Hugging Face's logo

Datasets:
deeplumiere
/
EEE_datastore

Modalities:
Text
Formats:
parquet
Size:
1K - 10K
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
21
deeplumiere commited on
Commit
4dc706a
·
verified ·
1 Parent(s): 3753796

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. data/mmlu_pro/01-ai/yi-1.5-34b-chat/c9134421-80f0-42d8-982b-f0f40754dd85.json +590 -0
  2. data/mmlu_pro/01-ai/yi-1.5-6b-chat/4f29bf2d-a5f4-416f-8e90-f6c1a951bebc.json +590 -0
  3. data/mmlu_pro/01-ai/yi-1.5-9b-chat/443312f3-5866-48c6-9fcb-f591df683fa5.json +590 -0
  4. data/mmlu_pro/01-ai/yi-34b/c8bce2aa-e6b1-4dcf-a874-2b65a547a792.json +590 -0
  5. data/mmlu_pro/01-ai/yi-6b-chat/a4db3d05-bc5a-4b9c-95f8-f7c339ed8282.json +590 -0
  6. data/mmlu_pro/01-ai/yi-6b/eaa5ed91-b28a-41a4-aeaa-c3f188496e18.json +590 -0
  7. data/mmlu_pro/01-ai/yi-large/6d386bbf-1224-41f9-a6b6-86ea931ced12.json +590 -0
  8. data/mmlu_pro/01-ai/yi-lightning/4f15fe1a-baf2-46d7-a83b-8f75d03591c7.json +589 -0
  9. data/mmlu_pro/Gemini-3-Pro(11/gemini-3-pro-11-25/e5c93fdf-9028-4e42-956d-a2c34770ba4d.json +71 -0
  10. data/mmlu_pro/NewenAI/newenai-phi4-sft/dfc932df-3c64-42db-a134-74cea5fb624f.json +72 -0
  11. data/mmlu_pro/abacus-ai/llama3-smaug-8b/bbd6e3fc-269b-43de-8c07-f53d4cdd345a.json +590 -0
  12. data/mmlu_pro/google/gemini-1.5-flash-002/0d1d88b0-fa49-4463-a184-e4302ffa4571.json +589 -0
  13. data/mmlu_pro/google/gemini-1.5-flash/01bb7276-91e8-44eb-ba3a-c7df072f7a38.json +589 -0
  14. data/mmlu_pro/google/gemini-1.5-pro-002/b3ffcd12-3a74-4f3f-9acf-7ec11615a88d.json +589 -0
  15. data/mmlu_pro/google/gemini-1.5-pro/fd7ad2ae-dd54-4275-8db7-7da0110eb2be.json +589 -0
  16. data/mmlu_pro/google/gemini-2.0-flash-exp/b70dfa85-e4ee-4eaa-9d5b-7f58a6ac2413.json +589 -0
  17. data/mmlu_pro/google/gemini-2.0-flash-lite/841ad2b0-86e3-44e4-88c2-d010d6c9c52e.json +71 -0
  18. data/mmlu_pro/google/gemini-2.0-flash/1961f28e-d519-422d-8221-fd482b2f60f5.json +71 -0
  19. data/mmlu_pro/google/gemini-2.0-pro/34eecaad-e501-4b2b-9e9c-284804d1a581.json +71 -0
  20. data/mmlu_pro/google/gemini-2.5-pro-exp-03-25/976b7201-8e01-49f0-b9ab-884d61e16288.json +589 -0
  21. data/mmlu_pro/google/gemini-2.5-pro/4ac5ed20-3428-462d-b108-9d4bfb2fed2b.json +71 -0
  22. data/mmlu_pro/google/gemini-3.1-flash-lite-preview/9e560681-ebd4-4d96-92fa-ef8e8e6c64fe.json +71 -0
  23. data/mmlu_pro/google/gemini-3.1-pro/c03b9f04-fa5b-41e8-b58f-0bfe85be3c0f.json +589 -0
  24. data/mmlu_pro/google/gemma-2-27b-it/e8295867-8053-488f-9130-9eb99b97670e.json +590 -0
  25. data/mmlu_pro/google/gemma-2-2b-it/7461516d-9358-42db-9888-056fd84c525f.json +72 -0
  26. data/mmlu_pro/google/gemma-2-9b-it/c9fcb231-ddb3-4c0a-8d57-3a3782f0d752.json +590 -0
  27. data/mmlu_pro/google/gemma-2-9b/9c2c6d87-805f-47c0-9ddf-e0b81d0f69f8.json +590 -0
  28. data/mmlu_pro/google/gemma-2b/eb556f91-0cbd-4530-a7d1-d80e86d42588.json +590 -0
  29. data/mmlu_pro/google/gemma-3-12b-it/83198456-6a19-4574-abf8-0a0cc97287af.json +72 -0
  30. data/mmlu_pro/google/gemma-3-1b-it/e3c1d8e7-a98f-4fd4-ac24-3d0c7c3669a9.json +72 -0
  31. data/mmlu_pro/google/gemma-3-27b-it/6a53d694-91e0-40bd-a265-1f640756ae66.json +72 -0
  32. data/mmlu_pro/google/gemma-3-4b-it/e7c8cf64-4f7a-4a14-8517-00760409be7c.json +72 -0
  33. data/mmlu_pro/google/gemma-7b/1e21b40e-8ae9-4712-87ad-344e5423846b.json +590 -0
  34. data/mmlu_pro/ibm/granite-3.0-2b-base/47ec1fc0-0ac3-4d86-87f6-274667ef2a16.json +590 -0
  35. data/mmlu_pro/ibm/granite-3.0-8b-base/5e811a7a-5a2c-4484-9368-5d62dab1f13f.json +590 -0
  36. data/mmlu_pro/ibm/granite-3.1-1b-a400m-base/1f8d80d4-01a3-4818-9a8b-d00054dcab43.json +590 -0
  37. data/mmlu_pro/ibm/granite-3.1-1b-a400m-instruct/f3eeed4c-1d73-46f0-acac-eafd5fc9754f.json +590 -0
  38. data/mmlu_pro/ibm/granite-3.1-2b-base/ba525354-6d57-4b26-b777-e58271890ae6.json +590 -0
  39. data/mmlu_pro/ibm/granite-3.1-2b-instruct/7959f684-3532-439e-a9e0-5389dcaf795c.json +590 -0
  40. data/mmlu_pro/ibm/granite-3.1-3b-a800m-base/95d5b3bc-f7e5-4171-9be8-9656a902d92f.json +590 -0
  41. data/mmlu_pro/ibm/granite-3.1-3b-a800m-instruct/b186c499-e9e3-4fef-8bbf-3c43f7c71e28.json +590 -0
  42. data/mmlu_pro/ibm/granite-3.1-8b-base/6b91c3fc-4ec1-480c-9e78-99440cbfe97d.json +590 -0
  43. data/mmlu_pro/ibm/granite-3.1-8b-instruct/07107f2c-5b3e-4b5d-87ec-0acc108a68f9.json +590 -0
  44. data/mmlu_pro/microsoft/phi-3.5-mini-instruct/61ae9c04-3c9a-47e3-bcab-eea1268d14fe.json +590 -0
  45. data/mmlu_pro/microsoft/phi-4-mini/86cb1a4f-477f-4c08-8fb4-d0a6153984c3.json +72 -0
  46. data/mmlu_pro/microsoft/phi-4-reasoning-plus/7f6559ff-c8dd-4fe7-9cd4-c1cd34e65c8b.json +72 -0
  47. data/mmlu_pro/microsoft/phi-4-reasoning/a61f15ba-23b9-48d9-bf7c-c1911d765421.json +72 -0
  48. data/mmlu_pro/microsoft/phi-4/c6c805a9-7065-4811-a2ee-aa9b3e065e09.json +72 -0
  49. data/mmlu_pro/microsoft/phi3-medium-128k/cfba928a-8592-47f2-bc5d-bb97cd624352.json +590 -0
  50. data/mmlu_pro/microsoft/phi3-medium-4k/996cc51e-f338-4192-ac74-c3149c097648.json +590 -0
data/mmlu_pro/01-ai/yi-1.5-34b-chat/c9134421-80f0-42d8-982b-f0f40754dd85.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/01-ai_yi-1.5-34b-chat/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Yi-1.5-34B-Chat",
25
+ "id": "01-ai/yi-1.5-34b-chat",
26
+ "developer": "01-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Yi-1.5-34B-Chat",
29
+ "size_billions_parameters": "34.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.5229
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.7141
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.5843
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.4753
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.539
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.6457
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3437
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.5819
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.5276
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.3479
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.5618
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.4629
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.4935
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.6429
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.5162
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/01-ai/yi-1.5-6b-chat/4f29bf2d-a5f4-416f-8e90-f6c1a951bebc.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/01-ai_yi-1.5-6b-chat/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Yi-1.5-6B-Chat",
25
+ "id": "01-ai/yi-1.5-6b-chat",
26
+ "developer": "01-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Yi-1.5-6B-Chat",
29
+ "size_billions_parameters": "6.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.3823
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.5746
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.4766
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.3074
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.4366
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.5273
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.2683
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.3362
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.3176
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2198
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.4145
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.3327
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.3564
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.5013
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.382
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/01-ai/yi-1.5-9b-chat/443312f3-5866-48c6-9fcb-f591df683fa5.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/01-ai_yi-1.5-9b-chat/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Yi-1.5-9B-Chat",
25
+ "id": "01-ai/yi-1.5-9b-chat",
26
+ "developer": "01-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Yi-1.5-9B-Chat",
29
+ "size_billions_parameters": "9.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.4595
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.6667
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.5425
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.3949
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.5
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.6019
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3323
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.4352
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.4094
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2661
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.5248
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.4008
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.4142
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.594
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.4491
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/01-ai/yi-34b/c8bce2aa-e6b1-4dcf-a874-2b65a547a792.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/01-ai_yi-34b/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Yi-34B",
25
+ "id": "01-ai/yi-34b",
26
+ "developer": "01-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Yi-34B",
29
+ "size_billions_parameters": "34.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.4303
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.6527
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.4005
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.265
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.4366
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.5569
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3261
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.5379
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.5197
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.327
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.3175
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.477
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.3503
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.6253
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.5509
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/01-ai/yi-6b-chat/a4db3d05-bc5a-4b9c-95f8-f7c339ed8282.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/01-ai_yi-6b-chat/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Yi-6B-Chat",
25
+ "id": "01-ai/yi-6b-chat",
26
+ "developer": "01-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Yi-6B-Chat",
29
+ "size_billions_parameters": "6.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.2884
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.477
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.2826
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1661
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.2659
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.3969
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.1899
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.3521
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.315
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2162
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.2124
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.3367
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.2094
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.4912
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.3506
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/01-ai/yi-6b/eaa5ed91-b28a-41a4-aeaa-c3f188496e18.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/01-ai_yi-6b/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Yi-6B",
25
+ "id": "01-ai/yi-6b",
26
+ "developer": "01-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Yi-6B",
29
+ "size_billions_parameters": "6.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.2651
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.4226
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.2864
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1484
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.2732
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.3578
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.1796
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.3166
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.294
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.1953
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.1902
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.3186
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.1832
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.4286
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.3496
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/01-ai/yi-large/6d386bbf-1224-41f9-a6b6-86ea931ced12.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/01-ai_yi-large/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Yi-large",
25
+ "id": "01-ai/yi-large",
26
+ "developer": "01-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Yi-large",
29
+ "size_billions_parameters": "150.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.5809
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.6987
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.6413
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.6166
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.6341
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.6813
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.4541
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.6443
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.4961
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.3624
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.6481
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.5531
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.5704
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.5063
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.6472
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/01-ai/yi-lightning/4f15fe1a-baf2-46d7-a83b-8f75d03591c7.json ADDED
@@ -0,0 +1,589 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/01-ai_yi-lightning/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Yi-Lightning",
25
+ "id": "01-ai/yi-lightning",
26
+ "developer": "01-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Yi-Lightning",
29
+ "leaderboard_data_source": "TIGER-Lab"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.6238
68
+ }
69
+ },
70
+ {
71
+ "evaluation_result_id": "mmlu_pro/biology",
72
+ "evaluation_name": "MMLU-Pro (Biology)",
73
+ "source_data": {
74
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
75
+ "source_type": "hf_dataset",
76
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
77
+ "hf_split": "train",
78
+ "additional_details": {
79
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
80
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
81
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
82
+ "paper_url": "https://arxiv.org/abs/2406.01574",
83
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
84
+ "dataset_total_questions": "12000",
85
+ "prompt_style": "5-shot CoT"
86
+ }
87
+ },
88
+ "metric_config": {
89
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
90
+ "metric_id": "mmlu_pro/biology",
91
+ "metric_name": "MMLU-Pro (Biology)",
92
+ "metric_kind": "accuracy",
93
+ "metric_unit": "proportion",
94
+ "lower_is_better": false,
95
+ "score_type": "continuous",
96
+ "min_score": 0.0,
97
+ "max_score": 1.0,
98
+ "additional_details": {
99
+ "aggregation": "accuracy_over_subset",
100
+ "prompt_style": "5-shot CoT"
101
+ }
102
+ },
103
+ "score_details": {
104
+ "score": 0.7964
105
+ }
106
+ },
107
+ {
108
+ "evaluation_result_id": "mmlu_pro/business",
109
+ "evaluation_name": "MMLU-Pro (Business)",
110
+ "source_data": {
111
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
112
+ "source_type": "hf_dataset",
113
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
114
+ "hf_split": "train",
115
+ "additional_details": {
116
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
117
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
118
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
119
+ "paper_url": "https://arxiv.org/abs/2406.01574",
120
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
121
+ "dataset_total_questions": "12000",
122
+ "prompt_style": "5-shot CoT"
123
+ }
124
+ },
125
+ "metric_config": {
126
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
127
+ "metric_id": "mmlu_pro/business",
128
+ "metric_name": "MMLU-Pro (Business)",
129
+ "metric_kind": "accuracy",
130
+ "metric_unit": "proportion",
131
+ "lower_is_better": false,
132
+ "score_type": "continuous",
133
+ "min_score": 0.0,
134
+ "max_score": 1.0,
135
+ "additional_details": {
136
+ "aggregation": "accuracy_over_subset",
137
+ "prompt_style": "5-shot CoT"
138
+ }
139
+ },
140
+ "score_details": {
141
+ "score": 0.6907
142
+ }
143
+ },
144
+ {
145
+ "evaluation_result_id": "mmlu_pro/chemistry",
146
+ "evaluation_name": "MMLU-Pro (Chemistry)",
147
+ "source_data": {
148
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
149
+ "source_type": "hf_dataset",
150
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
151
+ "hf_split": "train",
152
+ "additional_details": {
153
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
154
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
155
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
156
+ "paper_url": "https://arxiv.org/abs/2406.01574",
157
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
158
+ "dataset_total_questions": "12000",
159
+ "prompt_style": "5-shot CoT"
160
+ }
161
+ },
162
+ "metric_config": {
163
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
164
+ "metric_id": "mmlu_pro/chemistry",
165
+ "metric_name": "MMLU-Pro (Chemistry)",
166
+ "metric_kind": "accuracy",
167
+ "metric_unit": "proportion",
168
+ "lower_is_better": false,
169
+ "score_type": "continuous",
170
+ "min_score": 0.0,
171
+ "max_score": 1.0,
172
+ "additional_details": {
173
+ "aggregation": "accuracy_over_subset",
174
+ "prompt_style": "5-shot CoT"
175
+ }
176
+ },
177
+ "score_details": {
178
+ "score": 0.6193
179
+ }
180
+ },
181
+ {
182
+ "evaluation_result_id": "mmlu_pro/computer_science",
183
+ "evaluation_name": "MMLU-Pro (Computer Science)",
184
+ "source_data": {
185
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
186
+ "source_type": "hf_dataset",
187
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
188
+ "hf_split": "train",
189
+ "additional_details": {
190
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
191
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
192
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
193
+ "paper_url": "https://arxiv.org/abs/2406.01574",
194
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
195
+ "dataset_total_questions": "12000",
196
+ "prompt_style": "5-shot CoT"
197
+ }
198
+ },
199
+ "metric_config": {
200
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
201
+ "metric_id": "mmlu_pro/computer_science",
202
+ "metric_name": "MMLU-Pro (Computer Science)",
203
+ "metric_kind": "accuracy",
204
+ "metric_unit": "proportion",
205
+ "lower_is_better": false,
206
+ "score_type": "continuous",
207
+ "min_score": 0.0,
208
+ "max_score": 1.0,
209
+ "additional_details": {
210
+ "aggregation": "accuracy_over_subset",
211
+ "prompt_style": "5-shot CoT"
212
+ }
213
+ },
214
+ "score_details": {
215
+ "score": 0.6439
216
+ }
217
+ },
218
+ {
219
+ "evaluation_result_id": "mmlu_pro/economics",
220
+ "evaluation_name": "MMLU-Pro (Economics)",
221
+ "source_data": {
222
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
223
+ "source_type": "hf_dataset",
224
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
225
+ "hf_split": "train",
226
+ "additional_details": {
227
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
228
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
229
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
230
+ "paper_url": "https://arxiv.org/abs/2406.01574",
231
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
232
+ "dataset_total_questions": "12000",
233
+ "prompt_style": "5-shot CoT"
234
+ }
235
+ },
236
+ "metric_config": {
237
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
238
+ "metric_id": "mmlu_pro/economics",
239
+ "metric_name": "MMLU-Pro (Economics)",
240
+ "metric_kind": "accuracy",
241
+ "metric_unit": "proportion",
242
+ "lower_is_better": false,
243
+ "score_type": "continuous",
244
+ "min_score": 0.0,
245
+ "max_score": 1.0,
246
+ "additional_details": {
247
+ "aggregation": "accuracy_over_subset",
248
+ "prompt_style": "5-shot CoT"
249
+ }
250
+ },
251
+ "score_details": {
252
+ "score": 0.731
253
+ }
254
+ },
255
+ {
256
+ "evaluation_result_id": "mmlu_pro/engineering",
257
+ "evaluation_name": "MMLU-Pro (Engineering)",
258
+ "source_data": {
259
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
260
+ "source_type": "hf_dataset",
261
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
262
+ "hf_split": "train",
263
+ "additional_details": {
264
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
265
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
266
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
267
+ "paper_url": "https://arxiv.org/abs/2406.01574",
268
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
269
+ "dataset_total_questions": "12000",
270
+ "prompt_style": "5-shot CoT"
271
+ }
272
+ },
273
+ "metric_config": {
274
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
275
+ "metric_id": "mmlu_pro/engineering",
276
+ "metric_name": "MMLU-Pro (Engineering)",
277
+ "metric_kind": "accuracy",
278
+ "metric_unit": "proportion",
279
+ "lower_is_better": false,
280
+ "score_type": "continuous",
281
+ "min_score": 0.0,
282
+ "max_score": 1.0,
283
+ "additional_details": {
284
+ "aggregation": "accuracy_over_subset",
285
+ "prompt_style": "5-shot CoT"
286
+ }
287
+ },
288
+ "score_details": {
289
+ "score": 0.4221
290
+ }
291
+ },
292
+ {
293
+ "evaluation_result_id": "mmlu_pro/health",
294
+ "evaluation_name": "MMLU-Pro (Health)",
295
+ "source_data": {
296
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
297
+ "source_type": "hf_dataset",
298
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
299
+ "hf_split": "train",
300
+ "additional_details": {
301
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
302
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
303
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
304
+ "paper_url": "https://arxiv.org/abs/2406.01574",
305
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
306
+ "dataset_total_questions": "12000",
307
+ "prompt_style": "5-shot CoT"
308
+ }
309
+ },
310
+ "metric_config": {
311
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
312
+ "metric_id": "mmlu_pro/health",
313
+ "metric_name": "MMLU-Pro (Health)",
314
+ "metric_kind": "accuracy",
315
+ "metric_unit": "proportion",
316
+ "lower_is_better": false,
317
+ "score_type": "continuous",
318
+ "min_score": 0.0,
319
+ "max_score": 1.0,
320
+ "additional_details": {
321
+ "aggregation": "accuracy_over_subset",
322
+ "prompt_style": "5-shot CoT"
323
+ }
324
+ },
325
+ "score_details": {
326
+ "score": 0.6553
327
+ }
328
+ },
329
+ {
330
+ "evaluation_result_id": "mmlu_pro/history",
331
+ "evaluation_name": "MMLU-Pro (History)",
332
+ "source_data": {
333
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
334
+ "source_type": "hf_dataset",
335
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
336
+ "hf_split": "train",
337
+ "additional_details": {
338
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
339
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
340
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
341
+ "paper_url": "https://arxiv.org/abs/2406.01574",
342
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
343
+ "dataset_total_questions": "12000",
344
+ "prompt_style": "5-shot CoT"
345
+ }
346
+ },
347
+ "metric_config": {
348
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
349
+ "metric_id": "mmlu_pro/history",
350
+ "metric_name": "MMLU-Pro (History)",
351
+ "metric_kind": "accuracy",
352
+ "metric_unit": "proportion",
353
+ "lower_is_better": false,
354
+ "score_type": "continuous",
355
+ "min_score": 0.0,
356
+ "max_score": 1.0,
357
+ "additional_details": {
358
+ "aggregation": "accuracy_over_subset",
359
+ "prompt_style": "5-shot CoT"
360
+ }
361
+ },
362
+ "score_details": {
363
+ "score": 0.5748
364
+ }
365
+ },
366
+ {
367
+ "evaluation_result_id": "mmlu_pro/law",
368
+ "evaluation_name": "MMLU-Pro (Law)",
369
+ "source_data": {
370
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
371
+ "source_type": "hf_dataset",
372
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
373
+ "hf_split": "train",
374
+ "additional_details": {
375
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
376
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
377
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
378
+ "paper_url": "https://arxiv.org/abs/2406.01574",
379
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
380
+ "dataset_total_questions": "12000",
381
+ "prompt_style": "5-shot CoT"
382
+ }
383
+ },
384
+ "metric_config": {
385
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
386
+ "metric_id": "mmlu_pro/law",
387
+ "metric_name": "MMLU-Pro (Law)",
388
+ "metric_kind": "accuracy",
389
+ "metric_unit": "proportion",
390
+ "lower_is_better": false,
391
+ "score_type": "continuous",
392
+ "min_score": 0.0,
393
+ "max_score": 1.0,
394
+ "additional_details": {
395
+ "aggregation": "accuracy_over_subset",
396
+ "prompt_style": "5-shot CoT"
397
+ }
398
+ },
399
+ "score_details": {
400
+ "score": 0.3751
401
+ }
402
+ },
403
+ {
404
+ "evaluation_result_id": "mmlu_pro/math",
405
+ "evaluation_name": "MMLU-Pro (Math)",
406
+ "source_data": {
407
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
408
+ "source_type": "hf_dataset",
409
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
410
+ "hf_split": "train",
411
+ "additional_details": {
412
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
413
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
414
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
415
+ "paper_url": "https://arxiv.org/abs/2406.01574",
416
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
417
+ "dataset_total_questions": "12000",
418
+ "prompt_style": "5-shot CoT"
