five

open-llm-leaderboard-old/details_vicgalle__zephyr-7b-truthy

收藏
Hugging Face2024-02-10 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_vicgalle__zephyr-7b-truthy
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在vicgalle/zephyr-7b-truthy模型的评估运行期间自动创建的,包含了63个配置,每个配置对应一个评估任务。数据集由一次运行创建,每个运行可以在每个配置中找到,使用运行的时间戳作为分割名称。此外,还有一个名为results的配置,存储了所有运行的聚合结果,用于计算和显示在Open LLM Leaderboard上的聚合指标。

该数据集是在vicgalle/zephyr-7b-truthy模型的评估运行期间自动创建的,包含了63个配置,每个配置对应一个评估任务。数据集由一次运行创建,每个运行可以在每个配置中找到,使用运行的时间戳作为分割名称。此外,还有一个名为results的配置,存储了所有运行的聚合结果,用于计算和显示在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

数据集摘要

该数据集是在评估模型 vicgalle/zephyr-7b-truthyOpen LLM Leaderboard 上的自动创建的。数据集包含 63 个配置,每个配置对应一个评估任务。数据集从 1 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 63 个配置,每个配置对应一个评估任务。每个配置包含多个分割,分割名称使用运行的时间戳。"train" 分割始终指向最新的结果。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_vicgalle__zephyr-7b-truthy", "harness_winogrande_5", split="train")

最新结果

以下是 最新结果 的摘要: python { "all": { "acc": 0.593174337288042, "acc_stderr": 0.033460408309810855, "acc_norm": 0.5997217687076803, "acc_norm_stderr": 0.034170774358741766, "mc1": 0.45532435740514077, "mc1_stderr": 0.017433490102538765, "mc2": 0.6330887790426952, "mc2_stderr": 0.01528797501626636 }, "harness|arc:challenge|25": { "acc": 0.5793515358361775, "acc_stderr": 0.0144262112525084, "acc_norm": 0.6075085324232082, "acc_norm_stderr": 0.014269634635670717 }, "harness|hellaswag|10": { "acc": 0.6409081856203943, "acc_stderr": 0.004787537385153002, "acc_norm": 0.8464449312885879, "acc_norm_stderr": 0.0035978491398150577 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.28, "acc_stderr": 0.04512608598542129, "acc_norm": 0.28, "acc_norm_stderr": 0.04512608598542129 }, # 其他任务的结果... }

配置详情

数据集包含多个配置,每个配置对应一个评估任务。以下是部分配置的示例:

  • config_name: harness_arc_challenge_25

    • split: 2024_02_10T14_59_44.699643
      • path: **/details_harness|arc:challenge|25_2024-02-10T14-59-44.699643.parquet
    • split: latest
      • path: **/details_harness|arc:challenge|25_2024-02-10T14-59-44.699643.parquet
  • config_name: harness_gsm8k_5

    • split: 2024_02_10T14_59_44.699643
      • path: **/details_harness|gsm8k|5_2024-02-10T14-59-44.699643.parquet
    • split: latest
      • path: **/details_harness|gsm8k|5_2024-02-10T14-59-44.699643.parquet
  • config_name: harness_hellaswag_10

    • split: 2024_02_10T14_59_44.699643
      • path: **/details_harness|hellaswag|10_2024-02-10T14-59-44.699643.parquet
    • split: latest
      • path: **/details_harness|hellaswag|10_2024-02-10T14-59-44.699643.parquet
  • config_name: harness_hendrycksTest_5

    • split: 2024_02_10T14_59_44.699643
      • path:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2024-02-10T14-59-44.699643.parquet
        • **/details_harness|hendrycksTest-high_school_biology|5_2024-02-10T14-59-44.699643.parquet
二维码
社区交流群
二维码
科研交流群
商业服务