Update README.md
Browse files
README.md
CHANGED
|
@@ -1,129 +1,129 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: llama3
|
| 3 |
-
inference: false
|
| 4 |
-
---
|
| 5 |
-
|
| 6 |
-
# Description
|
| 7 |
-
4 bit quantization of [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) using GPTQ. We use the config below for quantization/evaluation and [HuggingFaceH4/ultrachat_200k](https://huggingface.co/datasets/HuggingFaceH4/ultrachat_200k)as the calibration data. The code is available under [this repository.](https://github.com/IST-DASLab/marlin/tree/2f6d7c10e124b3c5fa29ff8d77d568bd7af3274c/gptq)
|
| 8 |
-
|
| 9 |
-
```yaml
|
| 10 |
-
bits: 4
|
| 11 |
-
damp_percent: 0.01
|
| 12 |
-
desc_act: true
|
| 13 |
-
exllama_config:
|
| 14 |
-
version: 2
|
| 15 |
-
group_size: 128
|
| 16 |
-
quant_method: gptq
|
| 17 |
-
static_groups: false
|
| 18 |
-
sym: true
|
| 19 |
-
true_sequential: true
|
| 20 |
-
```
|
| 21 |
-
|
| 22 |
-
## Evaluations
|
| 23 |
-
|
| 24 |
-
Below is a comprehensive evaluation and also comparison with [casperhansen/llama-3-8b-instruct-awq](https://huggingface.co/casperhansen/llama-3-8b-instruct-awq) using the awesome [mosaicml/llm-foundry](https://github.com/mosaicml/llm-foundry/tree/main/scripts/eval).
|
| 25 |
-
|
| 26 |
-
| model_name | core_average | world_knowledge | commonsense_reasoning | language_understanding | symbolic_problem_solving | reading_comprehension |
|
| 27 |
-
| :---------------------------------------- | -----------: | --------------: | --------------------: | ---------------------: | -----------------------: | --------------------: |
|
| 28 |
-
| ISTA-DASLab/Llama-3-8B-Instruct-GPTQ-4bit | 0.552944 | 0.584061 | 0.547598 | 0.663904 | 0.431017 | 0.538141 |
|
| 29 |
-
| casperhansen/llama-3-8b-instruct-awq | 0.531504 | 0.557663 | 0.528201 | 0.657211 | 0.391476 | 0.522971 |
|
| 30 |
-
|
| 31 |
-
| Category | Benchmark | Subtask | Accuracy GPTQ | Accuracy AWQ | Number few shot |
|
| 32 |
-
| :----------------------- | :--------------------------- | :---------------------------------- | ------------: | -----------: | :-------------- |
|
| 33 |
-
| symbolic_problem_solving | gsm8k | | 0.721759 | 0.59818 | 0-shot |
|
| 34 |
-
| commonsense_reasoning | copa | | 0.85 | 0.84 | 0-shot |
|
| 35 |
-
| commonsense_reasoning | commonsense_qa | | 0.78706 | 0.782146 | 0-shot |
|
| 36 |
-
| commonsense_reasoning | piqa | | 0.784004 | 0.781828 | 0-shot |
|
| 37 |
-
| commonsense_reasoning | bigbench_strange_stories | | 0.764368 | 0.752874 | 0-shot |
|
| 38 |
-
| commonsense_reasoning | bigbench_strategy_qa | | 0.680647 | 0.659677 | 0-shot |
|
| 39 |
-
| language_understanding | lambada_openai | | 0.716476 | 0.717834 | 0-shot |
|
| 40 |
-
| language_understanding | hellaswag | | 0.750647 | 0.753137 | 0-shot |
|
| 41 |
-
