open-llm-leaderboard-old/details_uukuguy__speechless-orca-platypus-coig-lite-2k-0.6e-13b

该数据集是在模型[uukuguy/speechless-orca-platypus-coig-lite-2k-0.6e-13b](https://huggingface.co/uukuguy/speechless-orca-platypus-coig-lite-2k-0.6e-13b)在[Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)上的评估运行期间自动创建的。数据集由64个配置组成，每个配置对应一个评估任务。数据集是从2次运行中创建的，每次运行都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。此外，一个名为"results"的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](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)上的聚合指标。

原始信息汇总

数据集概述

该数据集是在评估模型 uukuguy/speechless-orca-platypus-coig-lite-2k-0.6e-13b 的过程中自动创建的。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。

数据集结构

配置详情

• config_name: harness_arc_challenge_25

  • data_files:
    • split: 2023_08_30T17_54_51.197545
      • path: **/details_harness|arc:challenge|25_2023-08-30T17:54:51.197545.parquet
    • split: latest
      • path: **/details_harness|arc:challenge|25_2023-08-30T17:54:51.197545.parquet

• config_name: harness_drop_3

  • data_files:
    • split: 2023_10_19T07_05_10.941569
      • path: **/details_harness|drop|3_2023-10-19T07-05-10.941569.parquet
    • split: latest
      • path: **/details_harness|drop|3_2023-10-19T07-05-10.941569.parquet

• config_name: harness_gsm8k_5

  • data_files:
    • split: 2023_10_19T07_05_10.941569
      • path: **/details_harness|gsm8k|5_2023-10-19T07-05-10.941569.parquet
    • split: latest
      • path: **/details_harness|gsm8k|5_2023-10-19T07-05-10.941569.parquet

• config_name: harness_hellaswag_10

  • data_files:
    • split: 2023_08_30T17_54_51.197545
      • path: **/details_harness|hellaswag|10_2023-08-30T17:54:51.197545.parquet
    • split: latest
      • path: **/details_harness|hellaswag|10_2023-08-30T17:54:51.197545.parquet

• config_name: harness_hendrycksTest_5

  • data_files:
    • split: 2023_08_30T17_54_51.197545
      • path:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-30T17:54:51.197545.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2023-08-30T17:54:51.197545.parquet
        • ... (其他路径省略)
    • split: latest
      • path:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-30T17:54:51.197545.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2023-08-30T17:54:51.197545.parquet
        • ... (其他路径省略)

结果配置

• config_name: results
  • description: 存储所有运行结果的聚合结果，用于计算和显示在 Open LLM Leaderboard 上的聚合指标。

最新结果

以下是 2023-10-19T07:05:10.941569 运行的最新结果：

python { "all": { "em": 0.40740352348993286, "em_stderr": 0.005031895046041088, "f1": 0.46848573825503476, "f1_stderr": 0.004828956071080464, "acc": 0.4270311930602038, "acc_stderr": 0.00945942839156503 }, "harness|drop|3": { "em": 0.40740352348993286, "em_stderr": 0.005031895046041088, "f1": 0.46848573825503476, "f1_stderr": 0.004828956071080464 }, "harness|gsm8k|5": { "acc": 0.07505686125852919, "acc_stderr": 0.007257633145486643 }, "harness|winogrande|5": { "acc": 0.7790055248618785, "acc_stderr": 0.011661223637643416 } }