2778985500335693,
      "learning_rate": 1.1757916727840502e-06,
      "loss": 0.6102,
      "step": 20500
    },
    {
      "epoch": 2.5620246852013464,
      "grad_norm": 3.335165023803711,
      "learning_rate": 1.1436717175822976e-06,
      "loss": 0.61,
      "step": 20550
    },
    {
      "epoch": 2.5682583219049993,
      "grad_norm": 3.118311882019043,
      "learning_rate": 1.1119699558978525e-06,
      "loss": 0.605,
      "step": 20600
    },
    {
      "epoch": 2.574491958608652,
      "grad_norm": 3.186270236968994,
      "learning_rate": 1.0806878846796454e-06,
      "loss": 0.615,
      "step": 20650
    },
    {
      "epoch": 2.5807255953123054,
      "grad_norm": 3.2636191844940186,
      "learning_rate": 1.0498269810589501e-06,
      "loss": 0.6069,
      "step": 20700
    },
    {
      "epoch": 2.5869592320159582,
      "grad_norm": 3.335686683654785,
      "learning_rate": 1.019388702279599e-06,
      "loss": 0.6169,
      "step": 20750
    },
    {
      "epoch": 2.593192868719611,
      "grad_norm": 3.2197108268737793,
      "learning_rate": 9.89374485629202e-07,
      "loss": 0.6029,
      "step": 20800
    },
    {
      "epoch": 2.599426505423264,
      "grad_norm": 3.1290664672851562,
      "learning_rate": 9.59785748371257e-07,
      "loss": 0.6122,
      "step": 20850
    },
    {
      "epoch": 2.6056601421269168,
      "grad_norm": 3.439748525619507,
      "learning_rate": 9.306238876782381e-07,
      "loss": 0.6118,
      "step": 20900
    },
    {
      "epoch": 2.6118937788305696,
      "grad_norm": 3.250145196914673,
      "learning_rate": 9.018902805656249e-07,
      "loss": 0.6052,
      "step": 20950
    },
    {
      "epoch": 2.6181274155342225,
      "grad_norm": 3.2482573986053467,
      "learning_rate": 8.735862838268638e-07,
      "loss": 0.604,
      "step": 21000
    },
    {
      "epoch": 2.6243610522378757,
      "grad_norm": 2.9533278942108154,
      "learning_rate": 8.457132339693231e-07,
      "loss": 0.5987,
      "step": 21050
    },
    {
      "epoch": 2.6305946889415286,
      "grad_norm": 3.3683395385742188,
      "learning_rate": 8.182724471511605e-07,
      "loss": 0.6065,
      "step": 21100
    },
    {
      "epoch": 2.6368283256451814,
      "grad_norm": 3.0807478427886963,
      "learning_rate": 7.912652191191905e-07,
      "loss": 0.61,
      "step": 21150
    },
    {
      "epoch": 2.6430619623488343,
      "grad_norm": 3.1802971363067627,
      "learning_rate": 7.64692825147696e-07,
      "loss": 0.6138,
      "step": 21200
    },
    {
      "epoch": 2.649295599052487,
      "grad_norm": 3.037116289138794,
      "learning_rate": 7.385565199782063e-07,
      "loss": 0.6098,
      "step": 21250
    },
    {
      "epoch": 2.65552923575614,
      "grad_norm": 3.420164108276367,
      "learning_rate": 7.128575377602509e-07,
      "loss": 0.6116,
      "step": 21300
    },
    {
      "epoch": 2.661762872459793,
      "grad_norm": 2.8981919288635254,
      "learning_rate": 6.87597091993083e-07,
      "loss": 0.613,
      "step": 21350
    },
    {
      "epoch": 2.667996509163446,
      "grad_norm": 3.189523458480835,
      "learning_rate": 6.627763754683824e-07,
      "loss": 0.6123,
      "step": 21400
    },
    {
      "epoch": 2.674230145867099,
      "grad_norm": 3.060560703277588,
      "learning_rate": 6.383965602139253e-07,