| reading_comprehension | coqa | | 0.198797 | 0.109733 | 0-shot |
|
| 42 |
-
| reading_comprehension | boolq | | 0.8263 | 0.836391 | 0-shot |
|
| 43 |
-
| world_knowledge | triviaqa_sm_sub | | 0.590667 | 0.511333 | 3-shot |
|
| 44 |
-
| world_knowledge | jeopardy | Average | 0.4975 | 0.489451 | 3-shot |
|
| 45 |
-
| world_knowledge | | american_history | 0.535109 | 0.544794 | 3-shot |
|
| 46 |
-
| world_knowledge | | literature | 0.622449 | 0.626531 | 3-shot |
|
| 47 |
-
| world_knowledge | | science | 0.420168 | 0.390756 | 3-shot |
|
| 48 |
-
| world_knowledge | | word_origins | 0.293151 | 0.271233 | 3-shot |
|
| 49 |
-
| world_knowledge | | world_history | 0.616622 | 0.613941 | 3-shot |
|
| 50 |
-
| world_knowledge | bigbench_qa_wikidata | | 0.684366 | 0.644358 | 3-shot |
|
| 51 |
-
| world_knowledge | arc_easy | | 0.808923 | 0.808081 | 3-shot |
|
| 52 |
-
| world_knowledge | arc_challenge | | 0.571672 | 0.571672 | 3-shot |
|
| 53 |
-
| commonsense_reasoning | siqa | | 0.827533 | 0.814227 | 3-shot |
|
| 54 |
-
| language_understanding | winograd | | 0.871795 | 0.860806 | 3-shot |
|
| 55 |
-
| symbolic_problem_solving | bigbench_operators | | 0.547619 | 0.552381 | 3-shot |
|
| 56 |
-
| reading_comprehension | squad | | 0.581552 | 0.58789 | 3-shot |
|
| 57 |
-
| symbolic_problem_solving | svamp | | 0.68 | 0.57 | 5-shot |
|
| 58 |
-
| world_knowledge | mmlu | Average | 0.668279 | 0.645874 | 5-shot |
|
| 59 |
-
| world_knowledge | | abstract_algebra | 0.29 | 0.33 | 5-shot |
|
| 60 |
-
| world_knowledge | | anatomy | 0.681481 | 0.651852 | 5-shot |
|
| 61 |
-
| world_knowledge | | astronomy | 0.703947 | 0.671053 | 5-shot |
|
| 62 |
-
| world_knowledge | | business_ethics | 0.67 | 0.68 | 5-shot |
|
| 63 |
-
| world_knowledge | | clinical_knowledge | 0.750943 | 0.701887 | 5-shot |
|
| 64 |
-
| world_knowledge | | college_biology | 0.784722 | 0.729167 | 5-shot |
|
| 65 |
-
| world_knowledge | | college_chemistry | 0.47 | 0.46 | 5-shot |
|
| 66 |
-
| world_knowledge | | college_computer_science | 0.56 | 0.54 | 5-shot |
|
| 67 |
-
| world_knowledge | | college_mathematics | 0.36 | 0.28 | 5-shot |
|
| 68 |
-
| world_knowledge | | college_medicine | 0.653179 | 0.635838 | 5-shot |
|
| 69 |
-
| world_knowledge | | college_physics | 0.5 | 0.431373 | 5-shot |
|
| 70 |
-
| world_knowledge | | computer_security | 0.78 | 0.75 | 5-shot |
|
| 71 |
-
| world_knowledge | | conceptual_physics | 0.548936 | 0.557447 | 5-shot |
|
| 72 |
-
| world_knowledge | | econometrics | 0.45614 | 0.482456 | 5-shot |
|
| 73 |
-
| world_knowledge | | electrical_engineering | 0.668966 | 0.586207 | 5-shot |
|
| 74 |
-
| world_knowledge | | elementary_mathematics | 0.439153 | 0.417989 | 5-shot |
|
| 75 |
-
| world_knowledge | | formal_logic | 0.47619 | 0.412698 | 5-shot |
|
| 76 |
-
| world_knowledge | | global_facts | 0.37 | 0.41 | 5-shot |
|