419
+ }
420
+ },
421
+ "metric_config": {
422
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
423
+ "metric_id": "mmlu_pro/math",
424
+ "metric_name": "MMLU-Pro (Math)",
425
+ "metric_kind": "accuracy",
426
+ "metric_unit": "proportion",
427
+ "lower_is_better": false,
428
+ "score_type": "continuous",
429
+ "min_score": 0.0,
430
+ "max_score": 1.0,
431
+ "additional_details": {
432
+ "aggregation": "accuracy_over_subset",
433
+ "prompt_style": "5-shot CoT"
434
+ }
435
+ },
436
+ "score_details": {
437
+ "score": 0.6913
438
+ }
439
+ },
440
+ {
441
+ "evaluation_result_id": "mmlu_pro/philosophy",
442
+ "evaluation_name": "MMLU-Pro (Philosophy)",
443
+ "source_data": {
444
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
445
+ "source_type": "hf_dataset",
446
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
447
+ "hf_split": "train",
448
+ "additional_details": {
449
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
450
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
451
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
452
+ "paper_url": "https://arxiv.org/abs/2406.01574",
453
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
454
+ "dataset_total_questions": "12000",
455
+ "prompt_style": "5-shot CoT"
456
+ }
457
+ },
458
+ "metric_config": {
459
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
460
+ "metric_id": "mmlu_pro/philosophy",
461
+ "metric_name": "MMLU-Pro (Philosophy)",
462
+ "metric_kind": "accuracy",
463
+ "metric_unit": "proportion",
464
+ "lower_is_better": false,
465
+ "score_type": "continuous",
466
+ "min_score": 0.0,
467
+ "max_score": 1.0,
468
+ "additional_details": {
469
+ "aggregation": "accuracy_over_subset",
470
+ "prompt_style": "5-shot CoT"
471
+ }
472
+ },
473
+ "score_details": {
474
+ "score": 0.5711
475
+ }
476
+ },
477
+ {
478
+ "evaluation_result_id": "mmlu_pro/physics",
479
+ "evaluation_name": "MMLU-Pro (Physics)",
480
+ "source_data": {
481
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
482
+ "source_type": "hf_dataset",
483
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
484
+ "hf_split": "train",
485
+ "additional_details": {
486
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
487
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
488
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
489
+ "paper_url": "https://arxiv.org/abs/2406.01574",
490
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
491
+ "dataset_total_questions": "12000",
492
+ "prompt_style": "5-shot CoT"
493
+ }
494
+ },
495
+ "metric_config": {
496
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
497
+ "metric_id": "mmlu_pro/physics",
498
+ "metric_name": "MMLU-Pro (Physics)",
499
+ "metric_kind": "accuracy",
500
+ "metric_unit": "proportion",
501
+ "lower_is_better": false,
502
+ "score_type": "continuous",
503
+ "min_score": 0.0,
504
+ "max_score": 1.0,
505
+ "additional_details": {
506
+ "aggregation": "accuracy_over_subset",
507
+ "prompt_style": "5-shot CoT"
508
+ }
509
+ },
510
+ "score_details": {
511
+ "score": 0.6251
512
+ }
513
+ },
514
+ {
515
+ "evaluation_result_id": "mmlu_pro/psychology",
516
+ "evaluation_name": "MMLU-Pro (Psychology)",
517
+ "source_data": {
518
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
519
+ "source_type": "hf_dataset",
520
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
521
+ "hf_split": "train",
522
+ "additional_details": {
523
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
524
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
525
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
526
+ "paper_url": "https://arxiv.org/abs/2406.01574",
527
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
528
+ "dataset_total_questions": "12000",
529
+ "prompt_style": "5-shot CoT"
530
+ }
531
+ },
532
+ "metric_config": {
533
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
534
+ "metric_id": "mmlu_pro/psychology",
535
+ "metric_name": "MMLU-Pro (Psychology)",
536
+ "metric_kind": "accuracy",
537
+ "metric_unit": "proportion",
538
+ "lower_is_better": false,
539
+ "score_type": "continuous",
540
+ "min_score": 0.0,
541
+ "max_score": 1.0,
542
+ "additional_details": {
543
+ "aggregation": "accuracy_over_subset",
544
+ "prompt_style": "5-shot CoT"
545
+ }
546
+ },
547
+ "score_details": {
548
+ "score": 0.7293
549
+ }
550
+ },
551
+ {
552
+ "evaluation_result_id": "mmlu_pro/other",
553
+ "evaluation_name": "MMLU-Pro (Other)",
554
+ "source_data": {
555
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
556
+ "source_type": "hf_dataset",
557
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
558
+ "hf_split": "train",
559
+ "additional_details": {
560
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
561
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
562
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
563
+ "paper_url": "https://arxiv.org/abs/2406.01574",
564
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
565
+ "dataset_total_questions": "12000",
566
+ "prompt_style": "5-shot CoT"
567
+ }
568
+ },
569
+ "metric_config": {
570
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
571
+ "metric_id": "mmlu_pro/other",
572
+ "metric_name": "MMLU-Pro (Other)",
573
+ "metric_kind": "accuracy",
574
+ "metric_unit": "proportion",
575
+ "lower_is_better": false,
576
+ "score_type": "continuous",
577
+ "min_score": 0.0,
578
+ "max_score": 1.0,
579
+ "additional_details": {
580
+ "aggregation": "accuracy_over_subset",
581
+ "prompt_style": "5-shot CoT"
582
+ }
583
+ },
584
+ "score_details": {
585
+ "score": 0.6677
586
+ }
587
+ }
588
+ ]
589
+ }
data/mmlu_pro/Gemini-3-Pro(11/gemini-3-pro-11-25/e5c93fdf-9028-4e42-956d-a2c34770ba4d.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/Gemini-3-Pro(11_gemini-3-pro-11-25/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-3-Pro(11/25)",
25
+ "id": "Gemini-3-Pro(11/gemini-3-pro-11-25",
26
+ "developer": "Gemini-3-Pro(11",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-3-Pro(11/25)",
29
+ "leaderboard_data_source": "Self-Reported"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.901
68
+ }
69
+ }
70
+ ]
71
+ }
data/mmlu_pro/NewenAI/newenai-phi4-sft/dfc932df-3c64-42db-a134-74cea5fb624f.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/NewenAI_newenai-phi4-sft/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "NewenAI/Phi4-sft",
25
+ "id": "NewenAI/newenai-phi4-sft",
26
+ "developer": "NewenAI",
27
+ "additional_details": {
28
+ "raw_model_name": "NewenAI/Phi4-sft",
29
+ "size_billions_parameters": "14.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.577
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/abacus-ai/llama3-smaug-8b/bbd6e3fc-269b-43de-8c07-f53d4cdd345a.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/abacus-ai_llama3-smaug-8b/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Llama3-Smaug-8B",
25
+ "id": "abacus-ai/llama3-smaug-8b",
26
+ "developer": "abacus-ai",
27
+ "additional_details": {
28
+ "raw_model_name": "Llama3-Smaug-8B",
29
+ "size_billions_parameters": "8.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.3693
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.622
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.3738
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.2305
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.3658
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.4917
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.1981
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.4327
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.4199
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2652
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.3316
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.4502
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.3727
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.2856
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.5739
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/google/gemini-1.5-flash-002/0d1d88b0-fa49-4463-a184-e4302ffa4571.json ADDED
@@ -0,0 +1,589 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-1.5-flash-002/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-1.5-Flash-002",
25
+ "id": "google/gemini-1.5-flash-002",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-1.5-Flash-002",
29
+ "leaderboard_data_source": "TIGER-Lab"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.6409
68
+ }
69
+ },
70
+ {
71
+ "evaluation_result_id": "mmlu_pro/biology",
72
+ "evaluation_name": "MMLU-Pro (Biology)",
73
+ "source_data": {
74
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
75
+ "source_type": "hf_dataset",
76
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
77
+ "hf_split": "train",
78
+ "additional_details": {
79
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
80
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
81
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
82
+ "paper_url": "https://arxiv.org/abs/2406.01574",
83
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
84
+ "dataset_total_questions": "12000",
85
+ "prompt_style": "5-shot CoT"
86
+ }
87
+ },
88
+ "metric_config": {
89
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
90
+ "metric_id": "mmlu_pro/biology",
91
+ "metric_name": "MMLU-Pro (Biology)",
92
+ "metric_kind": "accuracy",
93
+ "metric_unit": "proportion",
94
+ "lower_is_better": false,
95
+ "score_type": "continuous",
96
+ "min_score": 0.0,
97
+ "max_score": 1.0,
98
+ "additional_details": {
99
+ "aggregation": "accuracy_over_subset",
100
+ "prompt_style": "5-shot CoT"
101
+ }
102
+ },
103
+ "score_details": {
104
+ "score": 0.8368
105
+ }
106
+ },
107
+ {
108
+ "evaluation_result_id": "mmlu_pro/business",
109
+ "evaluation_name": "MMLU-Pro (Business)",
110
+ "source_data": {
111
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
112
+ "source_type": "hf_dataset",
113
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
114
+ "hf_split": "train",
115
+ "additional_details": {
116
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
117
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
118
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
119
+ "paper_url": "https://arxiv.org/abs/2406.01574",
120
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
121
+ "dataset_total_questions": "12000",
122
+ "prompt_style": "5-shot CoT"
123
+ }
124
+ },
125
+ "metric_config": {
126
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
127
+ "metric_id": "mmlu_pro/business",
128
+ "metric_name": "MMLU-Pro (Business)",
129
+ "metric_kind": "accuracy",
130
+ "metric_unit": "proportion",
131
+ "lower_is_better": false,
132
+ "score_type": "continuous",
133
+ "min_score": 0.0,
134
+ "max_score": 1.0,
135
+ "additional_details": {
136
+ "aggregation": "accuracy_over_subset",
137
+ "prompt_style": "5-shot CoT"
138
+ }
139
+ },
140
+ "score_details": {
141
+ "score": 0.7145
142
+ }
143
+ },
144
+ {
145
+ "evaluation_result_id": "mmlu_pro/chemistry",
146
+ "evaluation_name": "MMLU-Pro (Chemistry)",
147
+ "source_data": {
148
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
149
+ "source_type": "hf_dataset",
150
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
151
+ "hf_split": "train",
152
+ "additional_details": {
153
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
154
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
155
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
156
+ "paper_url": "https://arxiv.org/abs/2406.01574",
157
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
158
+ "dataset_total_questions": "12000",
159
+ "prompt_style": "5-shot CoT"
160
+ }
161
+ },
162
+ "metric_config": {
163
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
164
+ "metric_id": "mmlu_pro/chemistry",
165
+ "metric_name": "MMLU-Pro (Chemistry)",
166
+ "metric_kind": "accuracy",
167
+ "metric_unit": "proportion",
168
+ "lower_is_better": false,
169
+ "score_type": "continuous",
170
+ "min_score": 0.0,
171
+ "max_score": 1.0,
172
+ "additional_details": {
173
+ "aggregation": "accuracy_over_subset",
174
+ "prompt_style": "5-shot CoT"
175
+ }
176
+ },
177
+ "score_details": {
178
+ "score": 0.6708
179
+ }
180
+ },
181
+ {
182
+ "evaluation_result_id": "mmlu_pro/computer_science",
183
+ "evaluation_name": "MMLU-Pro (Computer Science)",
184
+ "source_data": {
185
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
186
+ "source_type": "hf_dataset",
187
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
188
+ "hf_split": "train",
189
+ "additional_details": {
190
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
191
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
192
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
193
+ "paper_url": "https://arxiv.org/abs/2406.01574",
194
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
195
+ "dataset_total_questions": "12000",
196
+ "prompt_style": "5-shot CoT"
197
+ }
198
+ },
199
+ "metric_config": {
200
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
201
+ "metric_id": "mmlu_pro/computer_science",
202
+ "metric_name": "MMLU-Pro (Computer Science)",
203
+ "metric_kind": "accuracy",
204
+ "metric_unit": "proportion",
205
+ "lower_is_better": false,
206
+ "score_type": "continuous",
207
+ "min_score": 0.0,
208
+ "max_score": 1.0,
209
+ "additional_details": {
210
+ "aggregation": "accuracy_over_subset",
211
+ "prompt_style": "5-shot CoT"
212
+ }
213
+ },
214
+ "score_details": {
215
+ "score": 0.6341
216
+ }
217
+ },
218
+ {
219
+ "evaluation_result_id": "mmlu_pro/economics",
220
+ "evaluation_name": "MMLU-Pro (Economics)",
221
+ "source_data": {
222
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
223
+ "source_type": "hf_dataset",
224
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
225
+ "hf_split": "train",
226
+ "additional_details": {
227
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
228
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
229
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
230
+ "paper_url": "https://arxiv.org/abs/2406.01574",
231
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
232
+ "dataset_total_questions": "12000",
233
+ "prompt_style": "5-shot CoT"
234
+ }
235
+ },
236
+ "metric_config": {
237
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
238
+ "metric_id": "mmlu_pro/economics",
239
+ "metric_name": "MMLU-Pro (Economics)",
240
+ "metric_kind": "accuracy",
241
+ "metric_unit": "proportion",
242
+ "lower_is_better": false,
243
+ "score_type": "continuous",
244
+ "min_score": 0.0,
245
+ "max_score": 1.0,
246
+ "additional_details": {
247
+ "aggregation": "accuracy_over_subset",
248
+ "prompt_style": "5-shot CoT"
249
+ }
250
+ },
251
+ "score_details": {
252
+ "score": 0.7628
253
+ }
254
+ },
255
+ {
256
+ "evaluation_result_id": "mmlu_pro/engineering",
257
+ "evaluation_name": "MMLU-Pro (Engineering)",
258
+ "source_data": {
259
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
260
+ "source_type": "hf_dataset",
261
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
262
+ "hf_split": "train",
263
+ "additional_details": {
264
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
265
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
266
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
267
+ "paper_url": "https://arxiv.org/abs/2406.01574",
268
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
269
+ "dataset_total_questions": "12000",
270
+ "prompt_style": "5-shot CoT"
271
+ }
272
+ },
273
+ "metric_config": {
274
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
275
+ "metric_id": "mmlu_pro/engineering",
276
+ "metric_name": "MMLU-Pro (Engineering)",
277
+ "metric_kind": "accuracy",
278
+ "metric_unit": "proportion",
279
+ "lower_is_better": false,
280
+ "score_type": "continuous",
281
+ "min_score": 0.0,
282
+ "max_score": 1.0,
283
+ "additional_details": {
284
+ "aggregation": "accuracy_over_subset",
285
+ "prompt_style": "5-shot CoT"
286
+ }
287
+ },
288
+ "score_details": {
289
+ "score": 0.407
290
+ }
291
+ },
292
+ {
293
+ "evaluation_result_id": "mmlu_pro/health",
294
+ "evaluation_name": "MMLU-Pro (Health)",
295
+ "source_data": {
296
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
297
+ "source_type": "hf_dataset",
298
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
299
+ "hf_split": "train",
300
+ "additional_details": {
301
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
302
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
303
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
304
+ "paper_url": "https://arxiv.org/abs/2406.01574",
305
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
306
+ "dataset_total_questions": "12000",
307
+ "prompt_style": "5-shot CoT"
308
+ }
309
+ },
310
+ "metric_config": {
311
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
312
+ "metric_id": "mmlu_pro/health",
313
+ "metric_name": "MMLU-Pro (Health)",
314
+ "metric_kind": "accuracy",
315
+ "metric_unit": "proportion",
316
+ "lower_is_better": false,
317
+ "score_type": "continuous",
318
+ "min_score": 0.0,
319
+ "max_score": 1.0,
320
+ "additional_details": {
321
+ "aggregation": "accuracy_over_subset",
322
+ "prompt_style": "5-shot CoT"
323
+ }
324
+ },
325
+ "score_details": {
326
+ "score": 0.6284
327
+ }
328
+ },
329
+ {
330
+ "evaluation_result_id": "mmlu_pro/history",
331
+ "evaluation_name": "MMLU-Pro (History)",
332
+ "source_data": {
333
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
334
+ "source_type": "hf_dataset",
335
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
336
+ "hf_split": "train",
337
+ "additional_details": {
338
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
339
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
340
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
341
+ "paper_url": "https://arxiv.org/abs/2406.01574",
342
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
343
+ "dataset_total_questions": "12000",
344
+ "prompt_style": "5-shot CoT"
345
+ }
346
+ },
347
+ "metric_config": {
348
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
349
+ "metric_id": "mmlu_pro/history",
350
+ "metric_name": "MMLU-Pro (History)",
351
+ "metric_kind": "accuracy",
352
+ "metric_unit": "proportion",
353
+ "lower_is_better": false,
354
+ "score_type": "continuous",
355
+ "min_score": 0.0,
356
+ "max_score": 1.0,
357
+ "additional_details": {
358
+ "aggregation": "accuracy_over_subset",
359
+ "prompt_style": "5-shot CoT"
360
+ }
361
+ },
362
+ "score_details": {
363
+ "score": 0.5932
364
+ }
365
+ },
366
+ {
367
+ "evaluation_result_id": "mmlu_pro/law",
368
+ "evaluation_name": "MMLU-Pro (Law)",
369
+ "source_data": {
370
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
371
+ "source_type": "hf_dataset",
372
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
373
+ "hf_split": "train",
374
+ "additional_details": {
375
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
376
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
377
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
378
+ "paper_url": "https://arxiv.org/abs/2406.01574",
379
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
380
+ "dataset_total_questions": "12000",
381
+ "prompt_style": "5-shot CoT"
382
+ }
383
+ },
384
+ "metric_config": {
385
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
386
+ "metric_id": "mmlu_pro/law",
387
+ "metric_name": "MMLU-Pro (Law)",
388
+ "metric_kind": "accuracy",
389
+ "metric_unit": "proportion",
390
+ "lower_is_better": false,
391
+ "score_type": "continuous",
392
+ "min_score": 0.0,
393
+ "max_score": 1.0,
394
+ "additional_details": {
395
+ "aggregation": "accuracy_over_subset",
396
+ "prompt_style": "5-shot CoT"
397
+ }
398
+ },
399
+ "score_details": {
400
+ "score": 0.4286
401
+ }
402
+ },
403
+ {
404
+ "evaluation_result_id": "mmlu_pro/math",
405
+ "evaluation_name": "MMLU-Pro (Math)",
406
+ "source_data": {
407
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
408
+ "source_type": "hf_dataset",
409
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
410
+ "hf_split": "train",
411
+ "additional_details": {
412
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
413
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
414
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
415
+ "paper_url": "https://arxiv.org/abs/2406.01574",
416
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
417
+ "dataset_total_questions": "12000",
418
+ "prompt_style": "5-shot CoT"
419
+ }
420
+ },
421
+ "metric_config": {
422
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
423
+ "metric_id": "mmlu_pro/math",
424
+ "metric_name": "MMLU-Pro (Math)",
425
+ "metric_kind": "accuracy",
426
+ "metric_unit": "proportion",
427
+ "lower_is_better": false,
428
+ "score_type": "continuous",
429
+ "min_score": 0.0,
430
+ "max_score": 1.0,
431
+ "additional_details": {
432
+ "aggregation": "accuracy_over_subset",
433
+ "prompt_style": "5-shot CoT"
434
+ }
435
+ },
436
+ "score_details": {
437
+ "score": 0.6255
438
+ }
439
+ },
440
+ {
441
+ "evaluation_result_id": "mmlu_pro/philosophy",
442
+ "evaluation_name": "MMLU-Pro (Philosophy)",
443
+ "source_data": {
444
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
445
+ "source_type": "hf_dataset",
446
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
447
+ "hf_split": "train",
448
+ "additional_details": {
449
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
450
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
451
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
452
+ "paper_url": "https://arxiv.org/abs/2406.01574",
453
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
454
+ "dataset_total_questions": "12000",
455
+ "prompt_style": "5-shot CoT"
456
+ }
457
+ },
458
+ "metric_config": {
459
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
460
+ "metric_id": "mmlu_pro/philosophy",
461
+ "metric_name": "MMLU-Pro (Philosophy)",
462
+ "metric_kind": "accuracy",
463
+ "metric_unit": "proportion",
464
+ "lower_is_better": false,
465
+ "score_type": "continuous",
466
+ "min_score": 0.0,
467
+ "max_score": 1.0,
468
+ "additional_details": {
469
+ "aggregation": "accuracy_over_subset",
470
+ "prompt_style": "5-shot CoT"
471
+ }
472
+ },
473
+ "score_details": {
474
+ "score": 0.6052
475
+ }
476
+ },
477
+ {
478
+ "evaluation_result_id": "mmlu_pro/physics",
479
+ "evaluation_name": "MMLU-Pro (Physics)",
480
+ "source_data": {
481
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
482
+ "source_type": "hf_dataset",
483
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
484
+ "hf_split": "train",
485
+ "additional_details": {
486
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
487
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
488
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
489
+ "paper_url": "https://arxiv.org/abs/2406.01574",
490
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
491
+ "dataset_total_questions": "12000",
492
+ "prompt_style": "5-shot CoT"
493
+ }
494
+ },
495
+ "metric_config": {
496
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
497
+ "metric_id": "mmlu_pro/physics",
498
+ "metric_name": "MMLU-Pro (Physics)",
499
+ "metric_kind": "accuracy",
500
+ "metric_unit": "proportion",
501
+ "lower_is_better": false,
502
+ "score_type": "continuous",
503
+ "min_score": 0.0,
504
+ "max_score": 1.0,
505
+ "additional_details": {
506
+ "aggregation": "accuracy_over_subset",
507
+ "prompt_style": "5-shot CoT"
508
+ }
509
+ },
510
+ "score_details": {
511
+ "score": 0.7141
512
+ }
513
+ },
514
+ {
515
+ "evaluation_result_id": "mmlu_pro/psychology",
516
+ "evaluation_name": "MMLU-Pro (Psychology)",
517
+ "source_data": {
518
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
519
+ "source_type": "hf_dataset",
520
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
521
+ "hf_split": "train",
522
+ "additional_details": {
523
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
524
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
525
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
526
+ "paper_url": "https://arxiv.org/abs/2406.01574",
527
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
528
+ "dataset_total_questions": "12000",
529
+ "prompt_style": "5-shot CoT"
530
+ }
531
+ },
532
+ "metric_config": {
533
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
534
+ "metric_id": "mmlu_pro/psychology",
535
+ "metric_name": "MMLU-Pro (Psychology)",
536
+ "metric_kind": "accuracy",
537
+ "metric_unit": "proportion",
538
+ "lower_is_better": false,
539
+ "score_type": "continuous",
540
+ "min_score": 0.0,
541
+ "max_score": 1.0,
542
+ "additional_details": {
543
+ "aggregation": "accuracy_over_subset",
544
+ "prompt_style": "5-shot CoT"
545
+ }
546
+ },
547
+ "score_details": {
548
+ "score": 0.7623
549
+ }
550
+ },
551
+ {
552
+ "evaluation_result_id": "mmlu_pro/other",
553
+ "evaluation_name": "MMLU-Pro (Other)",
554
+ "source_data": {
555
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
556
+ "source_type": "hf_dataset",
557
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
558
+ "hf_split": "train",
559
+ "additional_details": {
560
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
561
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
562
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
563
+ "paper_url": "https://arxiv.org/abs/2406.01574",
564
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
565
+ "dataset_total_questions": "12000",
566
+ "prompt_style": "5-shot CoT"
567
+ }
568
+ },
569
+ "metric_config": {
570
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
571
+ "metric_id": "mmlu_pro/other",
572
+ "metric_name": "MMLU-Pro (Other)",
573
+ "metric_kind": "accuracy",
574
+ "metric_unit": "proportion",
575
+ "lower_is_better": false,
576
+ "score_type": "continuous",
577
+ "min_score": 0.0,
578
+ "max_score": 1.0,
579
+ "additional_details": {
580
+ "aggregation": "accuracy_over_subset",
581
+ "prompt_style": "5-shot CoT"
582
+ }
583
+ },
584
+ "score_details": {
585
+ "score": 0.6453
586
+ }
587
+ }
588
+ ]
589
+ }
data/mmlu_pro/google/gemini-1.5-flash/01bb7276-91e8-44eb-ba3a-c7df072f7a38.json ADDED
@@ -0,0 +1,589 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-1.5-flash/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-1.5-Flash",
25
+ "id": "google/gemini-1.5-flash",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-1.5-Flash",
29
+ "leaderboard_data_source": "TIGER-Lab"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.5912