      "loss": 0.6085,
      "step": 21450
    },
    {
      "epoch": 2.680463782570752,
      "grad_norm": 3.2859272956848145,
      "learning_rate": 6.144587974382399e-07,
      "loss": 0.6029,
      "step": 21500
    },
    {
      "epoch": 2.6866974192744046,
      "grad_norm": 3.47271466255188,
      "learning_rate": 5.909642174762642e-07,
      "loss": 0.6034,
      "step": 21550
    },
    {
      "epoch": 2.6929310559780575,
      "grad_norm": 3.2543468475341797,
      "learning_rate": 5.679139297359448e-07,
      "loss": 0.6125,
      "step": 21600
    },
    {
      "epoch": 2.6991646926817108,
      "grad_norm": 3.462310552597046,
      "learning_rate": 5.453090226458758e-07,
      "loss": 0.6088,
      "step": 21650
    },
    {
      "epoch": 2.705398329385363,
      "grad_norm": 3.0585429668426514,
      "learning_rate": 5.231505636038881e-07,
      "loss": 0.6118,
      "step": 21700
    },
    {
      "epoch": 2.7116319660890165,
      "grad_norm": 3.213984489440918,
      "learning_rate": 5.014395989266496e-07,
      "loss": 0.6,
      "step": 21750
    },
    {
      "epoch": 2.7178656027926693,
      "grad_norm": 3.1287286281585693,
      "learning_rate": 4.801771538002687e-07,
      "loss": 0.6084,
      "step": 21800
    },
    {
      "epoch": 2.724099239496322,
      "grad_norm": 3.0341527462005615,
      "learning_rate": 4.59364232231867e-07,
      "loss": 0.6066,
      "step": 21850
    },
    {
      "epoch": 2.730332876199975,
      "grad_norm": 3.1067686080932617,
      "learning_rate": 4.3900181700219035e-07,
      "loss": 0.6155,
      "step": 21900
    },
    {
      "epoch": 2.736566512903628,
      "grad_norm": 3.143383741378784,
      "learning_rate": 4.190908696191853e-07,
      "loss": 0.6087,
      "step": 21950
    },
    {
      "epoch": 2.742800149607281,
      "grad_norm": 3.1124000549316406,
      "learning_rate": 3.9963233027260794e-07,
      "loss": 0.6039,
      "step": 22000
    },
    {
      "epoch": 2.742800149607281,
      "eval_loss": 0.621798038482666,
      "eval_runtime": 429.6773,
      "eval_samples_per_second": 42.667,
      "eval_steps_per_second": 5.334,
      "step": 22000
    },
    {
      "epoch": 2.7490337863109335,
      "grad_norm": 3.36039137840271,
      "learning_rate": 3.806271177896248e-07,
      "loss": 0.6065,
      "step": 22050
    },
    {
      "epoch": 2.755267423014587,
      "grad_norm": 3.3219449520111084,
      "learning_rate": 3.6207612959142213e-07,
      "loss": 0.6026,
      "step": 22100
    },
    {
      "epoch": 2.7615010597182397,
      "grad_norm": 3.0128390789031982,
      "learning_rate": 3.4398024165083864e-07,
      "loss": 0.6133,
      "step": 22150
    },
    {
      "epoch": 2.7677346964218925,
      "grad_norm": 3.38627290725708,
      "learning_rate": 3.2634030845099417e-07,
      "loss": 0.61,
      "step": 22200
    },
    {
      "epoch": 2.7739683331255454,
      "grad_norm": 2.9998061656951904,
      "learning_rate": 3.0915716294494193e-07,
      "loss": 0.6079,
      "step": 22250
    },
    {
      "epoch": 2.780201969829198,
      "grad_norm": 3.2025392055511475,
      "learning_rate": 2.9243161651634654e-07,
      "loss": 0.6082,
      "step": 22300
    },
    {
      "epoch": 2.7864356065328515,
      "grad_norm": 3.0767886638641357,
      "learning_rate": 2.7616445894115607e-07,
      "loss": 0.6023,
