遇见数据集

open-llm-leaderboard/details_totally-not-an-llm__PuddleJumper-13b

收藏
Hugging Face2023-09-17 更新2024-03-04 收录
官方服务:

资源简介:

--- pretty_name: Evaluation run of totally-not-an-llm/PuddleJumper-13b dataset_summary: "Dataset automatically created during the evaluation run of model\ \ [totally-not-an-llm/PuddleJumper-13b](https://huggingface.co/totally-not-an-llm/PuddleJumper-13b)\ \ on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).\n\ \nThe dataset is composed of 64 configuration, each one coresponding to one of the\ \ evaluated task.\n\nThe dataset has been created from 2 run(s). Each run can be\ \ found as a specific split in each configuration, the split being named using the\ \ timestamp of the run.The \"train\" split is always pointing to the latest results.\n\ \nAn additional configuration \"results\" store all the aggregated results of the\ \ run (and is used to compute and display the agregated metrics on the [Open LLM\ \ Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)).\n\ \nTo load the details from a run, you can for instance do the following:\n```python\n\ from datasets import load_dataset\ndata = load_dataset(\"open-llm-leaderboard/details_totally-not-an-llm__PuddleJumper-13b\"\ ,\n\t\"harness_winogrande_5\",\n\tsplit=\"train\")\n```\n\n## Latest results\n\n\ These are the [latest results from run 2023-09-17T13:22:43.977787](https://huggingface.co/datasets/open-llm-leaderboard/details_totally-not-an-llm__PuddleJumper-13b/blob/main/results_2023-09-17T13-22-43.977787.json)(note\ \ that their might be results for other tasks in the repos if successive evals didn't\ \ cover the same tasks. You find each in the results and the \"latest\" split for\ \ each eval):\n\n```python\n{\n \"all\": {\n \"em\": 0.08452181208053691,\n\ \ \"em_stderr\": 0.002848708763936303,\n \"f1\": 0.20933095637583904,\n\ \ \"f1_stderr\": 0.003279820666133777,\n \"acc\": 0.38053092049715975,\n\ \ \"acc_stderr\": 0.008728490320313857\n },\n \"harness|drop|3\": {\n\ \ \"em\": 0.08452181208053691,\n \"em_stderr\": 0.002848708763936303,\n\ \ \"f1\": 0.20933095637583904,\n \"f1_stderr\": 0.003279820666133777\n\ \ },\n \"harness|gsm8k|5\": {\n \"acc\": 0.03335860500379075,\n \ \ \"acc_stderr\": 0.004946282649173775\n },\n \"harness|winogrande|5\"\ : {\n \"acc\": 0.7277032359905288,\n \"acc_stderr\": 0.012510697991453937\n\ \ }\n}\n```" repo_url: https://huggingface.co/totally-not-an-llm/PuddleJumper-13b leaderboard_url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard point_of_contact: clementine@hf.co configs: - config_name: harness_arc_challenge_25 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|arc:challenge|25_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|arc:challenge|25_2023-08-23T00:36:46.680857.parquet' - config_name: harness_drop_3 data_files: - split: 2023_09_17T13_22_43.977787 path: - '**/details_harness|drop|3_2023-09-17T13-22-43.977787.parquet' - split: latest path: - '**/details_harness|drop|3_2023-09-17T13-22-43.977787.parquet' - config_name: harness_gsm8k_5 data_files: - split: 2023_09_17T13_22_43.977787 path: - '**/details_harness|gsm8k|5_2023-09-17T13-22-43.977787.parquet' - split: latest path: - '**/details_harness|gsm8k|5_2023-09-17T13-22-43.977787.parquet' - config_name: harness_hellaswag_10 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hellaswag|10_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hellaswag|10_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-anatomy|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-astronomy|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-business_ethics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_biology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_chemistry|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_computer_science|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_mathematics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_medicine|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_physics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-computer_security|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-econometrics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-formal_logic|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-global_facts|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_biology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_chemistry|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_computer_science|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_european_history|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_geography|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_mathematics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_physics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_psychology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_statistics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_us_history|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_world_history|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-human_aging|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-human_sexuality|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-international_law|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-jurisprudence|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-logical_fallacies|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-machine_learning|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-management|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-marketing|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-medical_genetics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-miscellaneous|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-moral_disputes|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-moral_scenarios|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-nutrition|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-philosophy|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-prehistory|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-professional_accounting|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-professional_law|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-professional_medicine|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-professional_psychology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-public_relations|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-security_studies|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-sociology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-us_foreign_policy|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-virology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-world_religions|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-anatomy|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-astronomy|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-business_ethics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_biology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_chemistry|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_computer_science|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_mathematics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_medicine|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-college_physics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-computer_security|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-econometrics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-formal_logic|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-global_facts|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_biology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_chemistry|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_computer_science|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_european_history|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_geography|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_mathematics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_physics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_psychology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_statistics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_us_history|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-high_school_world_history|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-human_aging|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-human_sexuality|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-international_law|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-jurisprudence|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-logical_fallacies|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-machine_learning|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-management|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-marketing|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-medical_genetics|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-miscellaneous|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-moral_disputes|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-moral_scenarios|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-nutrition|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-philosophy|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-prehistory|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-professional_accounting|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-professional_law|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-professional_medicine|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-professional_psychology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-public_relations|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-security_studies|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-sociology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-us_foreign_policy|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-virology|5_2023-08-23T00:36:46.680857.parquet' - '**/details_harness|hendrycksTest-world_religions|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_abstract_algebra_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_anatomy_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-anatomy|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-anatomy|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_astronomy_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-astronomy|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-astronomy|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_business_ethics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-business_ethics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-business_ethics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_clinical_knowledge_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_college_biology_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-college_biology|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-college_biology|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_college_chemistry_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-college_chemistry|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-college_chemistry|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_college_computer_science_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-college_computer_science|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-college_computer_science|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_college_mathematics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-college_mathematics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-college_mathematics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_college_medicine_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-college_medicine|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