68
+ }
69
+ },
70
+ {
71
+ "evaluation_result_id": "mmlu_pro/biology",
72
+ "evaluation_name": "MMLU-Pro (Biology)",
73
+ "source_data": {
74
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
75
+ "source_type": "hf_dataset",
76
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
77
+ "hf_split": "train",
78
+ "additional_details": {
79
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
80
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
81
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
82
+ "paper_url": "https://arxiv.org/abs/2406.01574",
83
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
84
+ "dataset_total_questions": "12000",
85
+ "prompt_style": "5-shot CoT"
86
+ }
87
+ },
88
+ "metric_config": {
89
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
90
+ "metric_id": "mmlu_pro/biology",
91
+ "metric_name": "MMLU-Pro (Biology)",
92
+ "metric_kind": "accuracy",
93
+ "metric_unit": "proportion",
94
+ "lower_is_better": false,
95
+ "score_type": "continuous",
96
+ "min_score": 0.0,
97
+ "max_score": 1.0,
98
+ "additional_details": {
99
+ "aggregation": "accuracy_over_subset",
100
+ "prompt_style": "5-shot CoT"
101
+ }
102
+ },
103
+ "score_details": {
104
+ "score": 0.8131
105
+ }
106
+ },
107
+ {
108
+ "evaluation_result_id": "mmlu_pro/business",
109
+ "evaluation_name": "MMLU-Pro (Business)",
110
+ "source_data": {
111
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
112
+ "source_type": "hf_dataset",
113
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
114
+ "hf_split": "train",
115
+ "additional_details": {
116
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
117
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
118
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
119
+ "paper_url": "https://arxiv.org/abs/2406.01574",
120
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
121
+ "dataset_total_questions": "12000",
122
+ "prompt_style": "5-shot CoT"
123
+ }
124
+ },
125
+ "metric_config": {
126
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
127
+ "metric_id": "mmlu_pro/business",
128
+ "metric_name": "MMLU-Pro (Business)",
129
+ "metric_kind": "accuracy",
130
+ "metric_unit": "proportion",
131
+ "lower_is_better": false,
132
+ "score_type": "continuous",
133
+ "min_score": 0.0,
134
+ "max_score": 1.0,
135
+ "additional_details": {
136
+ "aggregation": "accuracy_over_subset",
137
+ "prompt_style": "5-shot CoT"
138
+ }
139
+ },
140
+ "score_details": {
141
+ "score": 0.667
142
+ }
143
+ },
144
+ {
145
+ "evaluation_result_id": "mmlu_pro/chemistry",
146
+ "evaluation_name": "MMLU-Pro (Chemistry)",
147
+ "source_data": {
148
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
149
+ "source_type": "hf_dataset",
150
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
151
+ "hf_split": "train",
152
+ "additional_details": {
153
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
154
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
155
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
156
+ "paper_url": "https://arxiv.org/abs/2406.01574",
157
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
158
+ "dataset_total_questions": "12000",
159
+ "prompt_style": "5-shot CoT"
160
+ }
161
+ },
162
+ "metric_config": {
163
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
164
+ "metric_id": "mmlu_pro/chemistry",
165
+ "metric_name": "MMLU-Pro (Chemistry)",
166
+ "metric_kind": "accuracy",
167
+ "metric_unit": "proportion",
168
+ "lower_is_better": false,
169
+ "score_type": "continuous",
170
+ "min_score": 0.0,
171
+ "max_score": 1.0,
172
+ "additional_details": {
173
+ "aggregation": "accuracy_over_subset",
174
+ "prompt_style": "5-shot CoT"
175
+ }
176
+ },
177
+ "score_details": {
178
+ "score": 0.613
179
+ }
180
+ },
181
+ {
182
+ "evaluation_result_id": "mmlu_pro/computer_science",
183
+ "evaluation_name": "MMLU-Pro (Computer Science)",
184
+ "source_data": {
185
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
186
+ "source_type": "hf_dataset",
187
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
188
+ "hf_split": "train",
189
+ "additional_details": {
190
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
191
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
192
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
193
+ "paper_url": "https://arxiv.org/abs/2406.01574",
194
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
195
+ "dataset_total_questions": "12000",
196
+ "prompt_style": "5-shot CoT"
197
+ }
198
+ },
199
+ "metric_config": {
200
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
201
+ "metric_id": "mmlu_pro/computer_science",
202
+ "metric_name": "MMLU-Pro (Computer Science)",
203
+ "metric_kind": "accuracy",
204
+ "metric_unit": "proportion",
205
+ "lower_is_better": false,
206
+ "score_type": "continuous",
207
+ "min_score": 0.0,
208
+ "max_score": 1.0,
209
+ "additional_details": {
210
+ "aggregation": "accuracy_over_subset",
211
+ "prompt_style": "5-shot CoT"
212
+ }
213
+ },
214
+ "score_details": {
215
+ "score": 0.5951
216
+ }
217
+ },
218
+ {
219
+ "evaluation_result_id": "mmlu_pro/economics",
220
+ "evaluation_name": "MMLU-Pro (Economics)",
221
+ "source_data": {
222
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
223
+ "source_type": "hf_dataset",
224
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
225
+ "hf_split": "train",
226
+ "additional_details": {
227
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
228
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
229
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
230
+ "paper_url": "https://arxiv.org/abs/2406.01574",
231
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
232
+ "dataset_total_questions": "12000",
233
+ "prompt_style": "5-shot CoT"
234
+ }
235
+ },
236
+ "metric_config": {
237
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
238
+ "metric_id": "mmlu_pro/economics",
239
+ "metric_name": "MMLU-Pro (Economics)",
240
+ "metric_kind": "accuracy",
241
+ "metric_unit": "proportion",
242
+ "lower_is_better": false,
243
+ "score_type": "continuous",
244
+ "min_score": 0.0,
245
+ "max_score": 1.0,
246
+ "additional_details": {
247
+ "aggregation": "accuracy_over_subset",
248
+ "prompt_style": "5-shot CoT"
249
+ }
250
+ },
251
+ "score_details": {
252
+ "score": 0.6943
253
+ }
254
+ },
255
+ {
256
+ "evaluation_result_id": "mmlu_pro/engineering",
257
+ "evaluation_name": "MMLU-Pro (Engineering)",
258
+ "source_data": {
259
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
260
+ "source_type": "hf_dataset",
261
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
262
+ "hf_split": "train",
263
+ "additional_details": {
264
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
265
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
266
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
267
+ "paper_url": "https://arxiv.org/abs/2406.01574",
268
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
269
+ "dataset_total_questions": "12000",
270
+ "prompt_style": "5-shot CoT"
271
+ }
272
+ },
273
+ "metric_config": {
274
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
275
+ "metric_id": "mmlu_pro/engineering",
276
+ "metric_name": "MMLU-Pro (Engineering)",
277
+ "metric_kind": "accuracy",
278
+ "metric_unit": "proportion",
279
+ "lower_is_better": false,
280
+ "score_type": "continuous",
281
+ "min_score": 0.0,
282
+ "max_score": 1.0,
283
+ "additional_details": {
284
+ "aggregation": "accuracy_over_subset",
285
+ "prompt_style": "5-shot CoT"
286
+ }
287
+ },
288
+ "score_details": {
289
+ "score": 0.4416
290
+ }
291
+ },
292
+ {
293
+ "evaluation_result_id": "mmlu_pro/health",
294
+ "evaluation_name": "MMLU-Pro (Health)",
295
+ "source_data": {
296
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
297
+ "source_type": "hf_dataset",
298
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
299
+ "hf_split": "train",
300
+ "additional_details": {
301
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
302
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
303
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
304
+ "paper_url": "https://arxiv.org/abs/2406.01574",
305
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
306
+ "dataset_total_questions": "12000",
307
+ "prompt_style": "5-shot CoT"
308
+ }
309
+ },
310
+ "metric_config": {
311
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
312
+ "metric_id": "mmlu_pro/health",
313
+ "metric_name": "MMLU-Pro (Health)",
314
+ "metric_kind": "accuracy",
315
+ "metric_unit": "proportion",
316
+ "lower_is_better": false,
317
+ "score_type": "continuous",
318
+ "min_score": 0.0,
319
+ "max_score": 1.0,
320
+ "additional_details": {
321
+ "aggregation": "accuracy_over_subset",
322
+ "prompt_style": "5-shot CoT"
323
+ }
324
+ },
325
+ "score_details": {
326
+ "score": 0.6039
327
+ }
328
+ },
329
+ {
330
+ "evaluation_result_id": "mmlu_pro/history",
331
+ "evaluation_name": "MMLU-Pro (History)",
332
+ "source_data": {
333
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
334
+ "source_type": "hf_dataset",
335
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
336
+ "hf_split": "train",
337
+ "additional_details": {
338
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
339
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
340
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
341
+ "paper_url": "https://arxiv.org/abs/2406.01574",
342
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
343
+ "dataset_total_questions": "12000",
344
+ "prompt_style": "5-shot CoT"
345
+ }
346
+ },
347
+ "metric_config": {
348
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
349
+ "metric_id": "mmlu_pro/history",
350
+ "metric_name": "MMLU-Pro (History)",
351
+ "metric_kind": "accuracy",
352
+ "metric_unit": "proportion",
353
+ "lower_is_better": false,
354
+ "score_type": "continuous",
355
+ "min_score": 0.0,
356
+ "max_score": 1.0,
357
+ "additional_details": {
358
+ "aggregation": "accuracy_over_subset",
359
+ "prompt_style": "5-shot CoT"
360
+ }
361
+ },
362
+ "score_details": {
363
+ "score": 0.538
364
+ }
365
+ },
366
+ {
367
+ "evaluation_result_id": "mmlu_pro/law",
368
+ "evaluation_name": "MMLU-Pro (Law)",
369
+ "source_data": {
370
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
371
+ "source_type": "hf_dataset",
372
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
373
+ "hf_split": "train",
374
+ "additional_details": {
375
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
376
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
377
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
378
+ "paper_url": "https://arxiv.org/abs/2406.01574",
379
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
380
+ "dataset_total_questions": "12000",
381
+ "prompt_style": "5-shot CoT"
382
+ }
383
+ },
384
+ "metric_config": {
385
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
386
+ "metric_id": "mmlu_pro/law",
387
+ "metric_name": "MMLU-Pro (Law)",
388
+ "metric_kind": "accuracy",
389
+ "metric_unit": "proportion",
390
+ "lower_is_better": false,
391
+ "score_type": "continuous",
392
+ "min_score": 0.0,
393
+ "max_score": 1.0,
394
+ "additional_details": {
395
+ "aggregation": "accuracy_over_subset",
396
+ "prompt_style": "5-shot CoT"
397
+ }
398
+ },
399
+ "score_details": {
400
+ "score": 0.3732
401
+ }
402
+ },
403
+ {
404
+ "evaluation_result_id": "mmlu_pro/math",
405
+ "evaluation_name": "MMLU-Pro (Math)",
406
+ "source_data": {
407
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
408
+ "source_type": "hf_dataset",
409
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
410
+ "hf_split": "train",
411
+ "additional_details": {
412
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
413
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
414
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
415
+ "paper_url": "https://arxiv.org/abs/2406.01574",
416
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
417
+ "dataset_total_questions": "12000",
418
+ "prompt_style": "5-shot CoT"
419
+ }
420
+ },
421
+ "metric_config": {
422
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
423
+ "metric_id": "mmlu_pro/math",
424
+ "metric_name": "MMLU-Pro (Math)",
425
+ "metric_kind": "accuracy",
426
+ "metric_unit": "proportion",
427
+ "lower_is_better": false,
428
+ "score_type": "continuous",
429
+ "min_score": 0.0,
430
+ "max_score": 1.0,
431
+ "additional_details": {
432
+ "aggregation": "accuracy_over_subset",
433
+ "prompt_style": "5-shot CoT"
434
+ }
435
+ },
436
+ "score_details": {
437
+ "score": 0.5958
438
+ }
439
+ },
440
+ {
441
+ "evaluation_result_id": "mmlu_pro/philosophy",
442
+ "evaluation_name": "MMLU-Pro (Philosophy)",
443
+ "source_data": {
444
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
445
+ "source_type": "hf_dataset",
446
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
447
+ "hf_split": "train",
448
+ "additional_details": {
449
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
450
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
451
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
452
+ "paper_url": "https://arxiv.org/abs/2406.01574",
453
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
454
+ "dataset_total_questions": "12000",
455
+ "prompt_style": "5-shot CoT"
456
+ }
457
+ },
458
+ "metric_config": {
459
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
460
+ "metric_id": "mmlu_pro/philosophy",
461
+ "metric_name": "MMLU-Pro (Philosophy)",
462
+ "metric_kind": "accuracy",
463
+ "metric_unit": "proportion",
464
+ "lower_is_better": false,
465
+ "score_type": "continuous",
466
+ "min_score": 0.0,
467
+ "max_score": 1.0,
468
+ "additional_details": {
469
+ "aggregation": "accuracy_over_subset",
470
+ "prompt_style": "5-shot CoT"
471
+ }
472
+ },
473
+ "score_details": {
474
+ "score": 0.4949
475
+ }
476
+ },
477
+ {
478
+ "evaluation_result_id": "mmlu_pro/physics",
479
+ "evaluation_name": "MMLU-Pro (Physics)",
480
+ "source_data": {
481
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
482
+ "source_type": "hf_dataset",
483
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
484
+ "hf_split": "train",
485
+ "additional_details": {
486
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
487
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
488
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
489
+ "paper_url": "https://arxiv.org/abs/2406.01574",
490
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
491
+ "dataset_total_questions": "12000",
492
+ "prompt_style": "5-shot CoT"
493
+ }
494
+ },
495
+ "metric_config": {
496
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
497
+ "metric_id": "mmlu_pro/physics",
498
+ "metric_name": "MMLU-Pro (Physics)",
499
+ "metric_kind": "accuracy",
500
+ "metric_unit": "proportion",
501
+ "lower_is_better": false,
502
+ "score_type": "continuous",
503
+ "min_score": 0.0,
504
+ "max_score": 1.0,
505
+ "additional_details": {
506
+ "aggregation": "accuracy_over_subset",
507
+ "prompt_style": "5-shot CoT"
508
+ }
509
+ },
510
+ "score_details": {
511
+ "score": 0.612
512
+ }
513
+ },
514
+ {
515
+ "evaluation_result_id": "mmlu_pro/psychology",
516
+ "evaluation_name": "MMLU-Pro (Psychology)",
517
+ "source_data": {
518
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
519
+ "source_type": "hf_dataset",
520
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
521
+ "hf_split": "train",
522
+ "additional_details": {
523
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
524
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
525
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
526
+ "paper_url": "https://arxiv.org/abs/2406.01574",
527
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
528
+ "dataset_total_questions": "12000",
529
+ "prompt_style": "5-shot CoT"
530
+ }
531
+ },
532
+ "metric_config": {
533
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
534
+ "metric_id": "mmlu_pro/psychology",
535
+ "metric_name": "MMLU-Pro (Psychology)",
536
+ "metric_kind": "accuracy",
537
+ "metric_unit": "proportion",
538
+ "lower_is_better": false,
539
+ "score_type": "continuous",
540
+ "min_score": 0.0,
541
+ "max_score": 1.0,
542
+ "additional_details": {
543
+ "aggregation": "accuracy_over_subset",
544
+ "prompt_style": "5-shot CoT"
545
+ }
546
+ },
547
+ "score_details": {
548
+ "score": 0.7005
549
+ }
550
+ },
551
+ {
552
+ "evaluation_result_id": "mmlu_pro/other",
553
+ "evaluation_name": "MMLU-Pro (Other)",
554
+ "source_data": {
555
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
556
+ "source_type": "hf_dataset",
557
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
558
+ "hf_split": "train",
559
+ "additional_details": {
560
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
561
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
562
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
563
+ "paper_url": "https://arxiv.org/abs/2406.01574",
564
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
565
+ "dataset_total_questions": "12000",
566
+ "prompt_style": "5-shot CoT"
567
+ }
568
+ },
569
+ "metric_config": {
570
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
571
+ "metric_id": "mmlu_pro/other",
572
+ "metric_name": "MMLU-Pro (Other)",
573
+ "metric_kind": "accuracy",
574
+ "metric_unit": "proportion",
575
+ "lower_is_better": false,
576
+ "score_type": "continuous",
577
+ "min_score": 0.0,
578
+ "max_score": 1.0,
579
+ "additional_details": {
580
+ "aggregation": "accuracy_over_subset",
581
+ "prompt_style": "5-shot CoT"
582
+ }
583
+ },
584
+ "score_details": {
585
+ "score": 0.58
586
+ }
587
+ }
588
+ ]
589
+ }
data/mmlu_pro/google/gemini-1.5-pro-002/b3ffcd12-3a74-4f3f-9acf-7ec11615a88d.json ADDED
@@ -0,0 +1,589 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-1.5-pro-002/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-1.5-Pro-002",
25
+ "id": "google/gemini-1.5-pro-002",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-1.5-Pro-002",
29
+ "leaderboard_data_source": "TIGER-Lab"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.7025
68
+ }
69
+ },
70
+ {
71
+ "evaluation_result_id": "mmlu_pro/biology",
72
+ "evaluation_name": "MMLU-Pro (Biology)",
73
+ "source_data": {
74
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
75
+ "source_type": "hf_dataset",
76
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
77
+ "hf_split": "train",
78
+ "additional_details": {
79
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
80
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
81
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
82
+ "paper_url": "https://arxiv.org/abs/2406.01574",
83
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
84
+ "dataset_total_questions": "12000",
85
+ "prompt_style": "5-shot CoT"
86
+ }
87
+ },
88
+ "metric_config": {
89
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
90
+ "metric_id": "mmlu_pro/biology",
91
+ "metric_name": "MMLU-Pro (Biology)",
92
+ "metric_kind": "accuracy",
93
+ "metric_unit": "proportion",
94
+ "lower_is_better": false,
95
+ "score_type": "continuous",
96
+ "min_score": 0.0,
97
+ "max_score": 1.0,
98
+ "additional_details": {
99
+ "aggregation": "accuracy_over_subset",
100
+ "prompt_style": "5-shot CoT"
101
+ }
102
+ },
103
+ "score_details": {
104
+ "score": 0.8645
105
+ }
106
+ },
107
+ {
108
+ "evaluation_result_id": "mmlu_pro/business",
109
+ "evaluation_name": "MMLU-Pro (Business)",
110
+ "source_data": {
111
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
112
+ "source_type": "hf_dataset",
113
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
114
+ "hf_split": "train",
115
+ "additional_details": {
116
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
117
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
118
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
119
+ "paper_url": "https://arxiv.org/abs/2406.01574",
120
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
121
+ "dataset_total_questions": "12000",
122
+ "prompt_style": "5-shot CoT"
123
+ }
124
+ },
125
+ "metric_config": {
126
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
127
+ "metric_id": "mmlu_pro/business",
128
+ "metric_name": "MMLU-Pro (Business)",
129
+ "metric_kind": "accuracy",
130
+ "metric_unit": "proportion",
131
+ "lower_is_better": false,
132
+ "score_type": "continuous",
133
+ "min_score": 0.0,
134
+ "max_score": 1.0,
135
+ "additional_details": {
136
+ "aggregation": "accuracy_over_subset",
137
+ "prompt_style": "5-shot CoT"
138
+ }
139
+ },
140
+ "score_details": {
141
+ "score": 0.8094
142
+ }
143
+ },
144
+ {
145
+ "evaluation_result_id": "mmlu_pro/chemistry",
146
+ "evaluation_name": "MMLU-Pro (Chemistry)",
147
+ "source_data": {
148
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
149
+ "source_type": "hf_dataset",
150
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
151
+ "hf_split": "train",
152
+ "additional_details": {
153
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
154
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
155
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
156
+ "paper_url": "https://arxiv.org/abs/2406.01574",
157
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
158
+ "dataset_total_questions": "12000",
159
+ "prompt_style": "5-shot CoT"
160
+ }
161
+ },
162
+ "metric_config": {
163
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
164
+ "metric_id": "mmlu_pro/chemistry",
165
+ "metric_name": "MMLU-Pro (Chemistry)",
166
+ "metric_kind": "accuracy",
167
+ "metric_unit": "proportion",
168
+ "lower_is_better": false,
169
+ "score_type": "continuous",
170
+ "min_score": 0.0,
171
+ "max_score": 1.0,
172
+ "additional_details": {
173
+ "aggregation": "accuracy_over_subset",
174
+ "prompt_style": "5-shot CoT"
175
+ }
176
+ },
177
+ "score_details": {
178
+ "score": 0.6221
179
+ }
180
+ },
181
+ {
182
+ "evaluation_result_id": "mmlu_pro/computer_science",
183
+ "evaluation_name": "MMLU-Pro (Computer Science)",
184
+ "source_data": {
185
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
186
+ "source_type": "hf_dataset",
187
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
188
+ "hf_split": "train",
189
+ "additional_details": {
190
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
191
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
192
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
193
+ "paper_url": "https://arxiv.org/abs/2406.01574",
194
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
195
+ "dataset_total_questions": "12000",
196
+ "prompt_style": "5-shot CoT"
197
+ }
198
+ },
199
+ "metric_config": {
200
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
201
+ "metric_id": "mmlu_pro/computer_science",
202
+ "metric_name": "MMLU-Pro (Computer Science)",
203
+ "metric_kind": "accuracy",
204
+ "metric_unit": "proportion",
205
+ "lower_is_better": false,
206
+ "score_type": "continuous",
207
+ "min_score": 0.0,
208
+ "max_score": 1.0,
209
+ "additional_details": {
210
+ "aggregation": "accuracy_over_subset",
211
+ "prompt_style": "5-shot CoT"
212
+ }
213
+ },
214
+ "score_details": {
215
+ "score": 0.7122
216
+ }
217
+ },
218
+ {
219
+ "evaluation_result_id": "mmlu_pro/economics",
220
+ "evaluation_name": "MMLU-Pro (Economics)",
221
+ "source_data": {
222
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
223
+ "source_type": "hf_dataset",
224
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
225
+ "hf_split": "train",
226
+ "additional_details": {
227
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
228
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
229
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
230
+ "paper_url": "https://arxiv.org/abs/2406.01574",
231
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
232
+ "dataset_total_questions": "12000",
233
+ "prompt_style": "5-shot CoT"
234
+ }
235
+ },
236
+ "metric_config": {
237
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
238
+ "metric_id": "mmlu_pro/economics",
239
+ "metric_name": "MMLU-Pro (Economics)",
240
+ "metric_kind": "accuracy",
241
+ "metric_unit": "proportion",
242
+ "lower_is_better": false,
243
+ "score_type": "continuous",
244
+ "min_score": 0.0,
245
+ "max_score": 1.0,
246
+ "additional_details": {
247
+ "aggregation": "accuracy_over_subset",
248
+ "prompt_style": "5-shot CoT"
249
+ }
250
+ },
251
+ "score_details": {
252
+ "score": 0.8171
253
+ }
254
+ },
255
+ {
256
+ "evaluation_result_id": "mmlu_pro/engineering",
257
+ "evaluation_name": "MMLU-Pro (Engineering)",
258
+ "source_data": {
259
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
260
+ "source_type": "hf_dataset",
261
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
262
+ "hf_split": "train",
263
+ "additional_details": {
264
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
265
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
266
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
267
+ "paper_url": "https://arxiv.org/abs/2406.01574",
268
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
269
+ "dataset_total_questions": "12000",
270
+ "prompt_style": "5-shot CoT"
271
+ }
272
+ },
273
+ "metric_config": {
274
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
275
+ "metric_id": "mmlu_pro/engineering",
276
+ "metric_name": "MMLU-Pro (Engineering)",
277
+ "metric_kind": "accuracy",
278
+ "metric_unit": "proportion",
279
+ "lower_is_better": false,
280
+ "score_type": "continuous",
281
+ "min_score": 0.0,
282
+ "max_score": 1.0,
283
+ "additional_details": {
284
+ "aggregation": "accuracy_over_subset",
285
+ "prompt_style": "5-shot CoT"
286
+ }
287
+ },
288
+ "score_details": {
289
+ "score": 0.5899
290
+ }
291
+ },
292
+ {
293
+ "evaluation_result_id": "mmlu_pro/health",
294
+ "evaluation_name": "MMLU-Pro (Health)",
295
+ "source_data": {
296
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
297
+ "source_type": "hf_dataset",
298
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
299
+ "hf_split": "train",
300
+ "additional_details": {
301
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
302
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
303
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
304
+ "paper_url": "https://arxiv.org/abs/2406.01574",
305
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
306
+ "dataset_total_questions": "12000",
307
+ "prompt_style": "5-shot CoT"
308
+ }
309
+ },
310
+ "metric_config": {
311
