five

open-llm-leaderboard-old/details_FredrikBL__test-dare

收藏
Hugging Face2024-03-29 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_FredrikBL__test-dare
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在Open LLM Leaderboard上对模型FredrikBL/test-dare进行评估时自动创建的。数据集由63个配置组成,每个配置对应一个评估任务。数据集从1次运行中创建,每次运行可以在每个配置的特定分割中找到,分割以运行的时间戳命名。train分割始终指向最新结果。此外,results配置存储了所有运行的聚合结果,并用于计算和显示Open LLM Leaderboard上的聚合指标。

该数据集是在Open LLM Leaderboard上对模型FredrikBL/test-dare进行评估时自动创建的。数据集由63个配置组成,每个配置对应一个评估任务。数据集从1次运行中创建,每次运行可以在每个配置的特定分割中找到,分割以运行的时间戳命名。train分割始终指向最新结果。此外,results配置存储了所有运行的聚合结果,并用于计算和显示Open LLM Leaderboard上的聚合指标。
提供机构:
open-llm-leaderboard-old
原始信息汇总

数据集概述

该数据集是在对模型 FredrikBL/test-dare 进行评估运行期间自动创建的,用于 Open LLM Leaderboard

数据集组成

  • 数据集由 63 个配置组成,每个配置对应一个评估任务。
  • 数据集从 1 次运行中创建。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。

最新结果

以下是 2024-03-29T21:19:13.154293 运行的最新结果

python { "all": { "acc": 0.6460331293814567, "acc_stderr": 0.03224922642756375, "acc_norm": 0.6478142381132488, "acc_norm_stderr": 0.032903872728238165, "mc1": 0.3598531211750306, "mc1_stderr": 0.01680186046667715, "mc2": 0.5268971269435854, "mc2_stderr": 0.015057816486907058 }, ... }

配置详情

  • config_name: harness_arc_challenge_25

    • split: 2024_03_29T21_19_13.154293
      • path: **/details_harness|arc:challenge|25_2024-03-29T21-19-13.154293.parquet
    • split: latest
      • path: **/details_harness|arc:challenge|25_2024-03-29T21-19-13.154293.parquet
  • config_name: harness_gsm8k_5

    • split: 2024_03_29T21_19_13.154293
      • path: **/details_harness|gsm8k|5_2024-03-29T21-19-13.154293.parquet
    • split: latest
      • path: **/details_harness|gsm8k|5_2024-03-29T21-19-13.154293.parquet
  • config_name: harness_hellaswag_10

    • split: 2024_03_29T21_19_13.154293
      • path: **/details_harness|hellaswag|10_2024-03-29T21-19-13.154293.parquet
    • split: latest
      • path: **/details_harness|hellaswag|10_2024-03-29T21-19-13.154293.parquet
  • config_name: harness_hendrycksTest_5

    • split: 2024_03_29T21_19_13.154293
      • path:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2024-03-29T21-19-13.154293.parquet
        • **/details_harness|hendrycksTest-high_school_biology|5_2024-03-29T21-19-13.154293.parquet
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作