| 77 |
-
| world_knowledge | | high_school_biology | 0.790323 | 0.754839 | 5-shot |
|
| 78 |
-
| world_knowledge | | high_school_chemistry | 0.581281 | 0.507389 | 5-shot |
|
| 79 |
-
| world_knowledge | | high_school_computer_science | 0.71 | 0.74 | 5-shot |
|
| 80 |
-
| world_knowledge | | high_school_european_history | 0.745455 | 0.775758 | 5-shot |
|
| 81 |
-
| world_knowledge | | high_school_geography | 0.823232 | 0.823232 | 5-shot |
|
| 82 |
-
| world_knowledge | | high_school_government_and_politics | 0.917098 | 0.875648 | 5-shot |
|
| 83 |
-
| world_knowledge | | high_school_macroeconomics | 0.635897 | 0.620513 | 5-shot |
|
| 84 |
-
| world_knowledge | | high_school_mathematics | 0.407407 | 0.392593 | 5-shot |
|
| 85 |
-
| world_knowledge | | high_school_microeconomics | 0.726891 | 0.714286 | 5-shot |
|
| 86 |
-
| world_knowledge | | high_school_physics | 0.423841 | 0.410596 | 5-shot |
|
| 87 |
-
| world_knowledge | | high_school_psychology | 0.842202 | 0.838532 | 5-shot |
|
| 88 |
-
| world_knowledge | | high_school_statistics | 0.592593 | 0.513889 | 5-shot |
|
| 89 |
-
| world_knowledge | | high_school_us_history | 0.852941 | 0.852941 | 5-shot |
|
| 90 |
-
| world_knowledge | | high_school_world_history | 0.843882 | 0.831224 | 5-shot |
|
| 91 |
-
| world_knowledge | | human_aging | 0.717489 | 0.713004 | 5-shot |
|
| 92 |
-
| world_knowledge | | human_sexuality | 0.763359 | 0.70229 | 5-shot |
|
| 93 |
-
| world_knowledge | | international_law | 0.793388 | 0.77686 | 5-shot |
|
| 94 |
-
| world_knowledge | | jurisprudence | 0.814815 | 0.768519 | 5-shot |
|
| 95 |
-
| world_knowledge | | logical_fallacies | 0.754601 | 0.773006 | 5-shot |
|
| 96 |
-
| world_knowledge | | machine_learning | 0.553571 | 0.508929 | 5-shot |
|
| 97 |
-
| world_knowledge | | management | 0.84466 | 0.834951 | 5-shot |
|
| 98 |
-
| world_knowledge | | marketing | 0.92735 | 0.888889 | 5-shot |
|
| 99 |
-
| world_knowledge | | medical_genetics | 0.81 | 0.78 | 5-shot |
|
| 100 |
-
| world_knowledge | | miscellaneous | 0.825032 | 0.799489 | 5-shot |
|
| 101 |
-
| world_knowledge | | moral_disputes | 0.739884 | 0.722543 | 5-shot |
|
| 102 |
-
| world_knowledge | | moral_scenarios | 0.437989 | 0.38324 | 5-shot |
|
| 103 |
-
| world_knowledge | | nutrition | 0.764706 | 0.735294 | 5-shot |
|
| 104 |
-
| world_knowledge | | philosophy | 0.733119 | 0.713826 | 5-shot |
|
| 105 |
-
| world_knowledge | | prehistory | 0.719136 | 0.719136 | 5-shot |
|
| 106 |
-
| world_knowledge | | professional_accounting | 0.475177 | 0.485816 | 5-shot |
|
| 107 |
-
| world_knowledge | | professional_law | 0.480443 | 0.449153 | 5-shot |
|
| 108 |
-
| world_knowledge | | professional_medicine | 0.709559 | 0.676471 | 5-shot |
|
| 109 |
-
| world_knowledge | | professional_psychology | 0.694444 | 0.676471 | 5-shot |
|
| 110 |
-