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
312
+ "metric_id": "mmlu_pro/health",
313
+ "metric_name": "MMLU-Pro (Health)",
314
+ "metric_kind": "accuracy",
315
+ "metric_unit": "proportion",
316
+ "lower_is_better": false,
317
+ "score_type": "continuous",
318
+ "min_score": 0.0,
319
+ "max_score": 1.0,
320
+ "additional_details": {
321
+ "aggregation": "accuracy_over_subset",
322
+ "prompt_style": "5-shot CoT"
323
+ }
324
+ },
325
+ "score_details": {
326
+ "score": 0.7479
327
+ }
328
+ },
329
+ {
330
+ "evaluation_result_id": "mmlu_pro/history",
331
+ "evaluation_name": "MMLU-Pro (History)",
332
+ "source_data": {
333
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
334
+ "source_type": "hf_dataset",
335
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
336
+ "hf_split": "train",
337
+ "additional_details": {
338
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
339
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
340
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
341
+ "paper_url": "https://arxiv.org/abs/2406.01574",
342
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
343
+ "dataset_total_questions": "12000",
344
+ "prompt_style": "5-shot CoT"
345
+ }
346
+ },
347
+ "metric_config": {
348
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
349
+ "metric_id": "mmlu_pro/history",
350
+ "metric_name": "MMLU-Pro (History)",
351
+ "metric_kind": "accuracy",
352
+ "metric_unit": "proportion",
353
+ "lower_is_better": false,
354
+ "score_type": "continuous",
355
+ "min_score": 0.0,
356
+ "max_score": 1.0,
357
+ "additional_details": {
358
+ "aggregation": "accuracy_over_subset",
359
+ "prompt_style": "5-shot CoT"
360
+ }
361
+ },
362
+ "score_details": {
363
+ "score": 0.7008
364
+ }
365
+ },
366
+ {
367
+ "evaluation_result_id": "mmlu_pro/law",
368
+ "evaluation_name": "MMLU-Pro (Law)",
369
+ "source_data": {
370
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
371
+ "source_type": "hf_dataset",
372
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
373
+ "hf_split": "train",
374
+ "additional_details": {
375
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
376
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
377
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
378
+ "paper_url": "https://arxiv.org/abs/2406.01574",
379
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
380
+ "dataset_total_questions": "12000",
381
+ "prompt_style": "5-shot CoT"
382
+ }
383
+ },
384
+ "metric_config": {
385
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
386
+ "metric_id": "mmlu_pro/law",
387
+ "metric_name": "MMLU-Pro (Law)",
388
+ "metric_kind": "accuracy",
389
+ "metric_unit": "proportion",
390
+ "lower_is_better": false,
391
+ "score_type": "continuous",
392
+ "min_score": 0.0,
393
+ "max_score": 1.0,
394
+ "additional_details": {
395
+ "aggregation": "accuracy_over_subset",
396
+ "prompt_style": "5-shot CoT"
397
+ }
398
+ },
399
+ "score_details": {
400
+ "score": 0.5522
401
+ }
402
+ },
403
+ {
404
+ "evaluation_result_id": "mmlu_pro/math",
405
+ "evaluation_name": "MMLU-Pro (Math)",
406
+ "source_data": {
407
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
408
+ "source_type": "hf_dataset",
409
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
410
+ "hf_split": "train",
411
+ "additional_details": {
412
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
413
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
414
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
415
+ "paper_url": "https://arxiv.org/abs/2406.01574",
416
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
417
+ "dataset_total_questions": "12000",
418
+ "prompt_style": "5-shot CoT"
419
+ }
420
+ },
421
+ "metric_config": {
422
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
423
+ "metric_id": "mmlu_pro/math",
424
+ "metric_name": "MMLU-Pro (Math)",
425
+ "metric_kind": "accuracy",
426
+ "metric_unit": "proportion",
427
+ "lower_is_better": false,
428
+ "score_type": "continuous",
429
+ "min_score": 0.0,
430
+ "max_score": 1.0,
431
+ "additional_details": {
432
+ "aggregation": "accuracy_over_subset",
433
+ "prompt_style": "5-shot CoT"
434
+ }
435
+ },
436
+ "score_details": {
437
+ "score": 0.5174
438
+ }
439
+ },
440
+ {
441
+ "evaluation_result_id": "mmlu_pro/philosophy",
442
+ "evaluation_name": "MMLU-Pro (Philosophy)",
443
+ "source_data": {
444
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
445
+ "source_type": "hf_dataset",
446
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
447
+ "hf_split": "train",
448
+ "additional_details": {
449
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
450
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
451
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
452
+ "paper_url": "https://arxiv.org/abs/2406.01574",
453
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
454
+ "dataset_total_questions": "12000",
455
+ "prompt_style": "5-shot CoT"
456
+ }
457
+ },
458
+ "metric_config": {
459
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
460
+ "metric_id": "mmlu_pro/philosophy",
461
+ "metric_name": "MMLU-Pro (Philosophy)",
462
+ "metric_kind": "accuracy",
463
+ "metric_unit": "proportion",
464
+ "lower_is_better": false,
465
+ "score_type": "continuous",
466
+ "min_score": 0.0,
467
+ "max_score": 1.0,
468
+ "additional_details": {
469
+ "aggregation": "accuracy_over_subset",
470
+ "prompt_style": "5-shot CoT"
471
+ }
472
+ },
473
+ "score_details": {
474
+ "score": 0.7234
475
+ }
476
+ },
477
+ {
478
+ "evaluation_result_id": "mmlu_pro/physics",
479
+ "evaluation_name": "MMLU-Pro (Physics)",
480
+ "source_data": {
481
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
482
+ "source_type": "hf_dataset",
483
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
484
+ "hf_split": "train",
485
+ "additional_details": {
486
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
487
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
488
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
489
+ "paper_url": "https://arxiv.org/abs/2406.01574",
490
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
491
+ "dataset_total_questions": "12000",
492
+ "prompt_style": "5-shot CoT"
493
+ }
494
+ },
495
+ "metric_config": {
496
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
497
+ "metric_id": "mmlu_pro/physics",
498
+ "metric_name": "MMLU-Pro (Physics)",
499
+ "metric_kind": "accuracy",
500
+ "metric_unit": "proportion",
501
+ "lower_is_better": false,
502
+ "score_type": "continuous",
503
+ "min_score": 0.0,
504
+ "max_score": 1.0,
505
+ "additional_details": {
506
+ "aggregation": "accuracy_over_subset",
507
+ "prompt_style": "5-shot CoT"
508
+ }
509
+ },
510
+ "score_details": {
511
+ "score": 0.8072
512
+ }
513
+ },
514
+ {
515
+ "evaluation_result_id": "mmlu_pro/psychology",
516
+ "evaluation_name": "MMLU-Pro (Psychology)",
517
+ "source_data": {
518
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
519
+ "source_type": "hf_dataset",
520
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
521
+ "hf_split": "train",
522
+ "additional_details": {
523
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
524
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
525
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
526
+ "paper_url": "https://arxiv.org/abs/2406.01574",
527
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
528
+ "dataset_total_questions": "12000",
529
+ "prompt_style": "5-shot CoT"
530
+ }
531
+ },
532
+ "metric_config": {
533
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
534
+ "metric_id": "mmlu_pro/psychology",
535
+ "metric_name": "MMLU-Pro (Psychology)",
536
+ "metric_kind": "accuracy",
537
+ "metric_unit": "proportion",
538
+ "lower_is_better": false,
539
+ "score_type": "continuous",
540
+ "min_score": 0.0,
541
+ "max_score": 1.0,
542
+ "additional_details": {
543
+ "aggregation": "accuracy_over_subset",
544
+ "prompt_style": "5-shot CoT"
545
+ }
546
+ },
547
+ "score_details": {
548
+ "score": 0.8294
549
+ }
550
+ },
551
+ {
552
+ "evaluation_result_id": "mmlu_pro/other",
553
+ "evaluation_name": "MMLU-Pro (Other)",
554
+ "source_data": {
555
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
556
+ "source_type": "hf_dataset",
557
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
558
+ "hf_split": "train",
559
+ "additional_details": {
560
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
561
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
562
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
563
+ "paper_url": "https://arxiv.org/abs/2406.01574",
564
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
565
+ "dataset_total_questions": "12000",
566
+ "prompt_style": "5-shot CoT"
567
+ }
568
+ },
569
+ "metric_config": {
570
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
571
+ "metric_id": "mmlu_pro/other",
572
+ "metric_name": "MMLU-Pro (Other)",
573
+ "metric_kind": "accuracy",
574
+ "metric_unit": "proportion",
575
+ "lower_is_better": false,
576
+ "score_type": "continuous",
577
+ "min_score": 0.0,
578
+ "max_score": 1.0,
579
+ "additional_details": {
580
+ "aggregation": "accuracy_over_subset",
581
+ "prompt_style": "5-shot CoT"
582
+ }
583
+ },
584
+ "score_details": {
585
+ "score": 0.7359
586
+ }
587
+ }
588
+ ]
589
+ }
data/mmlu_pro/google/gemini-1.5-pro/fd7ad2ae-dd54-4275-8db7-7da0110eb2be.json ADDED
@@ -0,0 +1,589 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-1.5-pro/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-1.5-Pro",
25
+ "id": "google/gemini-1.5-pro",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-1.5-Pro",
29
+ "leaderboard_data_source": "Self-Reported"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.6903
68
+ }
69
+ },
70
+ {
71
+ "evaluation_result_id": "mmlu_pro/biology",
72
+ "evaluation_name": "MMLU-Pro (Biology)",
73
+ "source_data": {
74
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
75
+ "source_type": "hf_dataset",
76
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
77
+ "hf_split": "train",
78
+ "additional_details": {
79
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
80
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
81
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
82
+ "paper_url": "https://arxiv.org/abs/2406.01574",
83
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
84
+ "dataset_total_questions": "12000",
85
+ "prompt_style": "5-shot CoT"
86
+ }
87
+ },
88
+ "metric_config": {
89
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
90
+ "metric_id": "mmlu_pro/biology",
91
+ "metric_name": "MMLU-Pro (Biology)",
92
+ "metric_kind": "accuracy",
93
+ "metric_unit": "proportion",
94
+ "lower_is_better": false,
95
+ "score_type": "continuous",
96
+ "min_score": 0.0,
97
+ "max_score": 1.0,
98
+ "additional_details": {
99
+ "aggregation": "accuracy_over_subset",
100
+ "prompt_style": "5-shot CoT"
101
+ }
102
+ },
103
+ "score_details": {
104
+ "score": 0.8466
105
+ }
106
+ },
107
+ {
108
+ "evaluation_result_id": "mmlu_pro/business",
109
+ "evaluation_name": "MMLU-Pro (Business)",
110
+ "source_data": {
111
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
112
+ "source_type": "hf_dataset",
113
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
114
+ "hf_split": "train",
115
+ "additional_details": {
116
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
117
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
118
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
119
+ "paper_url": "https://arxiv.org/abs/2406.01574",
120
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
121
+ "dataset_total_questions": "12000",
122
+ "prompt_style": "5-shot CoT"
123
+ }
124
+ },
125
+ "metric_config": {
126
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
127
+ "metric_id": "mmlu_pro/business",
128
+ "metric_name": "MMLU-Pro (Business)",
129
+ "metric_kind": "accuracy",
130
+ "metric_unit": "proportion",
131
+ "lower_is_better": false,
132
+ "score_type": "continuous",
133
+ "min_score": 0.0,
134
+ "max_score": 1.0,
135
+ "additional_details": {
136
+ "aggregation": "accuracy_over_subset",
137
+ "prompt_style": "5-shot CoT"
138
+ }
139
+ },
140
+ "score_details": {
141
+ "score": 0.7288
142
+ }
143
+ },
144
+ {
145
+ "evaluation_result_id": "mmlu_pro/chemistry",
146
+ "evaluation_name": "MMLU-Pro (Chemistry)",
147
+ "source_data": {
148
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
149
+ "source_type": "hf_dataset",
150
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
151
+ "hf_split": "train",
152
+ "additional_details": {
153
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
154
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
155
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
156
+ "paper_url": "https://arxiv.org/abs/2406.01574",
157
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
158
+ "dataset_total_questions": "12000",
159
+ "prompt_style": "5-shot CoT"
160
+ }
161
+ },
162
+ "metric_config": {
163
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
164
+ "metric_id": "mmlu_pro/chemistry",
165
+ "metric_name": "MMLU-Pro (Chemistry)",
166
+ "metric_kind": "accuracy",
167
+ "metric_unit": "proportion",
168
+ "lower_is_better": false,
169
+ "score_type": "continuous",
170
+ "min_score": 0.0,
171
+ "max_score": 1.0,
172
+ "additional_details": {
173
+ "aggregation": "accuracy_over_subset",
174
+ "prompt_style": "5-shot CoT"
175
+ }
176
+ },
177
+ "score_details": {
178
+ "score": 0.7032
179
+ }
180
+ },
181
+ {
182
+ "evaluation_result_id": "mmlu_pro/computer_science",
183
+ "evaluation_name": "MMLU-Pro (Computer Science)",
184
+ "source_data": {
185
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
186
+ "source_type": "hf_dataset",
187
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
188
+ "hf_split": "train",
189
+ "additional_details": {
190
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
191
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
192
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
193
+ "paper_url": "https://arxiv.org/abs/2406.01574",
194
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
195
+ "dataset_total_questions": "12000",
196
+ "prompt_style": "5-shot CoT"
197
+ }
198
+ },
199
+ "metric_config": {
200
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
201
+ "metric_id": "mmlu_pro/computer_science",
202
+ "metric_name": "MMLU-Pro (Computer Science)",
203
+ "metric_kind": "accuracy",
204
+ "metric_unit": "proportion",
205
+ "lower_is_better": false,
206
+ "score_type": "continuous",
207
+ "min_score": 0.0,
208
+ "max_score": 1.0,
209
+ "additional_details": {
210
+ "aggregation": "accuracy_over_subset",
211
+ "prompt_style": "5-shot CoT"
212
+ }
213
+ },
214
+ "score_details": {
215
+ "score": 0.7293
216
+ }
217
+ },
218
+ {
219
+ "evaluation_result_id": "mmlu_pro/economics",
220
+ "evaluation_name": "MMLU-Pro (Economics)",
221
+ "source_data": {
222
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
223
+ "source_type": "hf_dataset",
224
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
225
+ "hf_split": "train",
226
+ "additional_details": {
227
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
228
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
229
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
230
+ "paper_url": "https://arxiv.org/abs/2406.01574",
231
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
232
+ "dataset_total_questions": "12000",
233
+ "prompt_style": "5-shot CoT"
234
+ }
235
+ },
236
+ "metric_config": {
237
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
238
+ "metric_id": "mmlu_pro/economics",
239
+ "metric_name": "MMLU-Pro (Economics)",
240
+ "metric_kind": "accuracy",
241
+ "metric_unit": "proportion",
242
+ "lower_is_better": false,
243
+ "score_type": "continuous",
244
+ "min_score": 0.0,
245
+ "max_score": 1.0,
246
+ "additional_details": {
247
+ "aggregation": "accuracy_over_subset",
248
+ "prompt_style": "5-shot CoT"
249
+ }
250
+ },
251
+ "score_details": {
252
+ "score": 0.7844
253
+ }
254
+ },
255
+ {
256
+ "evaluation_result_id": "mmlu_pro/engineering",
257
+ "evaluation_name": "MMLU-Pro (Engineering)",
258
+ "source_data": {
259
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
260
+ "source_type": "hf_dataset",
261
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
262
+ "hf_split": "train",
263
+ "additional_details": {
264
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
265
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
266
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
267
+ "paper_url": "https://arxiv.org/abs/2406.01574",
268
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
269
+ "dataset_total_questions": "12000",
270
+ "prompt_style": "5-shot CoT"
271
+ }
272
+ },
273
+ "metric_config": {
274
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
275
+ "metric_id": "mmlu_pro/engineering",
276
+ "metric_name": "MMLU-Pro (Engineering)",
277
+ "metric_kind": "accuracy",
278
+ "metric_unit": "proportion",
279
+ "lower_is_better": false,
280
+ "score_type": "continuous",
281
+ "min_score": 0.0,
282
+ "max_score": 1.0,
283
+ "additional_details": {
284
+ "aggregation": "accuracy_over_subset",
285
+ "prompt_style": "5-shot CoT"
286
+ }
287
+ },
288
+ "score_details": {
289
+ "score": 0.4871
290
+ }
291
+ },
292
+ {
293
+ "evaluation_result_id": "mmlu_pro/health",
294
+ "evaluation_name": "MMLU-Pro (Health)",
295
+ "source_data": {
296
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
297
+ "source_type": "hf_dataset",
298
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
299
+ "hf_split": "train",
300
+ "additional_details": {
301
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
302
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
303
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
304
+ "paper_url": "https://arxiv.org/abs/2406.01574",
305
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
306
+ "dataset_total_questions": "12000",
307
+ "prompt_style": "5-shot CoT"
308
+ }
309
+ },
310
+ "metric_config": {
311
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
312
+ "metric_id": "mmlu_pro/health",
313
+ "metric_name": "MMLU-Pro (Health)",
314
+ "metric_kind": "accuracy",
315
+ "metric_unit": "proportion",
316
+ "lower_is_better": false,
317
+ "score_type": "continuous",
318
+ "min_score": 0.0,
319
+ "max_score": 1.0,
320
+ "additional_details": {
321
+ "aggregation": "accuracy_over_subset",
322
+ "prompt_style": "5-shot CoT"
323
+ }
324
+ },
325
+ "score_details": {
326
+ "score": 0.7274
327
+ }
328
+ },
329
+ {
330
+ "evaluation_result_id": "mmlu_pro/history",
331
+ "evaluation_name": "MMLU-Pro (History)",
332
+ "source_data": {
333
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
334
+ "source_type": "hf_dataset",
335
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
336
+ "hf_split": "train",
337
+ "additional_details": {
338
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
339
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
340
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
341
+ "paper_url": "https://arxiv.org/abs/2406.01574",
342
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
343
+ "dataset_total_questions": "12000",
344
+ "prompt_style": "5-shot CoT"
345
+ }
346
+ },
347
+ "metric_config": {
348
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
349
+ "metric_id": "mmlu_pro/history",
350
+ "metric_name": "MMLU-Pro (History)",
351
+ "metric_kind": "accuracy",
352
+ "metric_unit": "proportion",
353
+ "lower_is_better": false,
354
+ "score_type": "continuous",
355
+ "min_score": 0.0,
356
+ "max_score": 1.0,
357
+ "additional_details": {
358
+ "aggregation": "accuracy_over_subset",
359
+ "prompt_style": "5-shot CoT"
360
+ }
361
+ },
362
+ "score_details": {
363
+ "score": 0.6562
364
+ }
365
+ },
366
+ {
367
+ "evaluation_result_id": "mmlu_pro/law",
368
+ "evaluation_name": "MMLU-Pro (Law)",
369
+ "source_data": {
370
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
371
+ "source_type": "hf_dataset",
372
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
373
+ "hf_split": "train",
374
+ "additional_details": {
375
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
376
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
377
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
378
+ "paper_url": "https://arxiv.org/abs/2406.01574",
379
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
380
+ "dataset_total_questions": "12000",
381
+ "prompt_style": "5-shot CoT"
382
+ }
383
+ },
384
+ "metric_config": {
385
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
386
+ "metric_id": "mmlu_pro/law",
387
+ "metric_name": "MMLU-Pro (Law)",
388
+ "metric_kind": "accuracy",
389
+ "metric_unit": "proportion",
390
+ "lower_is_better": false,
391
+ "score_type": "continuous",
392
+ "min_score": 0.0,
393
+ "max_score": 1.0,
394
+ "additional_details": {
395
+ "aggregation": "accuracy_over_subset",
396
+ "prompt_style": "5-shot CoT"
397
+ }
398
+ },
399
+ "score_details": {
400
+ "score": 0.5077
401
+ }
402
+ },
403
+ {
404
+ "evaluation_result_id": "mmlu_pro/math",
405
+ "evaluation_name": "MMLU-Pro (Math)",
406
+ "source_data": {
407
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
408
+ "source_type": "hf_dataset",
409
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
410
+ "hf_split": "train",
411
+ "additional_details": {
412
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
413
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
414
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
415
+ "paper_url": "https://arxiv.org/abs/2406.01574",
416
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
417
+ "dataset_total_questions": "12000",
418
+ "prompt_style": "5-shot CoT"
419
+ }
420
+ },
421
+ "metric_config": {
422
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
423
+ "metric_id": "mmlu_pro/math",
424
+ "metric_name": "MMLU-Pro (Math)",
425
+ "metric_kind": "accuracy",
426
+ "metric_unit": "proportion",
427
+ "lower_is_better": false,
428
+ "score_type": "continuous",
429
+ "min_score": 0.0,
430
+ "max_score": 1.0,
431
+ "additional_details": {
432
+ "aggregation": "accuracy_over_subset",
433
+ "prompt_style": "5-shot CoT"
434
+ }
435
+ },
436
+ "score_details": {
437
+ "score": 0.7276
438
+ }
439
+ },
440
+ {
441
+ "evaluation_result_id": "mmlu_pro/philosophy",
442
+ "evaluation_name": "MMLU-Pro (Philosophy)",
443
+ "source_data": {
444
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
445
+ "source_type": "hf_dataset",
446
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
447
+ "hf_split": "train",
448
+ "additional_details": {
449
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
450
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
451
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
452
+ "paper_url": "https://arxiv.org/abs/2406.01574",
453
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
454
+ "dataset_total_questions": "12000",
455
+ "prompt_style": "5-shot CoT"
456
+ }
457
+ },
458
+ "metric_config": {
459
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
460
+ "metric_id": "mmlu_pro/philosophy",
461
+ "metric_name": "MMLU-Pro (Philosophy)",
462
+ "metric_kind": "accuracy",
463
+ "metric_unit": "proportion",
464
+ "lower_is_better": false,
465
+ "score_type": "continuous",
466
+ "min_score": 0.0,
467
+ "max_score": 1.0,
468
+ "additional_details": {
469
+ "aggregation": "accuracy_over_subset",
470
+ "prompt_style": "5-shot CoT"
471
+ }
472
+ },
473
+ "score_details": {
474
+ "score": 0.6172
475
+ }
476
+ },
477
+ {
478
+ "evaluation_result_id": "mmlu_pro/physics",
479
+ "evaluation_name": "MMLU-Pro (Physics)",
480
+ "source_data": {
481
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
482
+ "source_type": "hf_dataset",
483
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
484
+ "hf_split": "train",
485
+ "additional_details": {
486
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
487
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
488
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
489
+ "paper_url": "https://arxiv.org/abs/2406.01574",
490
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
491
+ "dataset_total_questions": "12000",
492
+ "prompt_style": "5-shot CoT"
493
+ }
494
+ },
495
+ "metric_config": {
496
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
497
+ "metric_id": "mmlu_pro/physics",
498
+ "metric_name": "MMLU-Pro (Physics)",
499
+ "metric_kind": "accuracy",
500
+ "metric_unit": "proportion",
501
+ "lower_is_better": false,
502
+ "score_type": "continuous",
503
+ "min_score": 0.0,
504
+ "max_score": 1.0,
505
+ "additional_details": {
506
+ "aggregation": "accuracy_over_subset",
507
+ "prompt_style": "5-shot CoT"
508
+ }
509
+ },
510
+ "score_details": {
511
+ "score": 0.7036
512
+ }
513
+ },
514
+ {
515
+ "evaluation_result_id": "mmlu_pro/psychology",
516
+ "evaluation_name": "MMLU-Pro (Psychology)",
517
+ "source_data": {
518
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
519
+ "source_type": "hf_dataset",
520
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
521
+ "hf_split": "train",
522
+ "additional_details": {
523
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
524
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
525
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
526
+ "paper_url": "https://arxiv.org/abs/2406.01574",
527
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
528
+ "dataset_total_questions": "12000",
529
+ "prompt_style": "5-shot CoT"
530
+ }
531
+ },
532
+ "metric_config": {
533
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
534
+ "metric_id": "mmlu_pro/psychology",
535
+ "metric_name": "MMLU-Pro (Psychology)",
536
+ "metric_kind": "accuracy",
537
+ "metric_unit": "proportion",
538
+ "lower_is_better": false,
539
+ "score_type": "continuous",
540
+ "min_score": 0.0,
541
+ "max_score": 1.0,
542
+ "additional_details": {
543
+ "aggregation": "accuracy_over_subset",
544
+ "prompt_style": "5-shot CoT"
545
+ }
546
+ },
547
+ "score_details": {
548
+ "score": 0.772
549
+ }
550
+ },
551
+ {
552
+ "evaluation_result_id": "mmlu_pro/other",
553
+ "evaluation_name": "MMLU-Pro (Other)",
554
+ "source_data": {
555
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
556
+ "source_type": "hf_dataset",
557
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
558
+ "hf_split": "train",
559
+ "additional_details": {
560
