five

open-llm-leaderboard-old/details_Pirr__pythia-13b-deduped-green_devil

收藏
Hugging Face2023-10-28 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_Pirr__pythia-13b-deduped-green_devil
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上评估模型Pirr/pythia-13b-deduped-green_devil时自动生成的。数据集包含64个配置,每个配置对应一个评估任务。数据集由2次运行生成,每次运行的结果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上评估模型Pirr/pythia-13b-deduped-green_devil时自动生成的。数据集包含64个配置,每个配置对应一个评估任务。数据集由2次运行生成,每次运行的结果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

数据集来源

该数据集是在评估模型 Pirr/pythia-13b-deduped-green_devilOpen LLM Leaderboard 上的运行过程中自动创建的。

数据集结构

  • 配置数量:64个配置,每个配置对应一个评估任务。
  • 数据来源:数据集由2次运行结果组成,每次运行的结果作为一个特定的分割(split)存储,分割名称使用运行的时间戳。
  • 最新结果:"train" 分割始终指向最新的结果。
  • 汇总结果:一个额外的配置 "results" 存储所有运行的汇总结果,用于计算和显示在 Open LLM Leaderboard 上的聚合指标。

数据加载示例

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

最新结果

以下是 2023-10-28T10:52:35.771401 运行 的最新结果: python { "all": { "em": 0.001153523489932886, "em_stderr": 0.0003476179896857083, "f1": 0.04843435402684591, "f1_stderr": 0.0011963744934619489, "acc": 0.34526287822984214, "acc_stderr": 0.008596442508425665 }, "harness|drop|3": { "em": 0.001153523489932886, "em_stderr": 0.0003476179896857083, "f1": 0.04843435402684591, "f1_stderr": 0.0011963744934619489 }, "harness|gsm8k|5": { "acc": 0.02122820318423048, "acc_stderr": 0.003970449129848635 }, "harness|winogrande|5": { "acc": 0.6692975532754538, "acc_stderr": 0.013222435887002696 } }

配置详情

以下是部分配置的详细信息:

  • harness_arc_challenge_25

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|arc:challenge|25_2023-07-18T14:06:19.404792.parquet
  • harness_drop_3

    • 分割:2023_10_28T10_52_35.771401, latest
    • 路径:**/details_harness|drop|3_2023-10-28T10-52-35.771401.parquet
  • harness_gsm8k_5

    • 分割:2023_10_28T10_52_35.771401, latest
    • 路径:**/details_harness|gsm8k|5_2023-10-28T10-52-35.771401.parquet
  • harness_hellaswag_10

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hellaswag|10_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_anatomy_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_astronomy_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_college_biology_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_college_physics_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_computer_security_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_econometrics_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-18T14:06:19.404792.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分割:2023_07_18T14_06_19.404792, latest
    • 路径:**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-18T14:06:19.404792.parquet

以上是数据集的概述和部分配置的详细信息。如需更多配置详情,请参阅原始数据集详情页面。

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