| world_knowledge | | public_relations | 0.7 | 0.6 | 5-shot |
|
| 111 |
-
| world_knowledge | | security_studies | 0.730612 | 0.718367 | 5-shot |
|
| 112 |
-
| world_knowledge | | sociology | 0.830846 | 0.845771 | 5-shot |
|
| 113 |
-
| world_knowledge | | us_foreign_policy | 0.86 | 0.85 | 5-shot |
|
| 114 |
-
| world_knowledge | | virology | 0.542169 | 0.518072 | 5-shot |
|
| 115 |
-
| world_knowledge | | world_religions | 0.812865 | 0.795322 | 5-shot |
|
| 116 |
-
| symbolic_problem_solving | bigbench_dyck_languages | | 0.086 | 0.045 | 5-shot |
|
| 117 |
-
| language_understanding | winogrande | | 0.764009 | 0.759274 | 5-shot |
|
| 118 |
-
| symbolic_problem_solving | agi_eval_lsat_ar | | 0.3 | 0.278261 | 5-shot |
|
| 119 |
-
| symbolic_problem_solving | simple_arithmetic_nospaces | | 0.466 | 0.458 | 5-shot |
|
| 120 |
-
| symbolic_problem_solving | simple_arithmetic_withspaces | | 0.502 | 0.496 | 5-shot |
|
| 121 |
-
| reading_comprehension | agi_eval_lsat_rc | | 0.731343 | 0.708955 | 5-shot |
|
| 122 |
-
| reading_comprehension | agi_eval_lsat_lr | | 0.554902 | 0.560784 | 5-shot |
|
| 123 |
-
| reading_comprehension | agi_eval_sat_en | | 0.81068 | 0.805825 | 5-shot |
|
| 124 |
-
| world_knowledge | arc_challenge | | 0.582765 | 0.591297 | 25-shot |
|
| 125 |
-
| commonsense_reasoning | openbook_qa | | 0.478 | 0.468 | 10-shot |
|
| 126 |
-
| language_understanding | hellaswag | | 0.769468 | 0.771062 | 10-shot |
|
| 127 |
-
| | bigbench_cs_algorithms | | 0.715151 | 0.687879 | 10-shot |
|
| 128 |
-
| symbolic_problem_solving | bigbench_elementary_math_qa | | 0.533569 | 0.530922 | 1-shot |
|
| 129 |
-
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: llama3
|
| 3 |
+
inference: false
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
# Description
|
| 7 |
+
4 bit quantization of [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) using GPTQ. We use the config below for quantization/evaluation and [HuggingFaceH4/ultrachat_200k](https://huggingface.co/datasets/HuggingFaceH4/ultrachat_200k) as the calibration data. The code is available under [this repository.](https://github.com/IST-DASLab/marlin/tree/2f6d7c10e124b3c5fa29ff8d77d568bd7af3274c/gptq)
|
| 8 |
+
|
| 9 |
+
```yaml
|
| 10 |
+
bits: 4
|
| 11 |
+
damp_percent: 0.01
|
| 12 |
+
desc_act: true
|
| 13 |
+
exllama_config:
|
| 14 |
+
version: 2
|
| 15 |
+
group_size: 128
|
| 16 |
+
quant_method: gptq
|
| 17 |
+
static_groups: false
|
| 18 |
+
sym: true
|
| 19 |
+
true_sequential: true
|
| 20 |
+
```
|
| 21 |
+
|
| 22 |
+
## Evaluations
|
| 23 |
+
|
| 24 |
+
Below is a comprehensive evaluation and also comparison with [casperhansen/llama-3-8b-instruct-awq](https://huggingface.co/casperhansen/llama-3-8b-instruct-awq) using the awesome [mosaicml/llm-foundry](https://github.com/mosaicml/llm-foundry/tree/main/scripts/eval).