      "step": 22350
    },
    {
      "epoch": 2.7926692432365043,
      "grad_norm": 3.3690905570983887,
      "learning_rate": 2.6035645835032044e-07,
      "loss": 0.6051,
      "step": 22400
    },
    {
      "epoch": 2.798902879940157,
      "grad_norm": 3.1710546016693115,
      "learning_rate": 2.4500836119351503e-07,
      "loss": 0.6105,
      "step": 22450
    },
    {
      "epoch": 2.80513651664381,
      "grad_norm": 3.1593053340911865,
      "learning_rate": 2.301208922038911e-07,
      "loss": 0.6044,
      "step": 22500
    },
    {
      "epoch": 2.811370153347463,
      "grad_norm": 3.0642852783203125,
      "learning_rate": 2.1569475436386546e-07,
      "loss": 0.6063,
      "step": 22550
    },
    {
      "epoch": 2.8176037900511157,
      "grad_norm": 3.3536767959594727,
      "learning_rate": 2.0173062887190898e-07,
      "loss": 0.5933,
      "step": 22600
    },
    {
      "epoch": 2.8238374267547686,
      "grad_norm": 3.0567429065704346,
      "learning_rate": 1.8822917511039818e-07,
      "loss": 0.6152,
      "step": 22650
    },
    {
      "epoch": 2.830071063458422,
      "grad_norm": 3.240172863006592,
      "learning_rate": 1.7519103061446552e-07,
      "loss": 0.6209,
      "step": 22700
    },
    {
      "epoch": 2.8363047001620747,
      "grad_norm": 3.2891921997070312,
      "learning_rate": 1.626168110419013e-07,
      "loss": 0.5974,
      "step": 22750
    },
    {
      "epoch": 2.8425383368657275,
      "grad_norm": 3.288302421569824,
      "learning_rate": 1.505071101440836e-07,
      "loss": 0.6079,
      "step": 22800
    },
    {
      "epoch": 2.8487719735693804,
      "grad_norm": 3.1591663360595703,
      "learning_rate": 1.388624997379373e-07,
      "loss": 0.6111,
      "step": 22850
    },
    {
      "epoch": 2.8550056102730332,
      "grad_norm": 3.1807122230529785,
      "learning_rate": 1.2768352967893582e-07,
      "loss": 0.6063,
      "step": 22900
    },
    {
      "epoch": 2.861239246976686,
      "grad_norm": 3.043985366821289,
      "learning_rate": 1.169707278351373e-07,
      "loss": 0.6117,
      "step": 22950
    },
    {
      "epoch": 2.867472883680339,
      "grad_norm": 2.9105870723724365,
      "learning_rate": 1.0672460006225682e-07,
      "loss": 0.6125,
      "step": 23000
    },
    {
      "epoch": 2.873706520383992,
      "grad_norm": 3.0219972133636475,
      "learning_rate": 9.694563017978331e-08,
      "loss": 0.6107,
      "step": 23050
    },
    {
      "epoch": 2.879940157087645,
      "grad_norm": 3.188505172729492,
      "learning_rate": 8.763427994813112e-08,
      "loss": 0.6003,
      "step": 23100
    },
    {
      "epoch": 2.886173793791298,
      "grad_norm": 3.3090999126434326,
      "learning_rate": 7.879098904683303e-08,
      "loss": 0.6125,
      "step": 23150
    },
    {
      "epoch": 2.8924074304949507,
      "grad_norm": 3.0697832107543945,
      "learning_rate": 7.041617505378573e-08,
      "loss": 0.6094,
      "step": 23200
    },
    {
      "epoch": 2.8986410671986036,
      "grad_norm": 2.9027655124664307,
      "learning_rate": 6.251023342552787e-08,
      "loss": 0.6091,
      "step": 23250
    },
    {
      "epoch": 2.9048747039022564,
      "grad_norm": 3.0272111892700195,
      "learning_rate": 5.5073537478566034e-08,
      "loss": 0.6137,
      "step": 23300
    },
    {