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
561
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
562
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
563
+ "paper_url": "https://arxiv.org/abs/2406.01574",
564
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
565
+ "dataset_total_questions": "12000",
566
+ "prompt_style": "5-shot CoT"
567
+ }
568
+ },
569
+ "metric_config": {
570
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
571
+ "metric_id": "mmlu_pro/other",
572
+ "metric_name": "MMLU-Pro (Other)",
573
+ "metric_kind": "accuracy",
574
+ "metric_unit": "proportion",
575
+ "lower_is_better": false,
576
+ "score_type": "continuous",
577
+ "min_score": 0.0,
578
+ "max_score": 1.0,
579
+ "additional_details": {
580
+ "aggregation": "accuracy_over_subset",
581
+ "prompt_style": "5-shot CoT"
582
+ }
583
+ },
584
+ "score_details": {
585
+ "score": 0.7251
586
+ }
587
+ }
588
+ ]
589
+ }
data/mmlu_pro/google/gemini-2.0-flash-exp/b70dfa85-e4ee-4eaa-9d5b-7f58a6ac2413.json ADDED
@@ -0,0 +1,589 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-2.0-flash-exp/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-2.0-Flash-exp",
25
+ "id": "google/gemini-2.0-flash-exp",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-2.0-Flash-exp",
29
+ "leaderboard_data_source": "TIGER-Lab"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.7624
68
+ }
69
+ },
70
+ {
71
+ "evaluation_result_id": "mmlu_pro/biology",
72
+ "evaluation_name": "MMLU-Pro (Biology)",
73
+ "source_data": {
74
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
75
+ "source_type": "hf_dataset",
76
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
77
+ "hf_split": "train",
78
+ "additional_details": {
79
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
80
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
81
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
82
+ "paper_url": "https://arxiv.org/abs/2406.01574",
83
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
84
+ "dataset_total_questions": "12000",
85
+ "prompt_style": "5-shot CoT"
86
+ }
87
+ },
88
+ "metric_config": {
89
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
90
+ "metric_id": "mmlu_pro/biology",
91
+ "metric_name": "MMLU-Pro (Biology)",
92
+ "metric_kind": "accuracy",
93
+ "metric_unit": "proportion",
94
+ "lower_is_better": false,
95
+ "score_type": "continuous",
96
+ "min_score": 0.0,
97
+ "max_score": 1.0,
98
+ "additional_details": {
99
+ "aggregation": "accuracy_over_subset",
100
+ "prompt_style": "5-shot CoT"
101
+ }
102
+ },
103
+ "score_details": {
104
+ "score": 0.8836
105
+ }
106
+ },
107
+ {
108
+ "evaluation_result_id": "mmlu_pro/business",
109
+ "evaluation_name": "MMLU-Pro (Business)",
110
+ "source_data": {
111
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
112
+ "source_type": "hf_dataset",
113
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
114
+ "hf_split": "train",
115
+ "additional_details": {
116
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
117
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
118
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
119
+ "paper_url": "https://arxiv.org/abs/2406.01574",
120
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
121
+ "dataset_total_questions": "12000",
122
+ "prompt_style": "5-shot CoT"
123
+ }
124
+ },
125
+ "metric_config": {
126
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
127
+ "metric_id": "mmlu_pro/business",
128
+ "metric_name": "MMLU-Pro (Business)",
129
+ "metric_kind": "accuracy",
130
+ "metric_unit": "proportion",
131
+ "lower_is_better": false,
132
+ "score_type": "continuous",
133
+ "min_score": 0.0,
134
+ "max_score": 1.0,
135
+ "additional_details": {
136
+ "aggregation": "accuracy_over_subset",
137
+ "prompt_style": "5-shot CoT"
138
+ }
139
+ },
140
+ "score_details": {
141
+ "score": 0.7985
142
+ }
143
+ },
144
+ {
145
+ "evaluation_result_id": "mmlu_pro/chemistry",
146
+ "evaluation_name": "MMLU-Pro (Chemistry)",
147
+ "source_data": {
148
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
149
+ "source_type": "hf_dataset",
150
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
151
+ "hf_split": "train",
152
+ "additional_details": {
153
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
154
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
155
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
156
+ "paper_url": "https://arxiv.org/abs/2406.01574",
157
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
158
+ "dataset_total_questions": "12000",
159
+ "prompt_style": "5-shot CoT"
160
+ }
161
+ },
162
+ "metric_config": {
163
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
164
+ "metric_id": "mmlu_pro/chemistry",
165
+ "metric_name": "MMLU-Pro (Chemistry)",
166
+ "metric_kind": "accuracy",
167
+ "metric_unit": "proportion",
168
+ "lower_is_better": false,
169
+ "score_type": "continuous",
170
+ "min_score": 0.0,
171
+ "max_score": 1.0,
172
+ "additional_details": {
173
+ "aggregation": "accuracy_over_subset",
174
+ "prompt_style": "5-shot CoT"
175
+ }
176
+ },
177
+ "score_details": {
178
+ "score": 0.8004
179
+ }
180
+ },
181
+ {
182
+ "evaluation_result_id": "mmlu_pro/computer_science",
183
+ "evaluation_name": "MMLU-Pro (Computer Science)",
184
+ "source_data": {
185
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
186
+ "source_type": "hf_dataset",
187
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
188
+ "hf_split": "train",
189
+ "additional_details": {
190
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
191
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
192
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
193
+ "paper_url": "https://arxiv.org/abs/2406.01574",
194
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
195
+ "dataset_total_questions": "12000",
196
+ "prompt_style": "5-shot CoT"
197
+ }
198
+ },
199
+ "metric_config": {
200
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
201
+ "metric_id": "mmlu_pro/computer_science",
202
+ "metric_name": "MMLU-Pro (Computer Science)",
203
+ "metric_kind": "accuracy",
204
+ "metric_unit": "proportion",
205
+ "lower_is_better": false,
206
+ "score_type": "continuous",
207
+ "min_score": 0.0,
208
+ "max_score": 1.0,
209
+ "additional_details": {
210
+ "aggregation": "accuracy_over_subset",
211
+ "prompt_style": "5-shot CoT"
212
+ }
213
+ },
214
+ "score_details": {
215
+ "score": 0.799
216
+ }
217
+ },
218
+ {
219
+ "evaluation_result_id": "mmlu_pro/economics",
220
+ "evaluation_name": "MMLU-Pro (Economics)",
221
+ "source_data": {
222
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
223
+ "source_type": "hf_dataset",
224
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
225
+ "hf_split": "train",
226
+ "additional_details": {
227
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
228
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
229
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
230
+ "paper_url": "https://arxiv.org/abs/2406.01574",
231
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
232
+ "dataset_total_questions": "12000",
233
+ "prompt_style": "5-shot CoT"
234
+ }
235
+ },
236
+ "metric_config": {
237
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
238
+ "metric_id": "mmlu_pro/economics",
239
+ "metric_name": "MMLU-Pro (Economics)",
240
+ "metric_kind": "accuracy",
241
+ "metric_unit": "proportion",
242
+ "lower_is_better": false,
243
+ "score_type": "continuous",
244
+ "min_score": 0.0,
245
+ "max_score": 1.0,
246
+ "additional_details": {
247
+ "aggregation": "accuracy_over_subset",
248
+ "prompt_style": "5-shot CoT"
249
+ }
250
+ },
251
+ "score_details": {
252
+ "score": 0.8169
253
+ }
254
+ },
255
+ {
256
+ "evaluation_result_id": "mmlu_pro/engineering",
257
+ "evaluation_name": "MMLU-Pro (Engineering)",
258
+ "source_data": {
259
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
260
+ "source_type": "hf_dataset",
261
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
262
+ "hf_split": "train",
263
+ "additional_details": {
264
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
265
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
266
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
267
+ "paper_url": "https://arxiv.org/abs/2406.01574",
268
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
269
+ "dataset_total_questions": "12000",
270
+ "prompt_style": "5-shot CoT"
271
+ }
272
+ },
273
+ "metric_config": {
274
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
275
+ "metric_id": "mmlu_pro/engineering",
276
+ "metric_name": "MMLU-Pro (Engineering)",
277
+ "metric_kind": "accuracy",
278
+ "metric_unit": "proportion",
279
+ "lower_is_better": false,
280
+ "score_type": "continuous",
281
+ "min_score": 0.0,
282
+ "max_score": 1.0,
283
+ "additional_details": {
284
+ "aggregation": "accuracy_over_subset",
285
+ "prompt_style": "5-shot CoT"
286
+ }
287
+ },
288
+ "score_details": {
289
+ "score": 0.6155
290
+ }
291
+ },
292
+ {
293
+ "evaluation_result_id": "mmlu_pro/health",
294
+ "evaluation_name": "MMLU-Pro (Health)",
295
+ "source_data": {
296
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
297
+ "source_type": "hf_dataset",
298
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
299
+ "hf_split": "train",
300
+ "additional_details": {
301
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
302
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
303
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
304
+ "paper_url": "https://arxiv.org/abs/2406.01574",
305
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
306
+ "dataset_total_questions": "12000",
307
+ "prompt_style": "5-shot CoT"
308
+ }
309
+ },
310
+ "metric_config": {
311
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
312
+ "metric_id": "mmlu_pro/health",
313
+ "metric_name": "MMLU-Pro (Health)",
314
+ "metric_kind": "accuracy",
315
+ "metric_unit": "proportion",
316
+ "lower_is_better": false,
317
+ "score_type": "continuous",
318
+ "min_score": 0.0,
319
+ "max_score": 1.0,
320
+ "additional_details": {
321
+ "aggregation": "accuracy_over_subset",
322
+ "prompt_style": "5-shot CoT"
323
+ }
324
+ },
325
+ "score_details": {
326
+ "score": 0.7442
327
+ }
328
+ },
329
+ {
330
+ "evaluation_result_id": "mmlu_pro/history",
331
+ "evaluation_name": "MMLU-Pro (History)",
332
+ "source_data": {
333
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
334
+ "source_type": "hf_dataset",
335
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
336
+ "hf_split": "train",
337
+ "additional_details": {
338
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
339
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
340
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
341
+ "paper_url": "https://arxiv.org/abs/2406.01574",
342
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
343
+ "dataset_total_questions": "12000",
344
+ "prompt_style": "5-shot CoT"
345
+ }
346
+ },
347
+ "metric_config": {
348
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
349
+ "metric_id": "mmlu_pro/history",
350
+ "metric_name": "MMLU-Pro (History)",
351
+ "metric_kind": "accuracy",
352
+ "metric_unit": "proportion",
353
+ "lower_is_better": false,
354
+ "score_type": "continuous",
355
+ "min_score": 0.0,
356
+ "max_score": 1.0,
357
+ "additional_details": {
358
+ "aggregation": "accuracy_over_subset",
359
+ "prompt_style": "5-shot CoT"
360
+ }
361
+ },
362
+ "score_details": {
363
+ "score": 0.7008
364
+ }
365
+ },
366
+ {
367
+ "evaluation_result_id": "mmlu_pro/law",
368
+ "evaluation_name": "MMLU-Pro (Law)",
369
+ "source_data": {
370
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
371
+ "source_type": "hf_dataset",
372
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
373
+ "hf_split": "train",
374
+ "additional_details": {
375
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
376
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
377
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
378
+ "paper_url": "https://arxiv.org/abs/2406.01574",
379
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
380
+ "dataset_total_questions": "12000",
381
+ "prompt_style": "5-shot CoT"
382
+ }
383
+ },
384
+ "metric_config": {
385
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
386
+ "metric_id": "mmlu_pro/law",
387
+ "metric_name": "MMLU-Pro (Law)",
388
+ "metric_kind": "accuracy",
389
+ "metric_unit": "proportion",
390
+ "lower_is_better": false,
391
+ "score_type": "continuous",
392
+ "min_score": 0.0,
393
+ "max_score": 1.0,
394
+ "additional_details": {
395
+ "aggregation": "accuracy_over_subset",
396
+ "prompt_style": "5-shot CoT"
397
+ }
398
+ },
399
+ "score_details": {
400
+ "score": 0.5647
401
+ }
402
+ },
403
+ {
404
+ "evaluation_result_id": "mmlu_pro/math",
405
+ "evaluation_name": "MMLU-Pro (Math)",
406
+ "source_data": {
407
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
408
+ "source_type": "hf_dataset",
409
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
410
+ "hf_split": "train",
411
+ "additional_details": {
412
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
413
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
414
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
415
+ "paper_url": "https://arxiv.org/abs/2406.01574",
416
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
417
+ "dataset_total_questions": "12000",
418
+ "prompt_style": "5-shot CoT"
419
+ }
420
+ },
421
+ "metric_config": {
422
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
423
+ "metric_id": "mmlu_pro/math",
424
+ "metric_name": "MMLU-Pro (Math)",
425
+ "metric_kind": "accuracy",
426
+ "metric_unit": "proportion",
427
+ "lower_is_better": false,
428
+ "score_type": "continuous",
429
+ "min_score": 0.0,
430
+ "max_score": 1.0,
431
+ "additional_details": {
432
+ "aggregation": "accuracy_over_subset",
433
+ "prompt_style": "5-shot CoT"
434
+ }
435
+ },
436
+ "score_details": {
437
+ "score": 0.8638
438
+ }
439
+ },
440
+ {
441
+ "evaluation_result_id": "mmlu_pro/philosophy",
442
+ "evaluation_name": "MMLU-Pro (Philosophy)",
443
+ "source_data": {
444
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
445
+ "source_type": "hf_dataset",
446
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
447
+ "hf_split": "train",
448
+ "additional_details": {
449
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
450
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
451
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
452
+ "paper_url": "https://arxiv.org/abs/2406.01574",
453
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
454
+ "dataset_total_questions": "12000",
455
+ "prompt_style": "5-shot CoT"
456
+ }
457
+ },
458
+ "metric_config": {
459
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
460
+ "metric_id": "mmlu_pro/philosophy",
461
+ "metric_name": "MMLU-Pro (Philosophy)",
462
+ "metric_kind": "accuracy",
463
+ "metric_unit": "proportion",
464
+ "lower_is_better": false,
465
+ "score_type": "continuous",
466
+ "min_score": 0.0,
467
+ "max_score": 1.0,
468
+ "additional_details": {
469
+ "aggregation": "accuracy_over_subset",
470
+ "prompt_style": "5-shot CoT"
471
+ }
472
+ },
473
+ "score_details": {
474
+ "score": 0.6994
475
+ }
476
+ },
477
+ {
478
+ "evaluation_result_id": "mmlu_pro/physics",
479
+ "evaluation_name": "MMLU-Pro (Physics)",
480
+ "source_data": {
481
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
482
+ "source_type": "hf_dataset",
483
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
484
+ "hf_split": "train",
485
+ "additional_details": {
486
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
487
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
488
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
489
+ "paper_url": "https://arxiv.org/abs/2406.01574",
490
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
491
+ "dataset_total_questions": "12000",
492
+ "prompt_style": "5-shot CoT"
493
+ }
494
+ },
495
+ "metric_config": {
496
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
497
+ "metric_id": "mmlu_pro/physics",
498
+ "metric_name": "MMLU-Pro (Physics)",
499
+ "metric_kind": "accuracy",
500
+ "metric_unit": "proportion",
501
+ "lower_is_better": false,
502
+ "score_type": "continuous",
503
+ "min_score": 0.0,
504
+ "max_score": 1.0,
505
+ "additional_details": {
506
+ "aggregation": "accuracy_over_subset",
507
+ "prompt_style": "5-shot CoT"
508
+ }
509
+ },
510
+ "score_details": {
511
+ "score": 0.8127
512
+ }
513
+ },
514
+ {
515
+ "evaluation_result_id": "mmlu_pro/psychology",
516
+ "evaluation_name": "MMLU-Pro (Psychology)",
517
+ "source_data": {
518
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
519
+ "source_type": "hf_dataset",
520
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
521
+ "hf_split": "train",
522
+ "additional_details": {
523
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
524
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
525
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
526
+ "paper_url": "https://arxiv.org/abs/2406.01574",
527
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
528
+ "dataset_total_questions": "12000",
529
+ "prompt_style": "5-shot CoT"
530
+ }
531
+ },
532
+ "metric_config": {
533
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
534
+ "metric_id": "mmlu_pro/psychology",
535
+ "metric_name": "MMLU-Pro (Psychology)",
536
+ "metric_kind": "accuracy",
537
+ "metric_unit": "proportion",
538
+ "lower_is_better": false,
539
+ "score_type": "continuous",
540
+ "min_score": 0.0,
541
+ "max_score": 1.0,
542
+ "additional_details": {
543
+ "aggregation": "accuracy_over_subset",
544
+ "prompt_style": "5-shot CoT"
545
+ }
546
+ },
547
+ "score_details": {
548
+ "score": 0.7905
549
+ }
550
+ },
551
+ {
552
+ "evaluation_result_id": "mmlu_pro/other",
553
+ "evaluation_name": "MMLU-Pro (Other)",
554
+ "source_data": {
555
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
556
+ "source_type": "hf_dataset",
557
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
558
+ "hf_split": "train",
559
+ "additional_details": {
560
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
561
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
562
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
563
+ "paper_url": "https://arxiv.org/abs/2406.01574",
564
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
565
+ "dataset_total_questions": "12000",
566
+ "prompt_style": "5-shot CoT"
567
+ }
568
+ },
569
+ "metric_config": {
570
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
571
+ "metric_id": "mmlu_pro/other",
572
+ "metric_name": "MMLU-Pro (Other)",
573
+ "metric_kind": "accuracy",
574
+ "metric_unit": "proportion",
575
+ "lower_is_better": false,
576
+ "score_type": "continuous",
577
+ "min_score": 0.0,
578
+ "max_score": 1.0,
579
+ "additional_details": {
580
+ "aggregation": "accuracy_over_subset",
581
+ "prompt_style": "5-shot CoT"
582
+ }
583
+ },
584
+ "score_details": {
585
+ "score": 0.7476
586
+ }
587
+ }
588
+ ]
589
+ }
data/mmlu_pro/google/gemini-2.0-flash-lite/841ad2b0-86e3-44e4-88c2-d010d6c9c52e.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-2.0-flash-lite/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-2.0-Flash-Lite",
25
+ "id": "google/gemini-2.0-flash-lite",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-2.0-Flash-Lite",
29
+ "leaderboard_data_source": "Self-Reported"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.716
68
+ }
69
+ }
70
+ ]
71
+ }
data/mmlu_pro/google/gemini-2.0-flash/1961f28e-d519-422d-8221-fd482b2f60f5.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-2.0-flash/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-2.0-Flash",
25
+ "id": "google/gemini-2.0-flash",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-2.0-Flash",
29
+ "leaderboard_data_source": "Self-Reported"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.776
68
+ }
69
+ }
70
+ ]
71
+ }
data/mmlu_pro/google/gemini-2.0-pro/34eecaad-e501-4b2b-9e9c-284804d1a581.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-2.0-pro/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-2.0-Pro",
25
+ "id": "google/gemini-2.0-pro",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-2.0-Pro",
29
+ "leaderboard_data_source": "Self-Reported"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.791
68
+ }
69
+ }
70
+ ]
71
+ }
data/mmlu_pro/google/gemini-2.5-pro-exp-03-25/976b7201-8e01-49f0-b9ab-884d61e16288.json ADDED
@@ -0,0 +1,589 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-2.5-pro-exp-03-25/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-2.5-Pro-Exp-03-25",
25
+ "id": "google/gemini-2.5-pro-exp-03-25",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-2.5-Pro-Exp-03-25",
29
+ "leaderboard_data_source": "TIGER-Lab"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.8452
68
+ }
69
+ },
70
+ {
71
+ "evaluation_result_id": "mmlu_pro/biology",
72
+ "evaluation_name": "MMLU-Pro (Biology)",
73
+ "source_data": {
74
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
75
+ "source_type": "hf_dataset",
76
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
77
+ "hf_split": "train",
78
+ "additional_details": {
79
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
80
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
81
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
82
+ "paper_url": "https://arxiv.org/abs/2406.01574",
83
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
84
+ "dataset_total_questions": "12000",
85
+ "prompt_style": "5-shot CoT"
86
+ }
87
+ },
88
+ "metric_config": {
89
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
90
+ "metric_id": "mmlu_pro/biology",
91
+ "metric_name": "MMLU-Pro (Biology)",
92
+ "metric_kind": "accuracy",
93
+ "metric_unit": "proportion",
94
+ "lower_is_better": false,
95
+ "score_type": "continuous",
96
+ "min_score": 0.0,
97
+ "max_score": 1.0,
98
+ "additional_details": {
99
+ "aggregation": "accuracy_over_subset",
100
+ "prompt_style": "5-shot CoT"
101
+ }
102
+ },
103
+ "score_details": {
104
+ "score": 0.9299
105
+ }
106
+ },
107
+ {
108
+ "evaluation_result_id": "mmlu_pro/business",
109
+ "evaluation_name": "MMLU-Pro (Business)",
110
+ "source_data": {
111
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
112
+ "source_type": "hf_dataset",
113
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
114
+ "hf_split": "train",
115
+ "additional_details": {
116
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
117
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
118
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
119
+ "paper_url": "https://arxiv.org/abs/2406.01574",
120
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
121
+ "dataset_total_questions": "12000",
122
+ "prompt_style": "5-shot CoT"
123
+ }
124
+ },
125
+ "metric_config": {
126
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
127
+ "metric_id": "mmlu_pro/business",
128
+ "metric_name": "MMLU-Pro (Business)",
129
+ "metric_kind": "accuracy",
130
+ "metric_unit": "proportion",
131
+ "lower_is_better": false,
132
+ "score_type": "continuous",
133
+ "min_score": 0.0,
134
+ "max_score": 1.0,
135
+ "additional_details": {
136
+ "aggregation": "accuracy_over_subset",
137
+ "prompt_style": "5-shot CoT"
138
+ }
139
+ },
140
+ "score_details": {
141
+ "score": 0.893
142
+ }
143
+ },
144
+ {
145
+ "evaluation_result_id": "mmlu_pro/chemistry",
146
+ "evaluation_name": "MMLU-Pro (Chemistry)",
147
+ "source_data": {
148
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
149
+ "source_type": "hf_dataset",
150
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
151
+ "hf_split": "train",
152
+ "additional_details": {
153
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
154
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
155
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
156
+ "paper_url": "https://arxiv.org/abs/2406.01574",
157
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
158
+ "dataset_total_questions": "12000",
159
+ "prompt_style": "5-shot CoT"
160
+ }
161
+ },
162
+ "metric_config": {
163
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
164
+ "metric_id": "mmlu_pro/chemistry",
165
+ "metric_name": "MMLU-Pro (Chemistry)",
166
+ "metric_kind": "accuracy",
167
+ "metric_unit": "proportion",
168
+ "lower_is_better": false,
169
+ "score_type": "continuous",
170
+ "min_score": 0.0,
171
+ "max_score": 1.0,
172
+ "additional_details": {
173
+ "aggregation": "accuracy_over_subset",
174
+ "prompt_style": "5-shot CoT"
175
+ }
176
+ },
177
+ "score_details": {
178
+ "score": 0.8741
179
+ }
180
+ },
181
+ {
182
+ "evaluation_result_id": "mmlu_pro/computer_science",
183
+ "evaluation_name": "MMLU-Pro (Computer Science)",
184
+ "source_data": {
185
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
186
+ "source_type": "hf_dataset",
187
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
188
+ "hf_split": "train",
189
+ "additional_details": {
190
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
191
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
192
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
193
+ "paper_url": "https://arxiv.org/abs/2406.01574",
194
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
195
+ "dataset_total_questions": "12000",
196
+ "prompt_style": "5-shot CoT"
197
+ }
198
+ },
199
+ "metric_config": {
200
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
201
+ "metric_id": "mmlu_pro/computer_science",
202
+ "metric_name": "MMLU-Pro (Computer Science)",
203
+ "metric_kind": "accuracy",
204
+ "metric_unit": "proportion",
205
+ "lower_is_better": false,
206
+ "score_type": "continuous",
207
+ "min_score": 0.0,
208
+ "max_score": 1.0,
209
+ "additional_details": {
210
+ "aggregation": "accuracy_over_subset",
211
+ "prompt_style": "5-shot CoT"
212
+ }
213
+ },
214
+ "score_details": {
215
+ "score": 0.8556
216
+ }
217
+ },
218
+ {
219
+ "evaluation_result_id": "mmlu_pro/economics",
220
+ "evaluation_name": "MMLU-Pro (Economics)",
221
+ "source_data": {
222
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
223
+ "source_type": "hf_dataset",
224
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
225
+ "hf_split": "train",
226
+ "additional_details": {
227
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
228
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
229
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
230
+ "paper_url": "https://arxiv.org/abs/2406.01574",
231