-college_medicine|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_college_physics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-college_physics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-college_physics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_computer_security_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-computer_security|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-computer_security|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_conceptual_physics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_econometrics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-econometrics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-econometrics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_electrical_engineering_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_elementary_mathematics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_formal_logic_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-formal_logic|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-formal_logic|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_global_facts_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-global_facts|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-global_facts|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_biology_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_biology|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_biology|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_chemistry_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_chemistry|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_chemistry|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_computer_science_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_computer_science|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_computer_science|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_european_history_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_european_history|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_european_history|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_geography_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_geography|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_geography|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_government_and_politics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_macroeconomics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_mathematics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_mathematics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_mathematics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_microeconomics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_physics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_physics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_physics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_psychology_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_psychology|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_psychology|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_statistics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_statistics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_statistics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_us_history_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_us_history|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_us_history|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_high_school_world_history_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-high_school_world_history|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-high_school_world_history|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_human_aging_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-human_aging|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-human_aging|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_human_sexuality_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-human_sexuality|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-human_sexuality|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_international_law_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-international_law|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-international_law|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_jurisprudence_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-jurisprudence|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-jurisprudence|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_logical_fallacies_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-logical_fallacies|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-logical_fallacies|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_machine_learning_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-machine_learning|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-machine_learning|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_management_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-management|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-management|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_marketing_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-marketing|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-marketing|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_medical_genetics_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-medical_genetics|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-medical_genetics|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_miscellaneous_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-miscellaneous|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-miscellaneous|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_moral_disputes_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-moral_disputes|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-moral_disputes|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_moral_scenarios_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-moral_scenarios|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-moral_scenarios|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_nutrition_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-nutrition|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-nutrition|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_philosophy_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-philosophy|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-philosophy|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_prehistory_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-prehistory|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-prehistory|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_professional_accounting_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-professional_accounting|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-professional_accounting|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_professional_law_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-professional_law|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-professional_law|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_professional_medicine_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-professional_medicine|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-professional_medicine|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_professional_psychology_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-professional_psychology|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-professional_psychology|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_public_relations_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-public_relations|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-public_relations|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_security_studies_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-security_studies|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-security_studies|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_sociology_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-sociology|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-sociology|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_us_foreign_policy_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-us_foreign_policy|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-us_foreign_policy|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_virology_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-virology|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-virology|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_hendrycksTest_world_religions_5 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|hendrycksTest-world_religions|5_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|hendrycksTest-world_religions|5_2023-08-23T00:36:46.680857.parquet' - config_name: harness_truthfulqa_mc_0 data_files: - split: 2023_08_23T00_36_46.680857 path: - '**/details_harness|truthfulqa:mc|0_2023-08-23T00:36:46.680857.parquet' - split: latest path: - '**/details_harness|truthfulqa:mc|0_2023-08-23T00:36:46.680857.parquet' - config_name: harness_winogrande_5 data_files: - split: 2023_09_17T13_22_43.977787 path: - '**/details_harness|winogrande|5_2023-09-17T13-22-43.977787.parquet' - split: latest path: - '**/details_harness|winogrande|5_2023-09-17T13-22-43.977787.parquet' - config_name: results data_files: - split: 2023_09_17T13_22_43.977787 path: - results_2023-09-17T13-22-43.977787.parquet - split: latest path: - results_2023-09-17T13-22-43.977787.parquet --- # Dataset Card for Evaluation run of totally-not-an-llm/PuddleJumper-13b ## Dataset Description - **Homepage:** - **Repository:** https://huggingface.co/totally-not-an-llm/PuddleJumper-13b - **Paper:** - **Leaderboard:** https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard - **Point of Contact:** clementine@hf.co ### Dataset Summary Dataset automatically created during the evaluation run of model [totally-not-an-llm/PuddleJumper-13b](https://huggingface.co/totally-not-an-llm/PuddleJumper-13b) on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard). The dataset is composed of 64 configuration, each one coresponding to one of the evaluated task. The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results. An additional configuration "results" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)). To load the details from a run, you can for instance do the following: ```python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_totally-not-an-llm__PuddleJumper-13b", "harness_winogrande_5", split="train") ``` ## Latest results These are the [latest results from run 2023-09-17T13:22:43.977787](https://huggingface.co/datasets/open-llm-leaderboard/details_totally-not-an-llm__PuddleJumper-13b/blob/main/results_2023-09-17T13-22-43.977787.json)(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the "latest" split for each eval): ```python { "all": { "em": 0.08452181208053691, "em_stderr": 0.002848708763936303, "f1": 0.20933095637583904, "f1_stderr": 0.003279820666133777, "acc": 0.38053092049715975, "acc_stderr": 0.008728490320313857 }, "harness|drop|3": { "em": 0.08452181208053691, "em_stderr": 0.002848708763936303, "f1": 0.20933095637583904, "f1_stderr": 0.003279820666133777 }, "harness|gsm8k|5": { "acc": 0.03335860500379075, "acc_stderr": 0.004946282649173775 }, "harness|winogrande|5": { "acc": 0.7277032359905288, "acc_stderr": 0.012510697991453937 } } ``` ### Supported Tasks and Leaderboards [More Information Needed] ### Languages [More Information Needed] ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information [More Information Needed] ### Citation Information [More Information Needed] ### Contributions [More Information Needed]

