open-llm-leaderboard-old/details_yeontaek__Platypus2xOpenOrca-13B-IA3-ensemble
收藏数据集概述
数据集简介
该数据集是在评估模型yeontaek/Platypus2xOpenOrca-13B-IA3-ensemble在Open LLM Leaderboard上的自动创建的。数据集包含60个配置,每个配置对应一个评估任务。
数据集结构
数据集由1次运行创建,每个运行可以在每个配置中找到特定的分割,分割名称使用运行的时间戳。"train"分割始终指向最新的结果。
额外配置
一个额外的配置"results"存储所有运行的聚合结果,用于计算和显示在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_yeontaek__Platypus2xOpenOrca-13B-IA3-ensemble", "harness_truthfulqa_mc_0", split="train")
最新结果
最新结果来自2023-08-23T14:48:47.168259运行,包含多个任务的评估结果,例如: python { "all": { "acc": 0.591031165151078, "acc_stderr": 0.033910448517384374, "acc_norm": 0.5951164394626702, "acc_norm_stderr": 0.033889044058760844, "mc1": 0.3317013463892289, "mc1_stderr": 0.016482148810241477, "mc2": 0.47461350738527963, "mc2_stderr": 0.015202805791318129 }, "harness|arc:challenge|25": { "acc": 0.5844709897610921, "acc_stderr": 0.014401366641216386, "acc_norm": 0.621160409556314, "acc_norm_stderr": 0.014175915490000326 }, "harness|hellaswag|10": { "acc": 0.6185022903804023, "acc_stderr": 0.004847615216473461, "acc_norm": 0.8228440549691296, "acc_norm_stderr": 0.0038102033089010925 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.32, "acc_stderr": 0.046882617226215034, "acc_norm": 0.32, "acc_norm_stderr": 0.046882617226215034 }, # 其他任务的结果... }
配置详情
数据集包含多个配置,每个配置对应不同的任务和数据文件路径。例如:
-
config_name: harness_arc_challenge_25split: 2023_08_23T14_48_47.168259path: **/details_harness|arc:challenge|25_2023-08-23T14:48:47.168259.parquet
split: latestpath: **/details_harness|arc:challenge|25_2023-08-23T14:48:47.168259.parquet
-
config_name: harness_hellaswag_10split: 2023_08_23T14_48_47.168259path: **/details_harness|hellaswag|10_2023-08-23T14:48:47.168259.parquet
split: latestpath: **/details_harness|hellaswag|10_2023-08-23T14:48:47.168259.parquet
-
config_name: harness_hendrycksTest_5split: 2023_08_23T14_48_47.168259path: **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-anatomy|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-astronomy|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-business_ethics|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-college_biology|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-college_chemistry|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-college_computer_science|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-college_mathematics|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-college_medicine|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-college_physics|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-computer_security|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-conceptual_physics|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-econometrics|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-electrical_engineering|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-formal_logic|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-global_facts|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-high_school_biology|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-high_school_chemistry|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-high_school_computer_science|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-high_school_european_history|5_2023-08-23T14:48:47.168259.parquetpath: **/details_harness|hendrycksTest-high_school_geography|5_2023-08-23T14:48:47.168259.parquet
其他路径...
}



