open-llm-leaderboard-old/details_liminerity__M7-7b

该数据集是在Open LLM Leaderboard上对模型liminerity/M7-7b进行评估时自动创建的。它由63个配置组成，每个配置对应一个评估任务。数据集是从1次运行中创建的，每次运行作为每个配置中的一个特定分割，使用运行的时间戳命名。train分割始终指向最新的结果。一个额外的配置results存储了所有运行的聚合结果，用于计算和显示Open LLM Leaderboard上的聚合指标。

原始信息汇总

数据集概述

数据集名称

• 名称: Evaluation run of liminerity/M7-7b

数据集来源

• 创建背景: 该数据集是在对模型 liminerity/M7-7b 进行评估运行期间自动创建的，评估运行在 Open LLM Leaderboard 上进行。

数据集结构

• 配置数量: 该数据集包含 63 个配置，每个配置对应一个评估任务。
• 运行次数: 数据集从 1 次运行中创建。每个运行可以在每个配置中作为一个特定的分片找到，分片名称使用运行的时间戳。
• 最新结果: "train" 分片始终指向最新的结果。
• 结果汇总: 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 Open LLM Leaderboard 上的汇总指标。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_liminerity__M7-7b", "harness_winogrande_5", split="train")

最新结果

• 最新结果时间戳: 2024-03-10T04:12:29.468873
• 结果示例: python { "all": { "acc": 0.6518510677976063, "acc_stderr": 0.03203683002461246, "acc_norm": 0.6505703294705394, "acc_norm_stderr": 0.03271494006314174, "mc1": 0.631578947368421, "mc1_stderr": 0.016886551261046046, "mc2": 0.7792924999444771, "mc2_stderr": 0.013713064522592473 }, "harness|arc:challenge|25": { "acc": 0.7167235494880546, "acc_stderr": 0.013167478735134575, "acc_norm": 0.7286689419795221, "acc_norm_stderr": 0.012993807727545796 }, "harness|hellaswag|10": { "acc": 0.7177853017327226, "acc_stderr": 0.0044915745394418834, "acc_norm": 0.8914558852818164, "acc_norm_stderr": 0.003104306434972473 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.31, "acc_stderr": 0.04648231987117316, "acc_norm": 0.31, "acc_norm_stderr": 0.04648231987117316 },

  其他任务的结果...

}

配置详情

• 配置名称: harness_arc_challenge_25

  • 数据文件:
    • 分片: 2024_03_10T04_12_29.468873
      • 路径: **/details_harness|arc:challenge|25_2024-03-10T04-12-29.468873.parquet
    • 分片: latest
      • 路径: **/details_harness|arc:challenge|25_2024-03-10T04-12-29.468873.parquet

• 配置名称: harness_gsm8k_5

  • 数据文件:
    • 分片: 2024_03_10T04_12_29.468873
      • 路径: **/details_harness|gsm8k|5_2024-03-10T04-12-29.468873.parquet
    • 分片: latest
      • 路径: **/details_harness|gsm8k|5_2024-03-10T04-12-29.468873.parquet

• 配置名称: harness_hellaswag_10

  • 数据文件:
    • 分片: 2024_03_10T04_12_29.468873
      • 路径: **/details_harness|hellaswag|10_2024-03-10T04-12-29.468873.parquet
    • 分片: latest
      • 路径: **/details_harness|hellaswag|10_2024-03-10T04-12-29.468873.parquet

• 配置名称: harness_hendrycksTest_5

  • 数据文件:
    • 分片: 2024_03_10T04_12_29.468873
      • 路径:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-high_school_biology|5_2024-03-10T04-12-29.468873.parquet
        • **/details_harness|hendrycksTest-high_school_chemistry|5_2024-03-10T04-12-29.468873.parquet

        其他文件路径...

}