five

open-llm-leaderboard/details_dvruette__gpt-neox-20b-full-precision

收藏
Hugging Face2023-10-15 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_dvruette__gpt-neox-20b-full-precision
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在评估模型dvruette/gpt-neox-20b-full-precision时自动创建的,主要用于Open LLM Leaderboard的评估任务。数据集包含64个配置,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一个特定的分割,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,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。
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

该数据集是在对模型 dvruette/gpt-neox-20b-full-precision 进行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

数据集包含以下配置及其对应的数据文件:

  • harness_arc_challenge_25

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|arc:challenge|25_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|arc:challenge|25_2023-07-19T21:36:17.720122.parquet
  • harness_drop_3

    • 分割: 2023_10_15T10_49_53.793437
      • 路径: **/details_harness|drop|3_2023-10-15T10-49-53.793437.parquet
    • 分割: latest
      • 路径: **/details_harness|drop|3_2023-10-15T10-49-53.793437.parquet
  • harness_gsm8k_5

    • 分割: 2023_10_15T10_49_53.793437
      • 路径: **/details_harness|gsm8k|5_2023-10-15T10-49-53.793437.parquet
    • 分割: latest
      • 路径: **/details_harness|gsm8k|5_2023-10-15T10-49-53.793437.parquet
  • harness_hellaswag_10

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hellaswag|10_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hellaswag|10_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: 多个文件路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: 多个文件路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_anatomy_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_astronomy_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_college_biology_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_college_physics_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_computer_security_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_econometrics_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T21:36:17.720122.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分割: 2023_07_19T21_36_17.720122
      • 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T21:36:17.720122.parquet
    • 分割: latest
      • 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T21:36:17.720122.parquet

最新结果

以下是2023年10月15日10:49:53.793437运行的最新结果:

python { "all": { "em": 0.008179530201342282, "em_stderr": 0.0009224022743034369, "f1": 0.06148594798657739, "f1_stderr": 0.0015819609377213373, "acc": 0.35462516448027825, "acc_stderr": 0.008753822199298638 }, "harness|drop|3": { "em": 0.008179530201342282, "em_stderr": 0.0009224022743034369, "f1": 0.06148594798657739, "f1_stderr": 0.0015819609377213373 }, "harness|gsm8k|5": { "acc": 0.026535253980288095, "acc_stderr": 0.0044270459872651595 }, "harness|winogrande|5": { "acc": 0.6827150749802684

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