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
232
+ "dataset_total_questions": "12000",
233
+ "prompt_style": "5-shot CoT"
234
+ }
235
+ },
236
+ "metric_config": {
237
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
238
+ "metric_id": "mmlu_pro/economics",
239
+ "metric_name": "MMLU-Pro (Economics)",
240
+ "metric_kind": "accuracy",
241
+ "metric_unit": "proportion",
242
+ "lower_is_better": false,
243
+ "score_type": "continuous",
244
+ "min_score": 0.0,
245
+ "max_score": 1.0,
246
+ "additional_details": {
247
+ "aggregation": "accuracy_over_subset",
248
+ "prompt_style": "5-shot CoT"
249
+ }
250
+ },
251
+ "score_details": {
252
+ "score": 0.8829
253
+ }
254
+ },
255
+ {
256
+ "evaluation_result_id": "mmlu_pro/engineering",
257
+ "evaluation_name": "MMLU-Pro (Engineering)",
258
+ "source_data": {
259
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
260
+ "source_type": "hf_dataset",
261
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
262
+ "hf_split": "train",
263
+ "additional_details": {
264
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
265
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
266
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
267
+ "paper_url": "https://arxiv.org/abs/2406.01574",
268
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
269
+ "dataset_total_questions": "12000",
270
+ "prompt_style": "5-shot CoT"
271
+ }
272
+ },
273
+ "metric_config": {
274
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
275
+ "metric_id": "mmlu_pro/engineering",
276
+ "metric_name": "MMLU-Pro (Engineering)",
277
+ "metric_kind": "accuracy",
278
+ "metric_unit": "proportion",
279
+ "lower_is_better": false,
280
+ "score_type": "continuous",
281
+ "min_score": 0.0,
282
+ "max_score": 1.0,
283
+ "additional_details": {
284
+ "aggregation": "accuracy_over_subset",
285
+ "prompt_style": "5-shot CoT"
286
+ }
287
+ },
288
+ "score_details": {
289
+ "score": 0.7358
290
+ }
291
+ },
292
+ {
293
+ "evaluation_result_id": "mmlu_pro/health",
294
+ "evaluation_name": "MMLU-Pro (Health)",
295
+ "source_data": {
296
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
297
+ "source_type": "hf_dataset",
298
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
299
+ "hf_split": "train",
300
+ "additional_details": {
301
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
302
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
303
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
304
+ "paper_url": "https://arxiv.org/abs/2406.01574",
305
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
306
+ "dataset_total_questions": "12000",
307
+ "prompt_style": "5-shot CoT"
308
+ }
309
+ },
310
+ "metric_config": {
311
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
312
+ "metric_id": "mmlu_pro/health",
313
+ "metric_name": "MMLU-Pro (Health)",
314
+ "metric_kind": "accuracy",
315
+ "metric_unit": "proportion",
316
+ "lower_is_better": false,
317
+ "score_type": "continuous",
318
+ "min_score": 0.0,
319
+ "max_score": 1.0,
320
+ "additional_details": {
321
+ "aggregation": "accuracy_over_subset",
322
+ "prompt_style": "5-shot CoT"
323
+ }
324
+ },
325
+ "score_details": {
326
+ "score": 0.817
327
+ }
328
+ },
329
+ {
330
+ "evaluation_result_id": "mmlu_pro/history",
331
+ "evaluation_name": "MMLU-Pro (History)",
332
+ "source_data": {
333
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
334
+ "source_type": "hf_dataset",
335
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
336
+ "hf_split": "train",
337
+ "additional_details": {
338
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
339
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
340
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
341
+ "paper_url": "https://arxiv.org/abs/2406.01574",
342
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
343
+ "dataset_total_questions": "12000",
344
+ "prompt_style": "5-shot CoT"
345
+ }
346
+ },
347
+ "metric_config": {
348
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
349
+ "metric_id": "mmlu_pro/history",
350
+ "metric_name": "MMLU-Pro (History)",
351
+ "metric_kind": "accuracy",
352
+ "metric_unit": "proportion",
353
+ "lower_is_better": false,
354
+ "score_type": "continuous",
355
+ "min_score": 0.0,
356
+ "max_score": 1.0,
357
+ "additional_details": {
358
+ "aggregation": "accuracy_over_subset",
359
+ "prompt_style": "5-shot CoT"
360
+ }
361
+ },
362
+ "score_details": {
363
+ "score": 0.7553
364
+ }
365
+ },
366
+ {
367
+ "evaluation_result_id": "mmlu_pro/law",
368
+ "evaluation_name": "MMLU-Pro (Law)",
369
+ "source_data": {
370
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
371
+ "source_type": "hf_dataset",
372
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
373
+ "hf_split": "train",
374
+ "additional_details": {
375
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
376
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
377
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
378
+ "paper_url": "https://arxiv.org/abs/2406.01574",
379
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
380
+ "dataset_total_questions": "12000",
381
+ "prompt_style": "5-shot CoT"
382
+ }
383
+ },
384
+ "metric_config": {
385
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
386
+ "metric_id": "mmlu_pro/law",
387
+ "metric_name": "MMLU-Pro (Law)",
388
+ "metric_kind": "accuracy",
389
+ "metric_unit": "proportion",
390
+ "lower_is_better": false,
391
+ "score_type": "continuous",
392
+ "min_score": 0.0,
393
+ "max_score": 1.0,
394
+ "additional_details": {
395
+ "aggregation": "accuracy_over_subset",
396
+ "prompt_style": "5-shot CoT"
397
+ }
398
+ },
399
+ "score_details": {
400
+ "score": 0.7244
401
+ }
402
+ },
403
+ {
404
+ "evaluation_result_id": "mmlu_pro/math",
405
+ "evaluation_name": "MMLU-Pro (Math)",
406
+ "source_data": {
407
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
408
+ "source_type": "hf_dataset",
409
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
410
+ "hf_split": "train",
411
+ "additional_details": {
412
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
413
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
414
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
415
+ "paper_url": "https://arxiv.org/abs/2406.01574",
416
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
417
+ "dataset_total_questions": "12000",
418
+ "prompt_style": "5-shot CoT"
419
+ }
420
+ },
421
+ "metric_config": {
422
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
423
+ "metric_id": "mmlu_pro/math",
424
+ "metric_name": "MMLU-Pro (Math)",
425
+ "metric_kind": "accuracy",
426
+ "metric_unit": "proportion",
427
+ "lower_is_better": false,
428
+ "score_type": "continuous",
429
+ "min_score": 0.0,
430
+ "max_score": 1.0,
431
+ "additional_details": {
432
+ "aggregation": "accuracy_over_subset",
433
+ "prompt_style": "5-shot CoT"
434
+ }
435
+ },
436
+ "score_details": {
437
+ "score": 0.8884
438
+ }
439
+ },
440
+ {
441
+ "evaluation_result_id": "mmlu_pro/philosophy",
442
+ "evaluation_name": "MMLU-Pro (Philosophy)",
443
+ "source_data": {
444
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
445
+ "source_type": "hf_dataset",
446
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
447
+ "hf_split": "train",
448
+ "additional_details": {
449
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
450
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
451
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
452
+ "paper_url": "https://arxiv.org/abs/2406.01574",
453
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
454
+ "dataset_total_questions": "12000",
455
+ "prompt_style": "5-shot CoT"
456
+ }
457
+ },
458
+ "metric_config": {
459
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
460
+ "metric_id": "mmlu_pro/philosophy",
461
+ "metric_name": "MMLU-Pro (Philosophy)",
462
+ "metric_kind": "accuracy",
463
+ "metric_unit": "proportion",
464
+ "lower_is_better": false,
465
+ "score_type": "continuous",
466
+ "min_score": 0.0,
467
+ "max_score": 1.0,
468
+ "additional_details": {
469
+ "aggregation": "accuracy_over_subset",
470
+ "prompt_style": "5-shot CoT"
471
+ }
472
+ },
473
+ "score_details": {
474
+ "score": 0.8327
475
+ }
476
+ },
477
+ {
478
+ "evaluation_result_id": "mmlu_pro/physics",
479
+ "evaluation_name": "MMLU-Pro (Physics)",
480
+ "source_data": {
481
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
482
+ "source_type": "hf_dataset",
483
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
484
+ "hf_split": "train",
485
+ "additional_details": {
486
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
487
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
488
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
489
+ "paper_url": "https://arxiv.org/abs/2406.01574",
490
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
491
+ "dataset_total_questions": "12000",
492
+ "prompt_style": "5-shot CoT"
493
+ }
494
+ },
495
+ "metric_config": {
496
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
497
+ "metric_id": "mmlu_pro/physics",
498
+ "metric_name": "MMLU-Pro (Physics)",
499
+ "metric_kind": "accuracy",
500
+ "metric_unit": "proportion",
501
+ "lower_is_better": false,
502
+ "score_type": "continuous",
503
+ "min_score": 0.0,
504
+ "max_score": 1.0,
505
+ "additional_details": {
506
+ "aggregation": "accuracy_over_subset",
507
+ "prompt_style": "5-shot CoT"
508
+ }
509
+ },
510
+ "score_details": {
511
+ "score": 0.8831
512
+ }
513
+ },
514
+ {
515
+ "evaluation_result_id": "mmlu_pro/psychology",
516
+ "evaluation_name": "MMLU-Pro (Psychology)",
517
+ "source_data": {
518
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
519
+ "source_type": "hf_dataset",
520
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
521
+ "hf_split": "train",
522
+ "additional_details": {
523
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
524
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
525
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
526
+ "paper_url": "https://arxiv.org/abs/2406.01574",
527
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
528
+ "dataset_total_questions": "12000",
529
+ "prompt_style": "5-shot CoT"
530
+ }
531
+ },
532
+ "metric_config": {
533
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
534
+ "metric_id": "mmlu_pro/psychology",
535
+ "metric_name": "MMLU-Pro (Psychology)",
536
+ "metric_kind": "accuracy",
537
+ "metric_unit": "proportion",
538
+ "lower_is_better": false,
539
+ "score_type": "continuous",
540
+ "min_score": 0.0,
541
+ "max_score": 1.0,
542
+ "additional_details": {
543
+ "aggregation": "accuracy_over_subset",
544
+ "prompt_style": "5-shot CoT"
545
+ }
546
+ },
547
+ "score_details": {
548
+ "score": 0.8726
549
+ }
550
+ },
551
+ {
552
+ "evaluation_result_id": "mmlu_pro/other",
553
+ "evaluation_name": "MMLU-Pro (Other)",
554
+ "source_data": {
555
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
556
+ "source_type": "hf_dataset",
557
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
558
+ "hf_split": "train",
559
+ "additional_details": {
560
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
561
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
562
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
563
+ "paper_url": "https://arxiv.org/abs/2406.01574",
564
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
565
+ "dataset_total_questions": "12000",
566
+ "prompt_style": "5-shot CoT"
567
+ }
568
+ },
569
+ "metric_config": {
570
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
571
+ "metric_id": "mmlu_pro/other",
572
+ "metric_name": "MMLU-Pro (Other)",
573
+ "metric_kind": "accuracy",
574
+ "metric_unit": "proportion",
575
+ "lower_is_better": false,
576
+ "score_type": "continuous",
577
+ "min_score": 0.0,
578
+ "max_score": 1.0,
579
+ "additional_details": {
580
+ "aggregation": "accuracy_over_subset",
581
+ "prompt_style": "5-shot CoT"
582
+ }
583
+ },
584
+ "score_details": {
585
+ "score": 0.8295
586
+ }
587
+ }
588
+ ]
589
+ }
data/mmlu_pro/google/gemini-2.5-pro/4ac5ed20-3428-462d-b108-9d4bfb2fed2b.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-2.5-pro/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-2.5-Pro",
25
+ "id": "google/gemini-2.5-pro",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-2.5-Pro",
29
+ "leaderboard_data_source": "Self-Reported"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.86
68
+ }
69
+ }
70
+ ]
71
+ }
data/mmlu_pro/google/gemini-3.1-flash-lite-preview/9e560681-ebd4-4d96-92fa-ef8e8e6c64fe.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-3.1-flash-lite-preview/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-3.1-Flash-Lite-Preview",
25
+ "id": "google/gemini-3.1-flash-lite-preview",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-3.1-Flash-Lite-Preview",
29
+ "leaderboard_data_source": "Self-Reported"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.862
68
+ }
69
+ }
70
+ ]
71
+ }
data/mmlu_pro/google/gemini-3.1-pro/c03b9f04-fa5b-41e8-b58f-0bfe85be3c0f.json ADDED
@@ -0,0 +1,589 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemini-3.1-pro/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemini-3.1-Pro",
25
+ "id": "google/gemini-3.1-pro",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemini-3.1-Pro",
29
+ "leaderboard_data_source": "TIGER-Lab"
30
+ }
31
+ },
32
+ "evaluation_results": [
33
+ {
34
+ "evaluation_result_id": "mmlu_pro/overall",
35
+ "evaluation_name": "MMLU-Pro (overall)",
36
+ "source_data": {
37
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
38
+ "source_type": "hf_dataset",
39
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
40
+ "hf_split": "train",
41
+ "additional_details": {
42
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
43
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
44
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
45
+ "paper_url": "https://arxiv.org/abs/2406.01574",
46
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
47
+ "dataset_total_questions": "12000",
48
+ "prompt_style": "5-shot CoT"
49
+ }
50
+ },
51
+ "metric_config": {
52
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
53
+ "metric_id": "mmlu_pro/overall",
54
+ "metric_name": "MMLU-Pro (overall)",
55
+ "metric_kind": "accuracy",
56
+ "metric_unit": "proportion",
57
+ "lower_is_better": false,
58
+ "score_type": "continuous",
59
+ "min_score": 0.0,
60
+ "max_score": 1.0,
61
+ "additional_details": {
62
+ "aggregation": "accuracy_over_subset",
63
+ "prompt_style": "5-shot CoT"
64
+ }
65
+ },
66
+ "score_details": {
67
+ "score": 0.9116
68
+ }
69
+ },
70
+ {
71
+ "evaluation_result_id": "mmlu_pro/biology",
72
+ "evaluation_name": "MMLU-Pro (Biology)",
73
+ "source_data": {
74
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
75
+ "source_type": "hf_dataset",
76
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
77
+ "hf_split": "train",
78
+ "additional_details": {
79
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
80
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
81
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
82
+ "paper_url": "https://arxiv.org/abs/2406.01574",
83
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
84
+ "dataset_total_questions": "12000",
85
+ "prompt_style": "5-shot CoT"
86
+ }
87
+ },
88
+ "metric_config": {
89
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
90
+ "metric_id": "mmlu_pro/biology",
91
+ "metric_name": "MMLU-Pro (Biology)",
92
+ "metric_kind": "accuracy",
93
+ "metric_unit": "proportion",
94
+ "lower_is_better": false,
95
+ "score_type": "continuous",
96
+ "min_score": 0.0,
97
+ "max_score": 1.0,
98
+ "additional_details": {
99
+ "aggregation": "accuracy_over_subset",
100
+ "prompt_style": "5-shot CoT"
101
+ }
102
+ },
103
+ "score_details": {
104
+ "score": 0.9582
105
+ }
106
+ },
107
+ {
108
+ "evaluation_result_id": "mmlu_pro/business",
109
+ "evaluation_name": "MMLU-Pro (Business)",
110
+ "source_data": {
111
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
112
+ "source_type": "hf_dataset",
113
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
114
+ "hf_split": "train",
115
+ "additional_details": {
116
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
117
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
118
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
119
+ "paper_url": "https://arxiv.org/abs/2406.01574",
120
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
121
+ "dataset_total_questions": "12000",
122
+ "prompt_style": "5-shot CoT"
123
+ }
124
+ },
125
+ "metric_config": {
126
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
127
+ "metric_id": "mmlu_pro/business",
128
+ "metric_name": "MMLU-Pro (Business)",
129
+ "metric_kind": "accuracy",
130
+ "metric_unit": "proportion",
131
+ "lower_is_better": false,
132
+ "score_type": "continuous",
133
+ "min_score": 0.0,
134
+ "max_score": 1.0,
135
+ "additional_details": {
136
+ "aggregation": "accuracy_over_subset",
137
+ "prompt_style": "5-shot CoT"
138
+ }
139
+ },
140
+ "score_details": {
141
+ "score": 0.9328
142
+ }
143
+ },
144
+ {
145
+ "evaluation_result_id": "mmlu_pro/chemistry",
146
+ "evaluation_name": "MMLU-Pro (Chemistry)",
147
+ "source_data": {
148
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
149
+ "source_type": "hf_dataset",
150
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
151
+ "hf_split": "train",
152
+ "additional_details": {
153
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
154
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
155
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
156
+ "paper_url": "https://arxiv.org/abs/2406.01574",
157
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
158
+ "dataset_total_questions": "12000",
159
+ "prompt_style": "5-shot CoT"
160
+ }
161
+ },
162
+ "metric_config": {
163
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
164
+ "metric_id": "mmlu_pro/chemistry",
165
+ "metric_name": "MMLU-Pro (Chemistry)",
166
+ "metric_kind": "accuracy",
167
+ "metric_unit": "proportion",
168
+ "lower_is_better": false,
169
+ "score_type": "continuous",
170
+ "min_score": 0.0,
171
+ "max_score": 1.0,
172
+ "additional_details": {
173
+ "aggregation": "accuracy_over_subset",
174
+ "prompt_style": "5-shot CoT"
175
+ }
176
+ },
177
+ "score_details": {
178
+ "score": 0.9214
179
+ }
180
+ },
181
+ {
182
+ "evaluation_result_id": "mmlu_pro/computer_science",
183
+ "evaluation_name": "MMLU-Pro (Computer Science)",
184
+ "source_data": {
185
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
186
+ "source_type": "hf_dataset",
187
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
188
+ "hf_split": "train",
189
+ "additional_details": {
190
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
191
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
192
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
193
+ "paper_url": "https://arxiv.org/abs/2406.01574",
194
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
195
+ "dataset_total_questions": "12000",
196
+ "prompt_style": "5-shot CoT"
197
+ }
198
+ },
199
+ "metric_config": {
200
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
201
+ "metric_id": "mmlu_pro/computer_science",
202
+ "metric_name": "MMLU-Pro (Computer Science)",
203
+ "metric_kind": "accuracy",
204
+ "metric_unit": "proportion",
205
+ "lower_is_better": false,
206
+ "score_type": "continuous",
207
+ "min_score": 0.0,
208
+ "max_score": 1.0,
209
+ "additional_details": {
210
+ "aggregation": "accuracy_over_subset",
211
+ "prompt_style": "5-shot CoT"
212
+ }
213
+ },
214
+ "score_details": {
215
+ "score": 0.9366
216
+ }
217
+ },
218
+ {
219
+ "evaluation_result_id": "mmlu_pro/economics",
220
+ "evaluation_name": "MMLU-Pro (Economics)",
221
+ "source_data": {
222
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
223
+ "source_type": "hf_dataset",
224
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
225
+ "hf_split": "train",
226
+ "additional_details": {
227
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
228
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
229
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
230
+ "paper_url": "https://arxiv.org/abs/2406.01574",
231
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
232
+ "dataset_total_questions": "12000",
233
+ "prompt_style": "5-shot CoT"
234
+ }
235
+ },
236
+ "metric_config": {
237
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
238
+ "metric_id": "mmlu_pro/economics",
239
+ "metric_name": "MMLU-Pro (Economics)",
240
+ "metric_kind": "accuracy",
241
+ "metric_unit": "proportion",
242
+ "lower_is_better": false,
243
+ "score_type": "continuous",
244
+ "min_score": 0.0,
245
+ "max_score": 1.0,
246
+ "additional_details": {
247
+ "aggregation": "accuracy_over_subset",
248
+ "prompt_style": "5-shot CoT"
249
+ }
250
+ },
251
+ "score_details": {
252
+ "score": 0.9336
253
+ }
254
+ },
255
+ {
256
+ "evaluation_result_id": "mmlu_pro/engineering",
257
+ "evaluation_name": "MMLU-Pro (Engineering)",
258
+ "source_data": {
259
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
260
+ "source_type": "hf_dataset",
261
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
262
+ "hf_split": "train",
263
+ "additional_details": {
264
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
265
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
266
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
267
+ "paper_url": "https://arxiv.org/abs/2406.01574",
268
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
269
+ "dataset_total_questions": "12000",
270
+ "prompt_style": "5-shot CoT"
271
+ }
272
+ },
273
+ "metric_config": {
274
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
275
+ "metric_id": "mmlu_pro/engineering",
276
+ "metric_name": "MMLU-Pro (Engineering)",
277
+ "metric_kind": "accuracy",
278
+ "metric_unit": "proportion",
279
+ "lower_is_better": false,
280
+ "score_type": "continuous",
281
+ "min_score": 0.0,
282
+ "max_score": 1.0,
283
+ "additional_details": {
284
+ "aggregation": "accuracy_over_subset",
285
+ "prompt_style": "5-shot CoT"
286
+ }
287
+ },
288
+ "score_details": {
289
+ "score": 0.8733
290
+ }
291
+ },
292
+ {
293
+ "evaluation_result_id": "mmlu_pro/health",
294
+ "evaluation_name": "MMLU-Pro (Health)",
295
+ "source_data": {
296
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
297
+ "source_type": "hf_dataset",
298
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
299
+ "hf_split": "train",
300
+ "additional_details": {
301
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
302
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
303
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
304
+ "paper_url": "https://arxiv.org/abs/2406.01574",
305
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
306
+ "dataset_total_questions": "12000",
307
+ "prompt_style": "5-shot CoT"
308
+ }
309
+ },
310
+ "metric_config": {
311
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
312
+ "metric_id": "mmlu_pro/health",
313
+ "metric_name": "MMLU-Pro (Health)",
314
+ "metric_kind": "accuracy",
315
+ "metric_unit": "proportion",
316
+ "lower_is_better": false,
317
+ "score_type": "continuous",
318
+ "min_score": 0.0,
319
+ "max_score": 1.0,
320
+ "additional_details": {
321
+ "aggregation": "accuracy_over_subset",
322
+ "prompt_style": "5-shot CoT"
323
+ }
324
+ },
325
+ "score_details": {
326
+ "score": 0.8619
327
+ }
328
+ },
329
+ {
330
+ "evaluation_result_id": "mmlu_pro/history",
331
+ "evaluation_name": "MMLU-Pro (History)",
332
+ "source_data": {
333
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
334
+ "source_type": "hf_dataset",
335
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
336
+ "hf_split": "train",
337
+ "additional_details": {
338
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
339
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
340
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
341
+ "paper_url": "https://arxiv.org/abs/2406.01574",
342
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
343
+ "dataset_total_questions": "12000",
344
+ "prompt_style": "5-shot CoT"
345
+ }
346
+ },
347
+ "metric_config": {
348
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
349
+ "metric_id": "mmlu_pro/history",
350
+ "metric_name": "MMLU-Pro (History)",
351
+ "metric_kind": "accuracy",
352
+ "metric_unit": "proportion",
353
+ "lower_is_better": false,
354
+ "score_type": "continuous",
355
+ "min_score": 0.0,
356
+ "max_score": 1.0,
357
+ "additional_details": {
358
+ "aggregation": "accuracy_over_subset",
359
+ "prompt_style": "5-shot CoT"
360
+ }
361
+ },
362
+ "score_details": {
363
+ "score": 0.8556
364
+ }
365
+ },
366
+ {
367
+ "evaluation_result_id": "mmlu_pro/law",
368
+ "evaluation_name": "MMLU-Pro (Law)",
369
+ "source_data": {
370
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
371
+ "source_type": "hf_dataset",
372
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
373
+ "hf_split": "train",
374
+ "additional_details": {
375
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
376
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
377
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
378
+ "paper_url": "https://arxiv.org/abs/2406.01574",
379
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
380
+ "dataset_total_questions": "12000",
381
+ "prompt_style": "5-shot CoT"
382
+ }
383
+ },
384
+ "metric_config": {
385
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
386
+ "metric_id": "mmlu_pro/law",
387
+ "metric_name": "MMLU-Pro (Law)",
388
+ "metric_kind": "accuracy",
389
+ "metric_unit": "proportion",
390
+ "lower_is_better": false,
391
+ "score_type": "continuous",
392
+ "min_score": 0.0,
393
+ "max_score": 1.0,
394
+ "additional_details": {
395
+ "aggregation": "accuracy_over_subset",
396
+ "prompt_style": "5-shot CoT"
397
+ }
398
+ },
399
+ "score_details": {
400
+ "score": 0.8564
401
+ }
402
+ },
403
+ {
404
+ "evaluation_result_id": "mmlu_pro/math",
405
+ "evaluation_name": "MMLU-Pro (Math)",
406
+ "source_data": {
407
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
408
+ "source_type": "hf_dataset",
409
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
410
+ "hf_split": "train",
411
+ "additional_details": {
412
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
413
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
414
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
415
+ "paper_url": "https://arxiv.org/abs/2406.01574",
