open-llm-leaderboard-old/details_bn22__OpenHermes-2.5-Mistral-7B-MISALIGNED

该数据集是在Open LLM Leaderboard上对模型bn22/OpenHermes-2.5-Mistral-7B-MISALIGNED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。数据集包含1次运行的结果，每次运行在每个配置中表示为特定的拆分。train拆分始终指向最新结果。此外，还有一个名为results的配置，存储了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。README还提供了如何使用Python中的datasets库加载运行细节的示例。

原始信息汇总

数据集概述

该数据集是在对模型 bn22/OpenHermes-2.5-Mistral-7B-MISALIGNED 进行评估运行期间自动创建的。数据集包含 63 个配置，每个配置对应一个评估任务。数据集是从 1 次运行中创建的，每个运行的详细信息可以在每个配置中找到，并以运行的时间戳命名的特定分片形式存储。"train" 分片始终指向最新的结果。

数据集结构

配置详情

• 配置名称: harness_arc_challenge_25

  • 数据文件:
    • 分片: 2023_12_27T13_23_06.286047
      • 路径: **/details_harness|arc:challenge|25_2023-12-27T13-23-06.286047.parquet
    • 分片: latest
      • 路径: **/details_harness|arc:challenge|25_2023-12-27T13-23-06.286047.parquet

• 配置名称: harness_gsm8k_5

  • 数据文件:
    • 分片: 2023_12_27T13_23_06.286047
      • 路径: **/details_harness|gsm8k|5_2023-12-27T13-23-06.286047.parquet
    • 分片: latest
      • 路径: **/details_harness|gsm8k|5_2023-12-27T13-23-06.286047.parquet

• 配置名称: harness_hellaswag_10

  • 数据文件:
    • 分片: 2023_12_27T13_23_06.286047
      • 路径: **/details_harness|hellaswag|10_2023-12-27T13-23-06.286047.parquet
    • 分片: latest
      • 路径: **/details_harness|hellaswag|10_2023-12-27T13-23-06.286047.parquet

• 配置名称: harness_hendrycksTest_5

  • 数据文件:
    • 分片: 2023_12_27T13_23_06.286047
      • 路径:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2023-12-27T13-23-06.286047.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2023-12-27T13-23-06.286047.parquet
        • `**/details_harness|hendrycksTest-high_