five

open-llm-leaderboard/details_bunnycore__Starling-dolphin-E26-7B

收藏
Hugging Face2024-04-07 更新2024-06-11 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_bunnycore__Starling-dolphin-E26-7B
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在模型bunnycore/Starling-dolphin-E26-7B在Open LLM Leaderboard上的评估运行期间自动创建的。数据集由63个配置组成,每个配置对应一个评估任务。它包含一次运行的结果,每次运行在每个配置中表示为特定的分割。train分割始终指向最新结果。一个名为results的附加配置存储了所有运行的聚合结果,这些结果用于计算和显示Open LLM Leaderboard上的聚合指标。README还提供了一个如何使用Python中的datasets库加载运行细节的示例。

该数据集是在模型bunnycore/Starling-dolphin-E26-7B在Open LLM Leaderboard上的评估运行期间自动创建的。数据集由63个配置组成,每个配置对应一个评估任务。它包含一次运行的结果,每次运行在每个配置中表示为特定的分割。train分割始终指向最新结果。一个名为results的附加配置存储了所有运行的聚合结果,这些结果用于计算和显示Open LLM Leaderboard上的聚合指标。README还提供了一个如何使用Python中的datasets库加载运行细节的示例。
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

数据集名称

  • pretty_name: Evaluation run of bunnycore/Starling-dolphin-E26-7B

数据集描述

数据集组成

  • 数据结构: 包含63个配置,每个配置对应一个评估任务。
  • 数据生成: 数据集由1次运行创建,每次运行作为一个特定的分割,分割名使用运行的时间戳命名。
  • 特殊配置: 额外配置“results”存储所有运行的聚合结果,用于计算和显示聚合指标。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_bunnycore__Starling-dolphin-E26-7B", "harness_winogrande_5", split="train")

最新结果

  • 结果示例: 展示了不同任务的准确率(acc)和标准误差(acc_stderr)等指标。

数据集配置详情

配置列表

  • config_name: harness_arc_challenge_25

    • data_files:
      • split: 2024_04_07T16_30_50.523461
        • path: **/details_harness|arc:challenge|25_2024-04-07T16-30-50.523461.parquet
      • split: latest
        • path: **/details_harness|arc:challenge|25_2024-04-07T16-30-50.523461.parquet
  • config_name: harness_gsm8k_5

    • data_files:
      • split: 2024_04_07T16_30_50.523461
        • path: **/details_harness|gsm8k|5_2024-04-07T16-30-50.523461.parquet
      • split: latest
        • path: **/details_harness|gsm8k|5_2024-04-07T16-30-50.523461.parquet
  • config_name: harness_hellaswag_10

    • data_files:
      • split: 2024_04_07T16_30_50.523461
        • path: **/details_harness|hellaswag|10_2024-04-07T16-30-50.523461.parquet
      • split: latest
        • path: **/details_harness|hellaswag|10_2024-04-07T16-30-50.523461.parquet
  • config_name: harness_hendrycksTest_5

    • data_files:
      • split: 2024_04_07T16_30_50.523461
        • path:
          • **/details_harness|hendrycksTest-abstract_algebra|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-anatomy|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-astronomy|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-business_ethics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_biology|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_chemistry|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_computer_science|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_mathematics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_medicine|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_physics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-computer_security|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-conceptual_physics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-econometrics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-electrical_engineering|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-formal_logic|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-global_facts|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-high_school_biology|5_2024-04-07T16-30-50.523461.parquet
      • split: latest
        • path:
          • **/details_harness|hendrycksTest-abstract_algebra|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-anatomy|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-astronomy|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-business_ethics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_biology|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_chemistry|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_computer_science|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_mathematics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_medicine|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-college_physics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-computer_security|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-conceptual_physics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-econometrics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-electrical_engineering|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-formal_logic|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-global_facts|5_2024-04-07T16-30-50.523461.parquet
          • **/details_harness|hendrycksTest-high_school_biology|5_2024-04-07T16-30-50.523461.parquet

数据集使用说明

  • 加载数据: 使用load_dataset函数加载特定配置的数据,如示例所示。
  • 数据分割: 数据集中的每个配置都有基于时间戳的分割和“latest”分割,方便获取最新数据。

以上是对数据集的详细描述和配置信息,用户可根据需要加载和分析数据。

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