open-llm-leaderboard/details_allknowingroger__NexusRaven-15B-pass

该数据集是在模型allknowingroger/NexusRaven-15B-pass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。它由1次运行创建，每次运行作为每个配置中的一个特定分割，分割名称使用运行的时间戳。train分割始终指向最新的结果。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中的datasets库加载运行细节的示例。

原始信息汇总

数据集概述

数据集信息

• 名称: Evaluation run of allknowingroger/NexusRaven-15B-pass
• 来源: 自动创建于模型 allknowingroger/NexusRaven-15B-pass 在 Open LLM Leaderboard 的评估运行。

数据集组成

• 配置数量: 63 个配置，每个配置对应一个评估任务。
• 创建次数: 从 1 次运行中创建。每个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" 分割始终指向最新结果。
• 额外配置: "results" 配置存储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。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_allknowingroger__NexusRaven-15B-pass", "harness_winogrande_5", split="train")

最新结果

• 最新结果时间戳: 2024-04-11T07:30:47.900772
• 聚合结果: python { "all": { "acc": 0.4249753091336553, "acc_stderr": 0.034275379775475075, "acc_norm": 0.43178118203852844, "acc_norm_stderr": 0.0352092152747723, "mc1": 0.2802937576499388, "mc1_stderr": 0.015723139524608753, "mc2": 0.44836340333599356, "mc2_stderr": 0.016208658629744008 }, "harness|arc:challenge|25": { "acc": 0.3609215017064846, "acc_stderr": 0.014034761386175458, "acc_norm": 0.39078498293515357, "acc_norm_stderr": 0.014258563880513777 }, "harness|hellaswag|10": { "acc": 0.3986257717586138, "acc_stderr": 0.004886147907627405, "acc_norm": 0.518621788488349, "acc_norm_stderr": 0.0049863195875249604 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.3, "acc_stderr": 0.046056618647183814, "acc_norm": 0.3, "acc_norm_stderr": 0.046056618647183814 }, ... }

配置详情

• 配置名称: harness_arc_challenge_25

  • 数据文件:
    • 分割: 2024_04_11T07_30_47.900772
      • 路径: **/details_harness|arc:challenge|25_2024-04-11T07-30-47.900772.parquet
    • 分割: latest
      • 路径: **/details_harness|arc:challenge|25_2024-04-11T07-30-47.900772.parquet

• 配置名称: harness_gsm8k_5

  • 数据文件:
    • 分割: 2024_04_11T07_30_47.900772
      • 路径: **/details_harness|gsm8k|5_2024-04-11T07-30-47.900772.parquet
    • 分割: latest
      • 路径: **/details_harness|gsm8k|5_2024-04-11T07-30-47.900772.parquet

• 配置名称: harness_hellaswag_10

  • 数据文件:
    • 分割: 2024_04_11T07_30_47.900772
      • 路径: **/details_harness|hellaswag|10_2024-04-11T07-30-47.900772.parquet
    • 分割: latest
      • 路径: **/details_harness|hellaswag|10_2024-04-11T07-30-47.900772.parquet

• 配置名称: harness_hendrycksTest_5

  • 数据文件:
    • 分割: 2024_04_11T07_30_47.900772
      • 路径:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-04-11T07-30-47.900772.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2024-04-11T07-30-47.900772.parquet ...