|
| 25 |
+
|
| 26 |
+
| model_name | core_average | world_knowledge | commonsense_reasoning | language_understanding | symbolic_problem_solving | reading_comprehension |
|
| 27 |
+
| :---------------------------------------- | -----------: | --------------: | --------------------: | ---------------------: | -----------------------: | --------------------: |
|
| 28 |
+
| ISTA-DASLab/Llama-3-8B-Instruct-GPTQ-4bit | 0.552944 | 0.584061 | 0.547598 | 0.663904 | 0.431017 | 0.538141 |
|
| 29 |
+
| casperhansen/llama-3-8b-instruct-awq | 0.531504 | 0.557663 | 0.528201 | 0.657211 | 0.391476 | 0.522971 |
|
| 30 |
+
|
| 31 |
+
| Category | Benchmark | Subtask | Accuracy GPTQ | Accuracy AWQ | Number few shot |
|
| 32 |
+
| :----------------------- | :--------------------------- | :---------------------------------- | ------------: | -----------: | :-------------- |
|
| 33 |
+
| symbolic_problem_solving | gsm8k | | 0.721759 | 0.59818 | 0-shot |
|
| 34 |
+
| commonsense_reasoning | copa | | 0.85 | 0.84 | 0-shot |
|
| 35 |
+
| commonsense_reasoning | commonsense_qa | | 0.78706 | 0.782146 | 0-shot |
|
| 36 |
+
| commonsense_reasoning | piqa | | 0.784004 | 0.781828 | 0-shot |
|
| 37 |
+
| commonsense_reasoning | bigbench_strange_stories | | 0.764368 | 0.752874 | 0-shot |
|
| 38 |
+
| commonsense_reasoning | bigbench_strategy_qa | | 0.680647 | 0.659677 | 0-shot |
|
| 39 |
+
| language_understanding | lambada_openai | | 0.716476 | 0.717834 | 0-shot |
|
| 40 |
+
| language_understanding | hellaswag | | 0.750647 | 0.753137 | 0-shot |
|
| 41 |
+
| reading_comprehension | coqa | | 0.198797 | 0.109733 | 0-shot |
|
| 42 |
+
| reading_comprehension | boolq | | 0.8263 | 0.836391 | 0-shot |
|
| 43 |
+
| world_knowledge | triviaqa_sm_sub | | 0.590667 | 0.511333 | 3-shot |
|
| 44 |
+
| world_knowledge | jeopardy | Average | 0.4975 | 0.489451 | 3-shot |
|
| 45 |
+
| world_knowledge | | american_history | 0.535109 | 0.544794 | 3-shot |
|
| 46 |
+
| world_knowledge | | literature | 0.622449 | 0.626531 | 3-shot |
|
| 47 |
+
| world_knowledge | | science | 0.420168 | 0.390756 | 3-shot |
|
| 48 |
+
| world_knowledge | | word_origins | 0.293151 | 0.271233 | 3-shot |
|
| 49 |
+
| world_knowledge | | world_history | 0.616622 | 0.613941 | 3-shot |
|
| 50 |
+
| world_knowledge | bigbench_qa_wikidata | | 0.684366 | 0.644358 | 3-shot |
|
| 51 |
+
| world_knowledge | arc_easy | | 0.808923 | 0.808081 | 3-shot |
|
| 52 |
+
| world_knowledge | arc_challenge | | 0.571672 | 0.571672 | 3-shot |
|
| 53 |
+
| commonsense_reasoning | siqa | | 0.827533 | 0.814227 | 3-shot |
|
| 54 |
+
| language_understanding | winograd | | 0.871795 | 0.860806 | 3-shot |
|
| 55 |
+
| symbolic_problem_solving | bigbench_operators | | 0.547619 | 0.552381 | 3-shot |
|
| 56 |
+
| reading_comprehension | squad | | 0.581552 | 0.58789 | 3-shot |
|
| 57 |
+
| symbolic_problem_solving | svamp | | 0.68 | 0.57 | 5-shot |
|
| 58 |
+
| world_knowledge | mmlu | Average | 0.668279 | 0.645874 | 5-shot |
