five

open-llm-leaderboard/details_mosaicml__mpt-7b

收藏
Hugging Face2023-10-03 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_mosaicml__mpt-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mosaicml/mpt-7b的评估运行期间自动创建的,用于Open LLM Leaderboard的评估。数据集由122个配置组成,每个配置对应一个评估任务。数据集由4次运行创建,每次运行可以在每个配置中找到特定的分割,分割以运行的时间戳命名。"train"分割始终指向最新的结果。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
原始信息汇总

数据集概述

数据集名称

  • 名称: Evaluation run of mosaicml/mpt-7b

数据集组成

  • 配置数量: 122个配置
  • 每个配置对应: 一个评估任务
  • 创建来源: 从4次运行中创建
  • 分割命名: 使用运行的时间戳命名
  • 训练分割: 总是指向最新的结果
  • 额外配置: "results" 存储所有聚合结果,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_mosaicml__mpt-7b", "harness_truthfulqa_mc_0", split="train")

最新结果

  • 最新结果来源: 2023-10-03T22:10:31.153532
  • 结果示例: python { "all": { "acc": 0.28815728428182913, "acc_stderr": 0.032729017222815425, "acc_norm": 0.2923951167846347, "acc_norm_stderr": 0.032718180607395383, "mc1": 0.20563035495716034, "mc1_stderr": 0.014148482219460974, "mc2": 0.3354506043570123, "mc2_stderr": 0.013110323313593984 }, "harness|arc:challenge|25": { "acc": 0.42918088737201365, "acc_stderr": 0.014464085894870653, "acc_norm": 0.47696245733788395, "acc_norm_stderr": 0.014595873205358269 }, "harness|hellaswag|10": { "acc": 0.5730930093606851, "acc_stderr": 0.004936176784631949, "acc_norm": 0.7753435570603465, "acc_norm_stderr": 0.0041650291643616005 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.19, "acc_stderr": 0.03942772444036624, "acc_norm": 0.19, "acc_norm_stderr": 0.03942772444036624 },

    其他任务结果省略...

}

配置详情

  • 配置名称: harness_arc_challenge_25

    • 分割: 2023_08_22T15_05_51.358534
      • 路径: **/details_harness|arc:challenge|25_2023-08-22T15:05:51.358534.parquet
    • 分割: 2023_10_03T22_10_31.153532
      • 路径: **/details_harness|arc:challenge|25_2023-10-03T22-10-31.153532.parquet
    • 分割: latest
      • 路径: **/details_harness|arc:challenge|25_2023-10-03T22-10-31.153532.parquet
  • 配置名称: harness_drop_3

    • 分割: 2023_09_23T17_09_43.658606
      • 路径: **/details_harness|drop|3_2023-09-23T17-09-43.658606.parquet
    • 分割: latest
      • 路径: **/details_harness|drop|3_2023-09-23T17-09-43.658606.parquet
  • 配置名称: harness_gsm8k_5

    • 分割: 2023_09_23T17_09_43.658606
      • 路径: **/details_harness|gsm8k|5_2023-09-23T17-09-43.658606.parquet
    • 分割: latest
      • 路径: **/details_harness|gsm8k|5_2023-09-23T17-09-43.658606.parquet
  • 配置名称: harness_hellaswag_10

    • 分割: 2023_08_22T15_05_51.358534
      • 路径: **/details_harness|hellaswag|10_2023-08-22T15:05:51.358534.parquet
    • 分割: 2023_10_03T22_10_31.153532
      • 路径: **/details_harness|hellaswag|10_2023-10-03T22-10-31.153532.parquet
    • 分割: latest
      • 路径: **/details_harness|hellaswag|10_2023-10-03T22-10-31.153532.parquet
  • 配置名称: harness_hendrycksTest_5

    • 分割: 2023_08_22T15_05_51.358534
      • 路径:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2023-08-22T15:05:51.358534.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-22T15:05:51.358534.parquet
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