提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

该数据集是在对模型 totally-not-an-llm/PuddleJumper-13b 进行评估运行时自动创建的,用于 Open LLM Leaderboard

数据集组成

  • 数据集包含 64 个配置,每个配置对应一个评估任务。
  • 数据集从 2 次运行中创建,每次运行在每个配置中作为一个特定的分片存在,分片名称使用运行的时间戳。
  • "train" 分片始终指向最新的结果。
  • 一个额外的配置 "results" 存储所有运行的聚合结果,用于计算和显示 Open LLM Leaderboard 上的聚合指标。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_totally-not-an-llm__PuddleJumper-13b", "harness_winogrande_5", split="train")

最新结果

以下是 2023-09-17T13:22:43.977787 运行的最新结果: python { "all": { "em": 0.08452181208053691, "em_stderr": 0.002848708763936303, "f1": 0.20933095637583904, "f1_stderr": 0.003279820666133777, "acc": 0.38053092049715975, "acc_stderr": 0.008728490320313857 }, "harness|drop|3": { "em": 0.08452181208053691, "em_stderr": 0.002848708763936303, "f1": 0.20933095637583904, "f1_stderr": 0.003279820666133777 }, "harness|gsm8k|5": { "acc": 0.03335860500379075, "acc_stderr": 0.004946282649173775 }, "harness|winogrande|5": { "acc": 0.7277032359905288, "acc_stderr": 0.012510697991453937 } }

配置详情

  • harness_arc_challenge_25

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|arc:challenge|25_2023-08-23T00:36:46.680857.parquet
  • harness_drop_3

    • 分片:2023_09_17T13_22_43.977787, latest
    • 路径:**/details_harness|drop|3_2023-09-17T13-22-43.977787.parquet
  • harness_gsm8k_5

    • 分片:2023_09_17T13_22_43.977787, latest
    • 路径:**/details_harness|gsm8k|5_2023-09-17T13-22-43.977787.parquet
  • harness_hellaswag_10

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hellaswag|10_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_anatomy_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_astronomy_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_college_biology_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_college_physics_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_computer_security_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_econometrics_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-23T00:36:46.680857.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分片:2023_08_23T00_36_46.680857, latest
    • 路径:**/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-23T00:36:46.680857.parquet