|
| 59 |
+
| world_knowledge | | abstract_algebra | 0.29 | 0.33 | 5-shot |
|
| 60 |
+
| world_knowledge | | anatomy | 0.681481 | 0.651852 | 5-shot |
|
| 61 |
+
| world_knowledge | | astronomy | 0.703947 | 0.671053 | 5-shot |
|
| 62 |
+
| world_knowledge | | business_ethics | 0.67 | 0.68 | 5-shot |
|
| 63 |
+
| world_knowledge | | clinical_knowledge | 0.750943 | 0.701887 | 5-shot |
|
| 64 |
+
| world_knowledge | | college_biology | 0.784722 | 0.729167 | 5-shot |
|
| 65 |
+
| world_knowledge | | college_chemistry | 0.47 | 0.46 | 5-shot |
|
| 66 |
+
| world_knowledge | | college_computer_science | 0.56 | 0.54 | 5-shot |
|
| 67 |
+
| world_knowledge | | college_mathematics | 0.36 | 0.28 | 5-shot |
|
| 68 |
+
| world_knowledge | | college_medicine | 0.653179 | 0.635838 | 5-shot |
|
| 69 |
+
| world_knowledge | | college_physics | 0.5 | 0.431373 | 5-shot |
|
| 70 |
+
| world_knowledge | | computer_security | 0.78 | 0.75 | 5-shot |
|
| 71 |
+
| world_knowledge | | conceptual_physics | 0.548936 | 0.557447 | 5-shot |
|
| 72 |
+
| world_knowledge | | econometrics | 0.45614 | 0.482456 | 5-shot |
|
| 73 |
+
| world_knowledge | | electrical_engineering | 0.668966 | 0.586207 | 5-shot |
|
| 74 |
+
| world_knowledge | | elementary_mathematics | 0.439153 | 0.417989 | 5-shot |
|
| 75 |
+
| world_knowledge | | formal_logic | 0.47619 | 0.412698 | 5-shot |
|
| 76 |
+
| world_knowledge | | global_facts | 0.37 | 0.41 | 5-shot |
|
| 77 |
+
| world_knowledge | | high_school_biology | 0.790323 | 0.754839 | 5-shot |
|
| 78 |
+
| world_knowledge | | high_school_chemistry | 0.581281 | 0.507389 | 5-shot |
|
| 79 |
+
| world_knowledge | | high_school_computer_science | 0.71 | 0.74 | 5-shot |
|
| 80 |
+
| world_knowledge | | high_school_european_history | 0.745455 | 0.775758 | 5-shot |
|
| 81 |
+
| world_knowledge | | high_school_geography | 0.823232 | 0.823232 | 5-shot |
|
| 82 |
+
| world_knowledge | | high_school_government_and_politics | 0.917098 | 0.875648 | 5-shot |
|
| 83 |
+
| world_knowledge | | high_school_macroeconomics | 0.635897 | 0.620513 | 5-shot |
|
| 84 |
+
| world_knowledge | | high_school_mathematics | 0.407407 | 0.392593 | 5-shot |
|
| 85 |
+
| world_knowledge | | high_school_microeconomics | 0.726891 | 0.714286 | 5-shot |
|
| 86 |
+
| world_knowledge | | high_school_physics | 0.423841 | 0.410596 | 5-shot |
|
| 87 |
+
| world_knowledge | | high_school_psychology | 0.842202 | 0.838532 | 5-shot |
|
| 88 |
+
| world_knowledge | | high_school_statistics | 0.592593 | 0.513889 | 5-shot |
|
| 89 |
+
| world_knowledge | | high_school_us_history | 0.852941 | 0.852941 | 5-shot |
|
| 90 |
+
| world_knowledge | | high_school_world_history | 0.843882 | 0.831224 | 5-shot |
|
| 91 |
+
| world_knowledge | | human_aging | 0.717489 | 0.713004 | 5-shot |
|
| 92 |
+
| world_knowledge | | human_sexuality | 0.763359 | 0.70229 | 5-shot |
|
| 93 |
+
| world_knowledge | | international_law | 0.793388 | 0.77686 | 5-shot |
