Hermes 4 Evaluations
Collection
Evals from the Hermes-4 Technical Report
•
20 items
•
Updated
doc
dict | metric
dict | model_response
dict | task_name
stringclasses 1
value |
|---|---|---|---|
{"choices":["321"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":0.875,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":0.8749999999999(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["480"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":0.9375,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":0.96875,"math_(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["104"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":0.875,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":0.90625,"math_p(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["055"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":0.9375,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":0.96875,"math_(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["236"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":0.875,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":0.90625,"math_p(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["721"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":0.6875,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":0.843749999999(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["104"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":0.7499999999999999,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":0.(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["540"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":1.0,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":0.96875,"math_pas(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["294"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":1.0,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":1.0,"math_pass@1:(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
{"choices":["073"],"fewshot_samples":[],"fewshot_sorting_class":null,"generation_grammar":null,"gene(...TRUNCATED)
| {"math_pass@1:16_samples":1.0,"math_pass@1:1_samples":1.0,"math_pass@1:32_samples":1.0,"math_pass@1:(...TRUNCATED)
| {"argmax_logits_eq_gold":[],"input":[{"content":"You are a deep thinking AI, you may use extremely l(...TRUNCATED)
|
lighteval|aime24|0
|
End of preview. Expand
in Data Studio
36bpsychev2 Evaluation Results
Summary
| Benchmark | Score | Metric | Samples | Overlong rate |
|---|---|---|---|---|
| aime24 | 0.719 | math_pass@1:64_samples | 64 | 17.6% |
| aime25 | 0.693 | math_pass@1:64_samples | 64 | 18.8% |
| bbh_generative | 0.864 | extractive_match | 1 | 4.8% |
| drop_generative_nous | 0.835 | drop_acc | 1 | 2.7% |
| gpqa_diamond | 0.655 | gpqa_pass@1:8_samples | 8 | 2.2% |
| ifeval | 0.779 | inst_level_loose_acc | 1 | 7.8% |
| math_500 | 0.938 | math_pass@1:4_samples | 4 | 1.8% |
| mmlu_generative | 0.877 | extractive_match | 1 | 0.1% |
| mmlu_pro | 0.807 | pass@1:1_samples | 1 | 1.1% |
| musr_generative | 0.697 | extractive_match | 1 | 0.4% |
| obqa_generative | 0.966 | extractive_match | 1 | 0.2% |
| simpleqa_nous | 0.060 | fuzzy_match | 1 | 8.6% |
Overlong rate: 1,846 / 54,663 samples (3.4%) missing closing </think> tag
Detailed Results
aime24
| Metric | Score | Std Error |
|---|---|---|
| math_pass@1:1_samples | 0.767 | 0.079 |
| math_pass@1:4_samples | 0.700 | 0.068 |
| math_pass@1:8_samples | 0.713 | 0.064 |
| math_pass@1:16_samples | 0.719 | 0.061 |
| math_pass@1:32_samples | 0.731 | 0.061 |
| math_pass@1:64_samples | 0.719 | 0.062 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 02:57:47
Temperature: 0.6
Overlong samples: 17.6% (337 / 1920)
aime25
| Metric | Score | Std Error |
|---|---|---|
| math_pass@1:1_samples | 0.633 | 0.089 |
| math_pass@1:4_samples | 0.692 | 0.062 |
| math_pass@1:8_samples | 0.662 | 0.063 |
| math_pass@1:16_samples | 0.694 | 0.061 |
| math_pass@1:32_samples | 0.693 | 0.062 |
| math_pass@1:64_samples | 0.693 | 0.063 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 03:06:24
Temperature: 0.6
Overlong samples: 18.8% (360 / 1920)
bbh_generative
| Metric | Score | Std Error |
|---|---|---|
| extractive_match | 0.864 | 0.017 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 01:27:42
Temperature: 0.6
Overlong samples: 4.8% (262 / 5511)
drop_generative_nous
| Metric | Score | Std Error |
|---|---|---|
| drop_acc | 0.835 | 0.004 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 01:22:41
Temperature: 0.6
Overlong samples: 2.7% (255 / 9536)
gpqa_diamond
| Metric | Score | Std Error |
|---|---|---|
| gpqa_pass@1:1_samples | 0.641 | 0.034 |
| gpqa_pass@1:4_samples | 0.663 | 0.027 |
| gpqa_pass@1:8_samples | 0.655 | 0.026 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 01:18:07
Temperature: 0.6
Overlong samples: 2.2% (35 / 1584)
ifeval
| Metric | Score | Std Error |
|---|---|---|
| prompt_level_strict_acc | 0.654 | 0.020 |
| inst_level_strict_acc | 0.740 | 0.000 |
| prompt_level_loose_acc | 0.702 | 0.020 |
| inst_level_loose_acc | 0.779 | 0.000 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 00:25:46
Temperature: 0.6
Overlong samples: 7.8% (42 / 541)
math_500
| Metric | Score | Std Error |
|---|---|---|
| math_pass@1:1_samples | 0.942 | 0.010 |
| math_pass@1:4_samples | 0.938 | 0.008 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 00:33:46
Temperature: 0.6
Overlong samples: 1.8% (35 / 2000)
mmlu_generative
| Metric | Score | Std Error |
|---|---|---|
| extractive_match | 0.877 | 0.003 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 01:46:37
Temperature: 0.6
Overlong samples: 0.1% (16 / 14042)
mmlu_pro
| Metric | Score | Std Error |
|---|---|---|
| pass@1:1_samples | 0.807 | 0.004 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 04:43:51
Temperature: 0.6
Overlong samples: 1.1% (128 / 12032)
musr_generative
| Metric | Score | Std Error |
|---|---|---|
| extractive_match | 0.697 | 0.028 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 00:23:45
Temperature: 0.6
Overlong samples: 0.4% (3 / 756)
obqa_generative
| Metric | Score | Std Error |
|---|---|---|
| extractive_match | 0.966 | 0.008 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 00:17:21
Temperature: 0.6
Overlong samples: 0.2% (1 / 500)
simpleqa_nous
| Metric | Score | Std Error |
|---|---|---|
| exact_match | 0.039 | 0.003 |
| fuzzy_match | 0.060 | 0.004 |
Model: 36bpsychev2
Evaluation Time (hh:mm:ss): 01:27:06
Temperature: 0.6
Overlong samples: 8.6% (372 / 4321)
- Downloads last month
- 15
Size of downloaded dataset files:
616 MB
Size of the auto-converted Parquet files:
616 MB
Number of rows:
102,660