416
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
417
+ "dataset_total_questions": "12000",
418
+ "prompt_style": "5-shot CoT"
419
+ }
420
+ },
421
+ "metric_config": {
422
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
423
+ "metric_id": "mmlu_pro/math",
424
+ "metric_name": "MMLU-Pro (Math)",
425
+ "metric_kind": "accuracy",
426
+ "metric_unit": "proportion",
427
+ "lower_is_better": false,
428
+ "score_type": "continuous",
429
+ "min_score": 0.0,
430
+ "max_score": 1.0,
431
+ "additional_details": {
432
+ "aggregation": "accuracy_over_subset",
433
+ "prompt_style": "5-shot CoT"
434
+ }
435
+ },
436
+ "score_details": {
437
+ "score": 0.9548
438
+ }
439
+ },
440
+ {
441
+ "evaluation_result_id": "mmlu_pro/philosophy",
442
+ "evaluation_name": "MMLU-Pro (Philosophy)",
443
+ "source_data": {
444
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
445
+ "source_type": "hf_dataset",
446
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
447
+ "hf_split": "train",
448
+ "additional_details": {
449
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
450
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
451
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
452
+ "paper_url": "https://arxiv.org/abs/2406.01574",
453
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
454
+ "dataset_total_questions": "12000",
455
+ "prompt_style": "5-shot CoT"
456
+ }
457
+ },
458
+ "metric_config": {
459
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
460
+ "metric_id": "mmlu_pro/philosophy",
461
+ "metric_name": "MMLU-Pro (Philosophy)",
462
+ "metric_kind": "accuracy",
463
+ "metric_unit": "proportion",
464
+ "lower_is_better": false,
465
+ "score_type": "continuous",
466
+ "min_score": 0.0,
467
+ "max_score": 1.0,
468
+ "additional_details": {
469
+ "aggregation": "accuracy_over_subset",
470
+ "prompt_style": "5-shot CoT"
471
+ }
472
+ },
473
+ "score_details": {
474
+ "score": 0.9038
475
+ }
476
+ },
477
+ {
478
+ "evaluation_result_id": "mmlu_pro/physics",
479
+ "evaluation_name": "MMLU-Pro (Physics)",
480
+ "source_data": {
481
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
482
+ "source_type": "hf_dataset",
483
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
484
+ "hf_split": "train",
485
+ "additional_details": {
486
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
487
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
488
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
489
+ "paper_url": "https://arxiv.org/abs/2406.01574",
490
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
491
+ "dataset_total_questions": "12000",
492
+ "prompt_style": "5-shot CoT"
493
+ }
494
+ },
495
+ "metric_config": {
496
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
497
+ "metric_id": "mmlu_pro/physics",
498
+ "metric_name": "MMLU-Pro (Physics)",
499
+ "metric_kind": "accuracy",
500
+ "metric_unit": "proportion",
501
+ "lower_is_better": false,
502
+ "score_type": "continuous",
503
+ "min_score": 0.0,
504
+ "max_score": 1.0,
505
+ "additional_details": {
506
+ "aggregation": "accuracy_over_subset",
507
+ "prompt_style": "5-shot CoT"
508
+ }
509
+ },
510
+ "score_details": {
511
+ "score": 0.932
512
+ }
513
+ },
514
+ {
515
+ "evaluation_result_id": "mmlu_pro/psychology",
516
+ "evaluation_name": "MMLU-Pro (Psychology)",
517
+ "source_data": {
518
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
519
+ "source_type": "hf_dataset",
520
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
521
+ "hf_split": "train",
522
+ "additional_details": {
523
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
524
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
525
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
526
+ "paper_url": "https://arxiv.org/abs/2406.01574",
527
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
528
+ "dataset_total_questions": "12000",
529
+ "prompt_style": "5-shot CoT"
530
+ }
531
+ },
532
+ "metric_config": {
533
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
534
+ "metric_id": "mmlu_pro/psychology",
535
+ "metric_name": "MMLU-Pro (Psychology)",
536
+ "metric_kind": "accuracy",
537
+ "metric_unit": "proportion",
538
+ "lower_is_better": false,
539
+ "score_type": "continuous",
540
+ "min_score": 0.0,
541
+ "max_score": 1.0,
542
+ "additional_details": {
543
+ "aggregation": "accuracy_over_subset",
544
+ "prompt_style": "5-shot CoT"
545
+ }
546
+ },
547
+ "score_details": {
548
+ "score": 0.9185
549
+ }
550
+ },
551
+ {
552
+ "evaluation_result_id": "mmlu_pro/other",
553
+ "evaluation_name": "MMLU-Pro (Other)",
554
+ "source_data": {
555
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
556
+ "source_type": "hf_dataset",
557
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
558
+ "hf_split": "train",
559
+ "additional_details": {
560
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
561
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
562
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
563
+ "paper_url": "https://arxiv.org/abs/2406.01574",
564
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
565
+ "dataset_total_questions": "12000",
566
+ "prompt_style": "5-shot CoT"
567
+ }
568
+ },
569
+ "metric_config": {
570
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
571
+ "metric_id": "mmlu_pro/other",
572
+ "metric_name": "MMLU-Pro (Other)",
573
+ "metric_kind": "accuracy",
574
+ "metric_unit": "proportion",
575
+ "lower_is_better": false,
576
+ "score_type": "continuous",
577
+ "min_score": 0.0,
578
+ "max_score": 1.0,
579
+ "additional_details": {
580
+ "aggregation": "accuracy_over_subset",
581
+ "prompt_style": "5-shot CoT"
582
+ }
583
+ },
584
+ "score_details": {
585
+ "score": 0.8929
586
+ }
587
+ }
588
+ ]
589
+ }
data/mmlu_pro/google/gemma-2-27b-it/e8295867-8053-488f-9130-9eb99b97670e.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-2-27b-it/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-2-27B-it",
25
+ "id": "google/gemma-2-27b-it",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-2-27B-it",
29
+ "size_billions_parameters": "27.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.5654
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.7796
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.6008
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.5371
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.5683
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.6979
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3488
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.6186
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.5722
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.3951
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.5611
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.523
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.5296
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.7155
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.6115
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/google/gemma-2-2b-it/7461516d-9358-42db-9888-056fd84c525f.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-2-2b-it/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-2-2B-it",
25
+ "id": "google/gemma-2-2b-it",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-2-2B-it",
29
+ "size_billions_parameters": "2.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.156
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/google/gemma-2-9b-it/c9fcb231-ddb3-4c0a-8d57-3a3782f0d752.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-2-9b-it/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-2-9B-it",
25
+ "id": "google/gemma-2-9b-it",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-2-9B-it",
29
+ "size_billions_parameters": "9.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.5208
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.7587
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.5539
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.4664
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.5073
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.6552
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3622
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.5844
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.5354
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.3579
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.4944
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.495
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.4758
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.6617
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.5498
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/google/gemma-2-9b/9c2c6d87-805f-47c0-9ddf-e0b81d0f69f8.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-2-9b/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-2-9B",
25
+ "id": "google/gemma-2-9b",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-2-9B",
29
+ "size_billions_parameters": "9.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.451
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.6457
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.4284
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.3746
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.4122
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.5486
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3075
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.5232
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.4987
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2843
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.4041
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.485
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.4296
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.6353
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.5271
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/google/gemma-2b/eb556f91-0cbd-4530-a7d1-d80e86d42588.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-2b/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-2B",
25
+ "id": "google/gemma-2b",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-2B",
29
+ "size_billions_parameters": "2.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.1585
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.2482
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.1457
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1378
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.1414
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.1753
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.1269
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.177
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.154
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.123
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.163
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.1482
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.1563
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.1608
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.1817
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/google/gemma-3-12b-it/83198456-6a19-4574-abf8-0a0cc97287af.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-3-12b-it/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-3-12B-it",
25
+ "id": "google/gemma-3-12b-it",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-3-12B-it",
29
+ "size_billions_parameters": "12.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.606
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/google/gemma-3-1b-it/e3c1d8e7-a98f-4fd4-ac24-3d0c7c3669a9.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-3-1b-it/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-3-1B-it",
25
+ "id": "google/gemma-3-1b-it",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-3-1B-it",
29
+ "size_billions_parameters": "1.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.147
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/google/gemma-3-27b-it/6a53d694-91e0-40bd-a265-1f640756ae66.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-3-27b-it/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-3-27B-it",
25
+ "id": "google/gemma-3-27b-it",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-3-27B-it",
29
+ "size_billions_parameters": "27.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.675
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/google/gemma-3-4b-it/e7c8cf64-4f7a-4a14-8517-00760409be7c.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-3-4b-it/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-3-4B-it",
25
+ "id": "google/gemma-3-4b-it",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-3-4B-it",
29
+ "size_billions_parameters": "4.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.436
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/google/gemma-7b/1e21b40e-8ae9-4712-87ad-344e5423846b.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/google_gemma-7b/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Gemma-7B",
25
+ "id": "google/gemma-7b",
26
+ "developer": "google",
27
+ "additional_details": {
28
+ "raw_model_name": "Gemma-7B",
29
+ "size_billions_parameters": "7.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.3373
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.5649
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.3333
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.2624
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.3659
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.4242
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.227
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.3716
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.3675
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2171
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.2509
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.3908
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.2756
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.5175
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.4091
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.0-2b-base/47ec1fc0-0ac3-4d86-87f6-274667ef2a16.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.0-2b-base/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.0-2B-Base",
25
+ "id": "ibm/granite-3.0-2b-base",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.0-2B-Base",
29
+ "size_billions_parameters": "2.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.2172
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.3445
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.1977
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1564
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.2659
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.3033
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.1465
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.2298
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.1837
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.1653
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.2058
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.2385
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.164
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.3271
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.2327
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.0-8b-base/5e811a7a-5a2c-4484-9368-5d62dab1f13f.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.0-8b-base/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.0-8B-Base",
25
+ "id": "ibm/granite-3.0-8b-base",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.0-8B-Base",
29
+ "size_billions_parameters": "8.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.3103
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.4728
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.308
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.2217
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.3268
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.4135
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.2136
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.3863
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.3307
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2316
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.2805
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.3467
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.2494
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.4336
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.3149
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.1-1b-a400m-base/1f8d80d4-01a3-4818-9a8b-d00054dcab43.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.1-1b-a400m-base/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.1-1B-A400M-Base",
25
+ "id": "ibm/granite-3.1-1b-a400m-base",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.1-1B-A400M-Base",
29
+ "size_billions_parameters": "1.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.1234
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.1353
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.1153
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1246
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.1415
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.1422
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.098
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.1308
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.1234
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.1126
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.131
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.1683
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.1124
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.1078
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.1212
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.1-1b-a400m-instruct/f3eeed4c-1d73-46f0-acac-eafd5fc9754f.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.1-1b-a400m-instruct/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.1-1B-A400M-Instruct",
25
+ "id": "ibm/granite-3.1-1b-a400m-instruct",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.1-1B-A400M-Instruct",
29
+ "size_billions_parameters": "1.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.1327
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.1437
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.1267
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1148
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.161
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.1576
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.1125
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.1638
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.1129
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.1253
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.1303
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.1202
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.1209
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.1617
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.1288
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.1-2b-base/ba525354-6d57-4b26-b777-e58271890ae6.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.1-2b-base/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.1-2B-Base",
25
+ "id": "ibm/granite-3.1-2b-base",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.1-2B-Base",
29
+ "size_billions_parameters": "2.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.2389
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.3752
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.256
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1696
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.2439
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.3092
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.193
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.2604
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.2178
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.1253
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.2487
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.2525
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.1986
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.3421
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.2565
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.1-2b-instruct/7959f684-3532-439e-a9e0-5389dcaf795c.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.1-2b-instruct/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.1-2B-Instruct",
25
+ "id": "ibm/granite-3.1-2b-instruct",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.1-2B-Instruct",
29
+ "size_billions_parameters": "2.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.3197
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.5007
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.3308
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.2412
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.3707
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.4111
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.2528
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.3056
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.3045
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.218
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.3442
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.2846
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.2648
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.4411
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.3258
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.1-3b-a800m-base/95d5b3bc-f7e5-4171-9be8-9656a902d92f.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.1-3b-a800m-base/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.1-3B-A800M-Base",
25
+ "id": "ibm/granite-3.1-3b-a800m-base",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.1-3B-A800M-Base",
29
+ "size_billions_parameters": "3.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.2039
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.2957
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.1762
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1405
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.2268
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.2737
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.1527
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.2286
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.1995
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.1444
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.2198
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.2305
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.164
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.3083
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.1926
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.1-3b-a800m-instruct/b186c499-e9e3-4fef-8bbf-3c43f7c71e28.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.1-3b-a800m-instruct/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.1-3B-A800M-Instruct",
25
+ "id": "ibm/granite-3.1-3b-a800m-instruct",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.1-3B-A800M-Instruct",
29
+ "size_billions_parameters": "3.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.2542
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.3431
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.2725
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.1608
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.2829
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.3626
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.1744
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.2641
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.2415
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.1708
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.2754
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.2806
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.2017
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.381
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.2706
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.1-8b-base/6b91c3fc-4ec1-480c-9e78-99440cbfe97d.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.1-8b-base/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.1-8B-Base",
25
+ "id": "ibm/granite-3.1-8b-base",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.1-8B-Base",
29
+ "size_billions_parameters": "8.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.3308
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.4979
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.3181
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.2403
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.339
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.4372
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.2425
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.3716
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.3412
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2044
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.3249
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.3567
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.2748
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.4862
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.3636
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/ibm/granite-3.1-8b-instruct/07107f2c-5b3e-4b5d-87ec-0acc108a68f9.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/ibm_granite-3.1-8b-instruct/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Granite-3.1-8B-Instruct",
25
+ "id": "ibm/granite-3.1-8b-instruct",
26
+ "developer": "ibm",
27
+ "additional_details": {
28
+ "raw_model_name": "Granite-3.1-8B-Instruct",
29
+ "size_billions_parameters": "8.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.4103
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.5746
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.4563
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.3145
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.4244
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.5047
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.291
