open-llm-leaderboard-old/details_OpenBuddy__openbuddy-mixtral-7bx8-v17.3-32k

该数据集是在模型OpenBuddy/openbuddy-mixtral-7bx8-v17.3-32k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，每个配置对应一个被评估的任务。数据集是从1次运行中创建的，每次运行在每个配置中作为一个特定的分割找到。train分割始终指向最新的结果。一个额外的配置results存储了运行的所有聚合结果，用于计算和显示Open LLM Leaderboard上的聚合指标。README还提供了如何使用Python代码加载运行中的详细信息的示例，并包含了特定运行的最新结果。

原始信息汇总

数据集概述

数据集名称

• 名称: Evaluation run of OpenBuddy/openbuddy-mixtral-7bx8-v17.3-32k

数据集描述

• 创建目的: 该数据集是在模型 OpenBuddy/openbuddy-mixtral-7bx8-v17.3-32k 在 Open LLM Leaderboard 上的评估运行期间自动创建的。
• 组成: 数据集包含 63 个配置，每个配置对应一个评估任务。
• 创建次数: 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。"train" 分片始终指向最新的结果。
• 额外配置: 一个名为 "results" 的额外配置存储了所有运行的聚合结果，用于计算并在 Open LLM Leaderboard 上显示聚合指标。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_OpenBuddy__openbuddy-mixtral-7bx8-v17.3-32k", "harness_winogrande_5", split="train")

最新结果

• 最新结果时间戳: 2024-02-03T22:04:01.545858
• 结果示例: python { "all": { "acc": 0.6917313320960925, "acc_stderr": 0.030694742722936492, "acc_norm": 0.6985909537220241, "acc_norm_stderr": 0.03127101232040077, "mc1": 0.4222766217870257, "mc1_stderr": 0.017290733254248167, "mc2": 0.5913829725730599, "mc2_stderr": 0.015218394186481066 }, "harness|arc:challenge|25": { "acc": 0.6151877133105802, "acc_stderr": 0.014218371065251104, "acc_norm": 0.6450511945392492, "acc_norm_stderr": 0.013983036904094094 }, "harness|hellaswag|10": { "acc": 0.5156343357896833, "acc_stderr": 0.004987341485856662, "acc_norm": 0.6695877315275841, "acc_norm_stderr": 0.004694002781939541 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.35, "acc_stderr": 0.0479372485441102, "acc_norm": 0.35, "acc_norm_stderr": 0.0479372485441102 },

  更多任务结果...

}

配置详情

• 配置名称: harness_arc_challenge_25

  • 数据文件:
    • 分片: 2024_02_03T22_04_01.545858
      • 路径: **/details_harness|arc:challenge|25_2024-02-03T22-04-01.545858.parquet
    • 分片: latest
      • 路径: **/details_harness|arc:challenge|25_2024-02-03T22-04-01.545858.parquet

• 配置名称: harness_gsm8k_5

  • 数据文件:
    • 分片: 2024_02_03T22_04_01.545858
      • 路径: **/details_harness|gsm8k|5_2024-02-03T22-04-01.545858.parquet
    • 分片: latest
      • 路径: **/details_harness|gsm8k|5_2024-02-03T22-04-01.545858.parquet

• 配置名称: harness_hellaswag_10

  • 数据文件:
    • 分片: 2024_02_03T22_04_01.545858
      • 路径: **/details_harness|hellaswag|10_2024-02-03T22-04-01.545858.parquet
    • 分片: latest
      • 路径: **/details_harness|hellaswag|10_2024-02-03T22-04-01.545858.parquet

• 配置名称: harness_hendrycksTest_5

  • 数据文件:
    • 分片: 2024_02_03T22_04_01.545858
      • 路径:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-02-03T22-04-01.545858.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2024-02-03T22-04-01.545858.parquet

        更多路径...