以上是数据集的详细概述,包括数据集的组成、加载示例、最新结果以及各个配置的详细信息。

搜集汇总
数据集介绍
open-llm-leaderboard/details_totally-not-an-llm__PuddleJumper-13b 数据集图片
构建方式
该数据集是在Open LLM Leaderboard对totally-not-an-llm/PuddleJumper-13b模型进行评估的过程中自动生成的。它由64个配置组成,每个配置对应一个被评估的任务。数据集源自两次独立的运行,每次运行的结果被存储为特定配置中的一个分割,分割名称以运行的时间戳命名。此外,还包含一个名为“results”的额外配置,用于存储所有运行的聚合结果,这些聚合指标被用于在排行榜上计算和展示模型的整体表现。
特点
数据集的核心特点在于其结构化的多任务评估记录。每个任务(如ARC挑战、DROP、GSM8K、Winogrande等)均有独立的配置,便于研究者针对特定能力进行细粒度分析。每次运行的结果都以时间戳分割的形式保存,而“train”分割始终指向最新运行的结果,确保了数据的时效性。这种设计不仅支持对模型性能的纵向追踪,还通过聚合配置提供了宏观的评估概览,使得模型在不同基准上的表现一目了然。
使用方法
使用该数据集时,可通过Hugging Face的datasets库按需加载特定任务的评估细节。例如,加载Winogrande任务的训练分割数据,只需指定配置名“harness_winogrande_5”和分割为“train”。研究者可以灵活选择不同时间戳的分割来对比历史结果,或直接使用“latest”分割获取最新数据。对于需要整体性能分析的情况,可以直接访问“results”配置,其中包含了如准确率、F1分数等聚合指标,便于快速评估模型在多个任务上的综合表现。
背景与挑战
背景概述
在大型语言模型(LLM)能力评估领域,标准化基准测试的构建与演进一直是推动模型性能透明化与可比较性的核心动力。Open LLM Leaderboard由Hugging Face团队于2023年发起,旨在通过统一的多任务评估框架,系统性地衡量开源LLM在推理、常识理解、数学求解及知识问答等维度的表现。该数据集记录了模型totally-not-an-llm/PuddleJumper-13b在Leaderboard上的两次评估运行,涵盖ARC、HellaSwag、MMLU、GSM8K、Winogrande及DROP等经典任务,反映了2023年8月至9月间对该13B参数规模模型的综合能力刻画。作为社区驱动的评测工具,该数据集不仅为模型开发者提供了可复现的细粒度性能快照,更通过结构化存储运行时间戳与结果聚合,促进了LLM领域横向比较的标准化进程,对开源生态中模型迭代与基准对齐具有重要参考价值。
当前挑战
该数据集所应对的核心领域挑战在于多任务评估的公平性与泛化性:LLM在不同任务间表现差异显著,如PuddleJumper-13b在Winogrande上达到72.77%的准确率,却在GSM8K数学推理中仅获3.34%,凸显了模型在语言理解与符号计算间的能力鸿沟,亟需更均衡的评测设计以揭示真实短板。构建过程中面临的技术挑战包括:评估结果对采样种子与few-shot配置的敏感性,例如DROP任务使用3-shot设置而GSM8K使用5-shot,导致跨任务比较存在偏差;多次运行间的时间戳管理要求严格的数据版本控制,以避免结果混淆;此外,Parquet格式的细粒度细节文件与聚合结果的同步更新,对数据管道的自动化与一致性提出了较高要求,尤其在处理57个MMLU子任务时,需确保配置路径的完整性与错误回溯的可行性。
常用场景
经典使用场景
该数据集作为Open LLM Leaderboard的评估结果记录,核心用途是对大规模语言模型PuddleJumper-13b在多个标准化基准任务上的性能进行系统性评测。其经典使用场景涵盖常识推理(如Winogrande)、数学推理(如GSM8K)、阅读理解(如DROP)以及多领域知识问答(如MMLU)等任务。研究者可通过加载该数据集中的特定配置和分割,复现模型在各项任务上的详尽指标,包括准确率、F1分数和精确匹配等,从而对模型能力形成全面而细致的认知。
衍生相关工作
该数据集衍生出的经典工作集中体现在Open LLM Leaderboard这一社区驱动的评估生态中。基于其标准化评估流程,涌现了大量关于模型性能比较、训练策略优化(如指令微调、强化学习)以及数据集偏差分析的研究。例如,研究者利用该数据集的细粒度结果分析模型在特定子任务上的失败模式,进而提出针对性的数据增强或架构改进方案,这些工作共同推动了开源大语言模型在鲁棒性与泛化能力上的持续进步。
数据集最近研究
最新研究方向
在当前大语言模型(LLM)能力评估的浪潮中,Open LLM Leaderboard 已成为衡量模型综合性能的标杆平台。围绕 PuddleJumper-13B 这一模型,该数据集通过自动化的多任务评估流程,系统性地记录了其在 ARC、DROP、GSM8K、Winogrande 及 MMLU 等 64 项基准测试中的表现,覆盖了从常识推理到数学解题、从阅读理解到知识问答的广泛维度。前沿研究正聚焦于如何利用这类细粒度评估数据,深入剖析模型在零样本与少样本场景下的泛化边界,特别是在 GSM8K 上仅 3.3% 的准确率揭示了复杂数学推理仍是当前 13B 级别模型的显著瓶颈。该数据集的意义在于为社区提供了可复现的、带时间戳的评估快照,推动了对模型能力演变轨迹的追踪,并助力研究者识别特定任务上的弱点,从而引导未来模型训练与微调策略的优化方向。
以上内容由遇见数据集搜集并总结生成
二维码
社区交流群
二维码
科研交流群
商业服务