|
| 94 |
+
| world_knowledge | | jurisprudence | 0.814815 | 0.768519 | 5-shot |
|
| 95 |
+
| world_knowledge | | logical_fallacies | 0.754601 | 0.773006 | 5-shot |
|
| 96 |
+
| world_knowledge | | machine_learning | 0.553571 | 0.508929 | 5-shot |
|
| 97 |
+
| world_knowledge | | management | 0.84466 | 0.834951 | 5-shot |
|
| 98 |
+
| world_knowledge | | marketing | 0.92735 | 0.888889 | 5-shot |
|
| 99 |
+
| world_knowledge | | medical_genetics | 0.81 | 0.78 | 5-shot |
|
| 100 |
+
| world_knowledge | | miscellaneous | 0.825032 | 0.799489 | 5-shot |
|
| 101 |
+
| world_knowledge | | moral_disputes | 0.739884 | 0.722543 | 5-shot |
|
| 102 |
+
| world_knowledge | | moral_scenarios | 0.437989 | 0.38324 | 5-shot |
|
| 103 |
+
| world_knowledge | | nutrition | 0.764706 | 0.735294 | 5-shot |
|
| 104 |
+
| world_knowledge | | philosophy | 0.733119 | 0.713826 | 5-shot |
|
| 105 |
+
| world_knowledge | | prehistory | 0.719136 | 0.719136 | 5-shot |
|
| 106 |
+
| world_knowledge | | professional_accounting | 0.475177 | 0.485816 | 5-shot |
|
| 107 |
+
| world_knowledge | | professional_law | 0.480443 | 0.449153 | 5-shot |
|
| 108 |
+
| world_knowledge | | professional_medicine | 0.709559 | 0.676471 | 5-shot |
|
| 109 |
+
| world_knowledge | | professional_psychology | 0.694444 | 0.676471 | 5-shot |
|
| 110 |
+
| world_knowledge | | public_relations | 0.7 | 0.6 | 5-shot |
|
| 111 |
+
| world_knowledge | | security_studies | 0.730612 | 0.718367 | 5-shot |
|
| 112 |
+
| world_knowledge | | sociology | 0.830846 | 0.845771 | 5-shot |
|
| 113 |
+
| world_knowledge | | us_foreign_policy | 0.86 | 0.85 | 5-shot |
|
| 114 |
+
| world_knowledge | | virology | 0.542169 | 0.518072 | 5-shot |
|
| 115 |
+
| world_knowledge | | world_religions | 0.812865 | 0.795322 | 5-shot |
|
| 116 |
+
| symbolic_problem_solving | bigbench_dyck_languages | | 0.086 | 0.045 | 5-shot |
|
| 117 |
+
| language_understanding | winogrande | | 0.764009 | 0.759274 | 5-shot |
|
| 118 |
+
| symbolic_problem_solving | agi_eval_lsat_ar | | 0.3 | 0.278261 | 5-shot |
|
| 119 |
+
| symbolic_problem_solving | simple_arithmetic_nospaces | | 0.466 | 0.458 | 5-shot |
|
| 120 |
+
| symbolic_problem_solving | simple_arithmetic_withspaces | | 0.502 | 0.496 | 5-shot |
|
| 121 |
+
| reading_comprehension | agi_eval_lsat_rc | | 0.731343 | 0.708955 | 5-shot |
|
| 122 |
+
| reading_comprehension | agi_eval_lsat_lr | | 0.554902 | 0.560784 | 5-shot |
|
| 123 |
+
| reading_comprehension | agi_eval_sat_en | | 0.81068 | 0.805825 | 5-shot |
|
| 124 |
+
| world_knowledge | arc_challenge | | 0.582765 | 0.591297 | 25-shot |
|
| 125 |
+
| commonsense_reasoning | openbook_qa | | 0.478 | 0.468 | 10-shot |
|
| 126 |
+
| language_understanding | hellaswag | | 0.769468 | 0.771062 | 10-shot |
|
| 127 |
+
| | bigbench_cs_algorithms | | 0.715151 | 0.687879 | 10-shot |
|
| 128 |
+
| symbolic_problem_solving | bigbench_elementary_math_qa | | 0.533569 | 0.530922 | 1-shot |
|
| 129 |
+
|