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.4707
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.4121
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2607
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.4189
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.4329
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.3472
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.5739
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.4405
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/microsoft/phi-3.5-mini-instruct/61ae9c04-3c9a-47e3-bcab-eea1268d14fe.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/microsoft_phi-3.5-mini-instruct/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Phi-3.5-mini-instruct",
25
+ "id": "microsoft/phi-3.5-mini-instruct",
26
+ "developer": "microsoft",
27
+ "additional_details": {
28
+ "raw_model_name": "Phi-3.5-mini-instruct",
29
+ "size_billions_parameters": "3.8",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.4787
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.7057
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.5349
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.4125
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.5195
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.6386
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3075
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.5244
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.4252
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.2943
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.49
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.5
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.4509
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.4188
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.6353
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/microsoft/phi-4-mini/86cb1a4f-477f-4c08-8fb4-d0a6153984c3.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/microsoft_phi-4-mini/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Phi-4-mini",
25
+ "id": "microsoft/phi-4-mini",
26
+ "developer": "microsoft",
27
+ "additional_details": {
28
+ "raw_model_name": "Phi-4-mini",
29
+ "size_billions_parameters": "5.6",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.528
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/microsoft/phi-4-reasoning-plus/7f6559ff-c8dd-4fe7-9cd4-c1cd34e65c8b.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/microsoft_phi-4-reasoning-plus/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Phi-4-reasoning-plus",
25
+ "id": "microsoft/phi-4-reasoning-plus",
26
+ "developer": "microsoft",
27
+ "additional_details": {
28
+ "raw_model_name": "Phi-4-reasoning-plus",
29
+ "size_billions_parameters": "14.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.76
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/microsoft/phi-4-reasoning/a61f15ba-23b9-48d9-bf7c-c1911d765421.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/microsoft_phi-4-reasoning/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Phi-4-reasoning",
25
+ "id": "microsoft/phi-4-reasoning",
26
+ "developer": "microsoft",
27
+ "additional_details": {
28
+ "raw_model_name": "Phi-4-reasoning",
29
+ "size_billions_parameters": "14.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.743
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/microsoft/phi-4/c6c805a9-7065-4811-a2ee-aa9b3e065e09.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/microsoft_phi-4/self-reported/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "Self-Reported"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Phi-4",
25
+ "id": "microsoft/phi-4",
26
+ "developer": "microsoft",
27
+ "additional_details": {
28
+ "raw_model_name": "Phi-4",
29
+ "size_billions_parameters": "14.0",
30
+ "leaderboard_data_source": "Self-Reported"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.704
69
+ }
70
+ }
71
+ ]
72
+ }
data/mmlu_pro/microsoft/phi3-medium-128k/cfba928a-8592-47f2-bc5d-bb97cd624352.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/microsoft_phi3-medium-128k/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Phi3-medium-128k",
25
+ "id": "microsoft/phi3-medium-128k",
26
+ "developer": "microsoft",
27
+ "additional_details": {
28
+ "raw_model_name": "Phi3-medium-128k",
29
+ "size_billions_parameters": "14.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.5191
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.7336
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.564
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.4382
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.5171
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.6647
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3437
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.5856
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.5381
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.3597
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.4989
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.491
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.4519
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.7093
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.5639
587
+ }
588
+ }
589
+ ]
590
+ }
data/mmlu_pro/microsoft/phi3-medium-4k/996cc51e-f338-4192-ac74-c3149c097648.json ADDED
@@ -0,0 +1,590 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": "0.2.2",
3
+ "evaluation_id": "mmlu-pro/microsoft_phi3-medium-4k/tiger-lab/1783223217.608336",
4
+ "retrieved_timestamp": "1783223217.608336",
5
+ "source_metadata": {
6
+ "source_name": "MMLU-Pro Leaderboard",
7
+ "source_type": "documentation",
8
+ "source_organization_name": "TIGER-Lab",
9
+ "source_organization_url": "https://tiger-ai-lab.github.io",
10
+ "evaluator_relationship": "third_party",
11
+ "additional_details": {
12
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
13
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
14
+ "paper_url": "https://arxiv.org/abs/2406.01574",
15
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
16
+ "leaderboard_data_source": "TIGER-Lab"
17
+ }
18
+ },
19
+ "eval_library": {
20
+ "name": "MMLU-Pro leaderboard (TIGER-Lab)",
21
+ "version": "unknown"
22
+ },
23
+ "model_info": {
24
+ "name": "Phi3-medium-4k",
25
+ "id": "microsoft/phi3-medium-4k",
26
+ "developer": "microsoft",
27
+ "additional_details": {
28
+ "raw_model_name": "Phi3-medium-4k",
29
+ "size_billions_parameters": "14.0",
30
+ "leaderboard_data_source": "TIGER-Lab"
31
+ }
32
+ },
33
+ "evaluation_results": [
34
+ {
35
+ "evaluation_result_id": "mmlu_pro/overall",
36
+ "evaluation_name": "MMLU-Pro (overall)",
37
+ "source_data": {
38
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
39
+ "source_type": "hf_dataset",
40
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
41
+ "hf_split": "train",
42
+ "additional_details": {
43
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
44
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
45
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
46
+ "paper_url": "https://arxiv.org/abs/2406.01574",
47
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
48
+ "dataset_total_questions": "12000",
49
+ "prompt_style": "5-shot CoT"
50
+ }
51
+ },
52
+ "metric_config": {
53
+ "evaluation_description": "Overall accuracy across the ~12,000-question MMLU-Pro benchmark, evaluated 5-shot with chain-of-thought.",
54
+ "metric_id": "mmlu_pro/overall",
55
+ "metric_name": "MMLU-Pro (overall)",
56
+ "metric_kind": "accuracy",
57
+ "metric_unit": "proportion",
58
+ "lower_is_better": false,
59
+ "score_type": "continuous",
60
+ "min_score": 0.0,
61
+ "max_score": 1.0,
62
+ "additional_details": {
63
+ "aggregation": "accuracy_over_subset",
64
+ "prompt_style": "5-shot CoT"
65
+ }
66
+ },
67
+ "score_details": {
68
+ "score": 0.557
69
+ }
70
+ },
71
+ {
72
+ "evaluation_result_id": "mmlu_pro/biology",
73
+ "evaluation_name": "MMLU-Pro (Biology)",
74
+ "source_data": {
75
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
76
+ "source_type": "hf_dataset",
77
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
78
+ "hf_split": "train",
79
+ "additional_details": {
80
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
81
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
82
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
83
+ "paper_url": "https://arxiv.org/abs/2406.01574",
84
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
85
+ "dataset_total_questions": "12000",
86
+ "prompt_style": "5-shot CoT"
87
+ }
88
+ },
89
+ "metric_config": {
90
+ "evaluation_description": "Accuracy on the MMLU-Pro Biology subset, evaluated 5-shot with chain-of-thought.",
91
+ "metric_id": "mmlu_pro/biology",
92
+ "metric_name": "MMLU-Pro (Biology)",
93
+ "metric_kind": "accuracy",
94
+ "metric_unit": "proportion",
95
+ "lower_is_better": false,
96
+ "score_type": "continuous",
97
+ "min_score": 0.0,
98
+ "max_score": 1.0,
99
+ "additional_details": {
100
+ "aggregation": "accuracy_over_subset",
101
+ "prompt_style": "5-shot CoT"
102
+ }
103
+ },
104
+ "score_details": {
105
+ "score": 0.7587
106
+ }
107
+ },
108
+ {
109
+ "evaluation_result_id": "mmlu_pro/business",
110
+ "evaluation_name": "MMLU-Pro (Business)",
111
+ "source_data": {
112
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
113
+ "source_type": "hf_dataset",
114
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
115
+ "hf_split": "train",
116
+ "additional_details": {
117
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
118
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
119
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
120
+ "paper_url": "https://arxiv.org/abs/2406.01574",
121
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
122
+ "dataset_total_questions": "12000",
123
+ "prompt_style": "5-shot CoT"
124
+ }
125
+ },
126
+ "metric_config": {
127
+ "evaluation_description": "Accuracy on the MMLU-Pro Business subset, evaluated 5-shot with chain-of-thought.",
128
+ "metric_id": "mmlu_pro/business",
129
+ "metric_name": "MMLU-Pro (Business)",
130
+ "metric_kind": "accuracy",
131
+ "metric_unit": "proportion",
132
+ "lower_is_better": false,
133
+ "score_type": "continuous",
134
+ "min_score": 0.0,
135
+ "max_score": 1.0,
136
+ "additional_details": {
137
+ "aggregation": "accuracy_over_subset",
138
+ "prompt_style": "5-shot CoT"
139
+ }
140
+ },
141
+ "score_details": {
142
+ "score": 0.616
143
+ }
144
+ },
145
+ {
146
+ "evaluation_result_id": "mmlu_pro/chemistry",
147
+ "evaluation_name": "MMLU-Pro (Chemistry)",
148
+ "source_data": {
149
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
150
+ "source_type": "hf_dataset",
151
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
152
+ "hf_split": "train",
153
+ "additional_details": {
154
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
155
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
156
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
157
+ "paper_url": "https://arxiv.org/abs/2406.01574",
158
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
159
+ "dataset_total_questions": "12000",
160
+ "prompt_style": "5-shot CoT"
161
+ }
162
+ },
163
+ "metric_config": {
164
+ "evaluation_description": "Accuracy on the MMLU-Pro Chemistry subset, evaluated 5-shot with chain-of-thought.",
165
+ "metric_id": "mmlu_pro/chemistry",
166
+ "metric_name": "MMLU-Pro (Chemistry)",
167
+ "metric_kind": "accuracy",
168
+ "metric_unit": "proportion",
169
+ "lower_is_better": false,
170
+ "score_type": "continuous",
171
+ "min_score": 0.0,
172
+ "max_score": 1.0,
173
+ "additional_details": {
174
+ "aggregation": "accuracy_over_subset",
175
+ "prompt_style": "5-shot CoT"
176
+ }
177
+ },
178
+ "score_details": {
179
+ "score": 0.4991
180
+ }
181
+ },
182
+ {
183
+ "evaluation_result_id": "mmlu_pro/computer_science",
184
+ "evaluation_name": "MMLU-Pro (Computer Science)",
185
+ "source_data": {
186
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
187
+ "source_type": "hf_dataset",
188
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
189
+ "hf_split": "train",
190
+ "additional_details": {
191
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
192
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
193
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
194
+ "paper_url": "https://arxiv.org/abs/2406.01574",
195
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
196
+ "dataset_total_questions": "12000",
197
+ "prompt_style": "5-shot CoT"
198
+ }
199
+ },
200
+ "metric_config": {
201
+ "evaluation_description": "Accuracy on the MMLU-Pro Computer Science subset, evaluated 5-shot with chain-of-thought.",
202
+ "metric_id": "mmlu_pro/computer_science",
203
+ "metric_name": "MMLU-Pro (Computer Science)",
204
+ "metric_kind": "accuracy",
205
+ "metric_unit": "proportion",
206
+ "lower_is_better": false,
207
+ "score_type": "continuous",
208
+ "min_score": 0.0,
209
+ "max_score": 1.0,
210
+ "additional_details": {
211
+ "aggregation": "accuracy_over_subset",
212
+ "prompt_style": "5-shot CoT"
213
+ }
214
+ },
215
+ "score_details": {
216
+ "score": 0.5415
217
+ }
218
+ },
219
+ {
220
+ "evaluation_result_id": "mmlu_pro/economics",
221
+ "evaluation_name": "MMLU-Pro (Economics)",
222
+ "source_data": {
223
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
224
+ "source_type": "hf_dataset",
225
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
226
+ "hf_split": "train",
227
+ "additional_details": {
228
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
229
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
230
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
231
+ "paper_url": "https://arxiv.org/abs/2406.01574",
232
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
233
+ "dataset_total_questions": "12000",
234
+ "prompt_style": "5-shot CoT"
235
+ }
236
+ },
237
+ "metric_config": {
238
+ "evaluation_description": "Accuracy on the MMLU-Pro Economics subset, evaluated 5-shot with chain-of-thought.",
239
+ "metric_id": "mmlu_pro/economics",
240
+ "metric_name": "MMLU-Pro (Economics)",
241
+ "metric_kind": "accuracy",
242
+ "metric_unit": "proportion",
243
+ "lower_is_better": false,
244
+ "score_type": "continuous",
245
+ "min_score": 0.0,
246
+ "max_score": 1.0,
247
+ "additional_details": {
248
+ "aggregation": "accuracy_over_subset",
249
+ "prompt_style": "5-shot CoT"
250
+ }
251
+ },
252
+ "score_details": {
253
+ "score": 0.7038
254
+ }
255
+ },
256
+ {
257
+ "evaluation_result_id": "mmlu_pro/engineering",
258
+ "evaluation_name": "MMLU-Pro (Engineering)",
259
+ "source_data": {
260
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
261
+ "source_type": "hf_dataset",
262
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
263
+ "hf_split": "train",
264
+ "additional_details": {
265
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
266
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
267
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
268
+ "paper_url": "https://arxiv.org/abs/2406.01574",
269
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
270
+ "dataset_total_questions": "12000",
271
+ "prompt_style": "5-shot CoT"
272
+ }
273
+ },
274
+ "metric_config": {
275
+ "evaluation_description": "Accuracy on the MMLU-Pro Engineering subset, evaluated 5-shot with chain-of-thought.",
276
+ "metric_id": "mmlu_pro/engineering",
277
+ "metric_name": "MMLU-Pro (Engineering)",
278
+ "metric_kind": "accuracy",
279
+ "metric_unit": "proportion",
280
+ "lower_is_better": false,
281
+ "score_type": "continuous",
282
+ "min_score": 0.0,
283
+ "max_score": 1.0,
284
+ "additional_details": {
285
+ "aggregation": "accuracy_over_subset",
286
+ "prompt_style": "5-shot CoT"
287
+ }
288
+ },
289
+ "score_details": {
290
+ "score": 0.3787
291
+ }
292
+ },
293
+ {
294
+ "evaluation_result_id": "mmlu_pro/health",
295
+ "evaluation_name": "MMLU-Pro (Health)",
296
+ "source_data": {
297
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
298
+ "source_type": "hf_dataset",
299
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
300
+ "hf_split": "train",
301
+ "additional_details": {
302
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
303
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
304
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
305
+ "paper_url": "https://arxiv.org/abs/2406.01574",
306
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
307
+ "dataset_total_questions": "12000",
308
+ "prompt_style": "5-shot CoT"
309
+ }
310
+ },
311
+ "metric_config": {
312
+ "evaluation_description": "Accuracy on the MMLU-Pro Health subset, evaluated 5-shot with chain-of-thought.",
313
+ "metric_id": "mmlu_pro/health",
314
+ "metric_name": "MMLU-Pro (Health)",
315
+ "metric_kind": "accuracy",
316
+ "metric_unit": "proportion",
317
+ "lower_is_better": false,
318
+ "score_type": "continuous",
319
+ "min_score": 0.0,
320
+ "max_score": 1.0,
321
+ "additional_details": {
322
+ "aggregation": "accuracy_over_subset",
323
+ "prompt_style": "5-shot CoT"
324
+ }
325
+ },
326
+ "score_details": {
327
+ "score": 0.6357
328
+ }
329
+ },
330
+ {
331
+ "evaluation_result_id": "mmlu_pro/history",
332
+ "evaluation_name": "MMLU-Pro (History)",
333
+ "source_data": {
334
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
335
+ "source_type": "hf_dataset",
336
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
337
+ "hf_split": "train",
338
+ "additional_details": {
339
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
340
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
341
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
342
+ "paper_url": "https://arxiv.org/abs/2406.01574",
343
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
344
+ "dataset_total_questions": "12000",
345
+ "prompt_style": "5-shot CoT"
346
+ }
347
+ },
348
+ "metric_config": {
349
+ "evaluation_description": "Accuracy on the MMLU-Pro History subset, evaluated 5-shot with chain-of-thought.",
350
+ "metric_id": "mmlu_pro/history",
351
+ "metric_name": "MMLU-Pro (History)",
352
+ "metric_kind": "accuracy",
353
+ "metric_unit": "proportion",
354
+ "lower_is_better": false,
355
+ "score_type": "continuous",
356
+ "min_score": 0.0,
357
+ "max_score": 1.0,
358
+ "additional_details": {
359
+ "aggregation": "accuracy_over_subset",
360
+ "prompt_style": "5-shot CoT"
361
+ }
362
+ },
363
+ "score_details": {
364
+ "score": 0.5722
365
+ }
366
+ },
367
+ {
368
+ "evaluation_result_id": "mmlu_pro/law",
369
+ "evaluation_name": "MMLU-Pro (Law)",
370
+ "source_data": {
371
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
372
+ "source_type": "hf_dataset",
373
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
374
+ "hf_split": "train",
375
+ "additional_details": {
376
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
377
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
378
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
379
+ "paper_url": "https://arxiv.org/abs/2406.01574",
380
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
381
+ "dataset_total_questions": "12000",
382
+ "prompt_style": "5-shot CoT"
383
+ }
384
+ },
385
+ "metric_config": {
386
+ "evaluation_description": "Accuracy on the MMLU-Pro Law subset, evaluated 5-shot with chain-of-thought.",
387
+ "metric_id": "mmlu_pro/law",
388
+ "metric_name": "MMLU-Pro (Law)",
389
+ "metric_kind": "accuracy",
390
+ "metric_unit": "proportion",
391
+ "lower_is_better": false,
392
+ "score_type": "continuous",
393
+ "min_score": 0.0,
394
+ "max_score": 1.0,
395
+ "additional_details": {
396
+ "aggregation": "accuracy_over_subset",
397
+ "prompt_style": "5-shot CoT"
398
+ }
399
+ },
400
+ "score_details": {
401
+ "score": 0.3833
402
+ }
403
+ },
404
+ {
405
+ "evaluation_result_id": "mmlu_pro/math",
406
+ "evaluation_name": "MMLU-Pro (Math)",
407
+ "source_data": {
408
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
409
+ "source_type": "hf_dataset",
410
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
411
+ "hf_split": "train",
412
+ "additional_details": {
413
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
414
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
415
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
416
+ "paper_url": "https://arxiv.org/abs/2406.01574",
417
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
418
+ "dataset_total_questions": "12000",
419
+ "prompt_style": "5-shot CoT"
420
+ }
421
+ },
422
+ "metric_config": {
423
+ "evaluation_description": "Accuracy on the MMLU-Pro Math subset, evaluated 5-shot with chain-of-thought.",
424
+ "metric_id": "mmlu_pro/math",
425
+ "metric_name": "MMLU-Pro (Math)",
426
+ "metric_kind": "accuracy",
427
+ "metric_unit": "proportion",
428
+ "lower_is_better": false,
429
+ "score_type": "continuous",
430
+ "min_score": 0.0,
431
+ "max_score": 1.0,
432
+ "additional_details": {
433
+ "aggregation": "accuracy_over_subset",
434
+ "prompt_style": "5-shot CoT"
435
+ }
436
+ },
437
+ "score_details": {
438
+ "score": 0.5218
439
+ }
440
+ },
441
+ {
442
+ "evaluation_result_id": "mmlu_pro/philosophy",
443
+ "evaluation_name": "MMLU-Pro (Philosophy)",
444
+ "source_data": {
445
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
446
+ "source_type": "hf_dataset",
447
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
448
+ "hf_split": "train",
449
+ "additional_details": {
450
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
451
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
452
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
453
+ "paper_url": "https://arxiv.org/abs/2406.01574",
454
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
455
+ "dataset_total_questions": "12000",
456
+ "prompt_style": "5-shot CoT"
457
+ }
458
+ },
459
+ "metric_config": {
460
+ "evaluation_description": "Accuracy on the MMLU-Pro Philosophy subset, evaluated 5-shot with chain-of-thought.",
461
+ "metric_id": "mmlu_pro/philosophy",
462
+ "metric_name": "MMLU-Pro (Philosophy)",
463
+ "metric_kind": "accuracy",
464
+ "metric_unit": "proportion",
465
+ "lower_is_better": false,
466
+ "score_type": "continuous",
467
+ "min_score": 0.0,
468
+ "max_score": 1.0,
469
+ "additional_details": {
470
+ "aggregation": "accuracy_over_subset",
471
+ "prompt_style": "5-shot CoT"
472
+ }
473
+ },
474
+ "score_details": {
475
+ "score": 0.5511
476
+ }
477
+ },
478
+ {
479
+ "evaluation_result_id": "mmlu_pro/physics",
480
+ "evaluation_name": "MMLU-Pro (Physics)",
481
+ "source_data": {
482
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
483
+ "source_type": "hf_dataset",
484
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
485
+ "hf_split": "train",
486
+ "additional_details": {
487
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
488
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
489
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
490
+ "paper_url": "https://arxiv.org/abs/2406.01574",
491
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
492
+ "dataset_total_questions": "12000",
493
+ "prompt_style": "5-shot CoT"
494
+ }
495
+ },
496
+ "metric_config": {
497
+ "evaluation_description": "Accuracy on the MMLU-Pro Physics subset, evaluated 5-shot with chain-of-thought.",
498
+ "metric_id": "mmlu_pro/physics",
499
+ "metric_name": "MMLU-Pro (Physics)",
500
+ "metric_kind": "accuracy",
501
+ "metric_unit": "proportion",
502
+ "lower_is_better": false,
503
+ "score_type": "continuous",
504
+ "min_score": 0.0,
505
+ "max_score": 1.0,
506
+ "additional_details": {
507
+ "aggregation": "accuracy_over_subset",
508
+ "prompt_style": "5-shot CoT"
509
+ }
510
+ },
511
+ "score_details": {
512
+ "score": 0.4935
513
+ }
514
+ },
515
+ {
516
+ "evaluation_result_id": "mmlu_pro/psychology",
517
+ "evaluation_name": "MMLU-Pro (Psychology)",
518
+ "source_data": {
519
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
520
+ "source_type": "hf_dataset",
521
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
522
+ "hf_split": "train",
523
+ "additional_details": {
524
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
525
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
526
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
527
+ "paper_url": "https://arxiv.org/abs/2406.01574",
528
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
529
+ "dataset_total_questions": "12000",
530
+ "prompt_style": "5-shot CoT"
531
+ }
532
+ },
533
+ "metric_config": {
534
+ "evaluation_description": "Accuracy on the MMLU-Pro Psychology subset, evaluated 5-shot with chain-of-thought.",
535
+ "metric_id": "mmlu_pro/psychology",
536
+ "metric_name": "MMLU-Pro (Psychology)",
537
+ "metric_kind": "accuracy",
538
+ "metric_unit": "proportion",
539
+ "lower_is_better": false,
540
+ "score_type": "continuous",
541
+ "min_score": 0.0,
542
+ "max_score": 1.0,
543
+ "additional_details": {
544
+ "aggregation": "accuracy_over_subset",
545
+ "prompt_style": "5-shot CoT"
546
+ }
547
+ },
548
+ "score_details": {
549
+ "score": 0.7343
550
+ }
551
+ },
552
+ {
553
+ "evaluation_result_id": "mmlu_pro/other",
554
+ "evaluation_name": "MMLU-Pro (Other)",
555
+ "source_data": {
556
+ "dataset_name": "MMLU-Pro leaderboard submissions (TIGER-Lab)",
557
+ "source_type": "hf_dataset",
558
+ "hf_repo": "TIGER-Lab/mmlu_pro_leaderboard_submission",
559
+ "hf_split": "train",
560
+ "additional_details": {
561
+ "results_csv_url": "https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission/resolve/main/results.csv",
562
+ "leaderboard_space_url": "https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro",
563
+ "benchmark_hf_repo": "TIGER-Lab/MMLU-Pro",
564
+ "paper_url": "https://arxiv.org/abs/2406.01574",
565
+ "github_url": "https://github.com/TIGER-AI-Lab/MMLU-Pro",
566
+ "dataset_total_questions": "12000",
567
+ "prompt_style": "5-shot CoT"
568
+ }
569
+ },
570
+ "metric_config": {
571
+ "evaluation_description": "Accuracy on the MMLU-Pro Other subset, evaluated 5-shot with chain-of-thought.",
572
+ "metric_id": "mmlu_pro/other",
573
+ "metric_name": "MMLU-Pro (Other)",
574
+ "metric_kind": "accuracy",
575
+ "metric_unit": "proportion",
576
+ "lower_is_better": false,
577
+ "score_type": "continuous",
578
+ "min_score": 0.0,
579
+ "max_score": 1.0,
580
+ "additional_details": {
581
+ "aggregation": "accuracy_over_subset",
582
+ "prompt_style": "5-shot CoT"
583
+ }
584
+ },
585
+ "score_details": {
586
+ "score": 0.6028
587
+ }
588
+ }
589
+ ]
590
+ }