open-llm-leaderboard-old/details_42MARU__sitebunny-13b

该数据集是在模型42MARU/sitebunny-13b的评估运行期间自动创建的，用于Open LLM Leaderboard的评估。数据集由64个配置组成，每个配置对应一个评估任务。数据集由2次运行生成，每次运行的结果作为特定分割存储在配置中，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，还有一个名为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上的聚合指标。

原始信息汇总

数据集概述

该数据集是在对模型 42MARU/sitebunny-13b 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。
• 一个额外的配置 "results" 存储所有运行的聚合结果，用于计算和显示聚合指标在 Open LLM Leaderboard 上。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_42MARU__sitebunny-13b", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-23T17:55:45.857449 运行的最新结果：

python { "all": { "em": 0.3419672818791946, "em_stderr": 0.004857979450579366, "f1": 0.4322766359060419, "f1_stderr": 0.004628734048935794, "acc": 0.4305885746119642, "acc_stderr": 0.009958510446364506 }, "harness|drop|3": { "em": 0.3419672818791946, "em_stderr": 0.004857979450579366, "f1": 0.4322766359060419, "f1_stderr": 0.004628734048935794 }, "harness|gsm8k|5": { "acc": 0.09401061410159212, "acc_stderr": 0.008038819818872469 }, "harness|winogrande|5": { "acc": 0.7671665351223362, "acc_stderr": 0.011878201073856544 } }

配置详情

• harness_arc_challenge_25

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|arc:challenge|25_2023-10-03T14-18-14.630504.parquet

• harness_drop_3

  • 分割: 2023_10_23T17_55_45.857449, latest
  • 路径: **/details_harness|drop|3_2023-10-23T17-55-45.857449.parquet

• harness_gsm8k_5

  • 分割: 2023_10_23T17_55_45.857449, latest
  • 路径: **/details_harness|gsm8k|5_2023-10-23T17-55-45.857449.parquet

• harness_hellaswag_10

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hellaswag|10_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: 多个路径，例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_abstract_algebra_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_anatomy_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_astronomy_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_business_ethics_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_clinical_knowledge_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_college_biology_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_college_chemistry_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_college_computer_science_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_college_mathematics_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_college_medicine_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_college_physics_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_computer_security_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_conceptual_physics_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_econometrics_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_electrical_engineering_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-10-03T14-18-14.630504.parquet

• harness_hendrycksTest_elementary_mathematics_5

  • 分割: 2023_10_03T14_18_14.630504, latest
  • 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-10-03T14-18-14.630504.parquet