      "epoch": 2.9111083406059093,
      "grad_norm": 2.9046285152435303,
      "learning_rate": 4.810643837174667e-08,
      "loss": 0.6043,
      "step": 23350
    },
    {
      "epoch": 2.9173419773095626,
      "grad_norm": 3.4514291286468506,
      "learning_rate": 4.160926508967822e-08,
      "loss": 0.6108,
      "step": 23400
    },
    {
      "epoch": 2.9235756140132154,
      "grad_norm": 3.135082244873047,
      "learning_rate": 3.558232442719245e-08,
      "loss": 0.6124,
      "step": 23450
    },
    {
      "epoch": 2.9298092507168683,
      "grad_norm": 3.415313243865967,
      "learning_rate": 3.002590097485936e-08,
      "loss": 0.6037,
      "step": 23500
    },
    {
      "epoch": 2.936042887420521,
      "grad_norm": 3.178391456604004,
      "learning_rate": 2.4940257105547928e-08,
      "loss": 0.6029,
      "step": 23550
    },
    {
      "epoch": 2.942276524124174,
      "grad_norm": 3.341081380844116,
      "learning_rate": 2.0325632962039376e-08,
      "loss": 0.6069,
      "step": 23600
    },
    {
      "epoch": 2.948510160827827,
      "grad_norm": 3.4343740940093994,
      "learning_rate": 1.6182246445685114e-08,
      "loss": 0.6091,
      "step": 23650
    },
    {
      "epoch": 2.9547437975314796,
      "grad_norm": 3.357135772705078,
      "learning_rate": 1.2510293206118296e-08,
      "loss": 0.6145,
      "step": 23700
    },
    {
      "epoch": 2.960977434235133,
      "grad_norm": 3.356213331222534,
      "learning_rate": 9.309946632015676e-09,
      "loss": 0.6075,
      "step": 23750
    },
    {
      "epoch": 2.9672110709387858,
      "grad_norm": 3.239623546600342,
      "learning_rate": 6.581357842909697e-09,
      "loss": 0.6149,
      "step": 23800
    },
    {
      "epoch": 2.9734447076424386,
      "grad_norm": 3.2543535232543945,
      "learning_rate": 4.324655682051982e-09,
      "loss": 0.6061,
      "step": 23850
    },
    {
      "epoch": 2.9796783443460915,
      "grad_norm": 3.3732693195343018,
      "learning_rate": 2.5399467103337518e-09,
      "loss": 0.6059,
      "step": 23900
    },
    {
      "epoch": 2.9859119810497443,
      "grad_norm": 3.259169578552246,
      "learning_rate": 1.2273152012465262e-09,
      "loss": 0.6133,
      "step": 23950
    },
    {
      "epoch": 2.9921456177533976,
      "grad_norm": 2.9566566944122314,
      "learning_rate": 3.8682313690974194e-10,
      "loss": 0.613,
      "step": 24000
    },
    {
      "epoch": 2.9921456177533976,
      "eval_loss": 0.6216655969619751,
      "eval_runtime": 428.2552,
      "eval_samples_per_second": 42.809,
      "eval_steps_per_second": 5.352,
      "step": 24000
    },
    {
      "epoch": 2.99837925445705,
      "grad_norm": 3.3883161544799805,
      "learning_rate": 1.8510205138655424e-11,
      "loss": 0.6053,
      "step": 24050
    },
    {
      "epoch": 3.0,
      "step": 24063,
      "total_flos": 2.368215955426173e+18,
      "train_loss": 0.6610314225296434,
      "train_runtime": 35586.5982,
      "train_samples_per_second": 10.819,
      "train_steps_per_second": 0.676
    }
  ],
  "logging_steps": 50,
  "max_steps": 24063,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 2000,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 2,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.368215955426173e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}