open-llm-leaderboard-old/details_rizla__rizla-17

该数据集是在模型rizla/rizla-17的评估运行期间自动创建的，用于Open LLM Leaderboard的评估。数据集包含63个配置，每个配置对应一个评估任务。数据集由1次运行生成，每次运行的结果存储为特定配置中的一个分割，分割名称使用运行的时间戳。train分割始终指向最新的结果。此外，results配置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

数据集信息

• 数据集名称: Evaluation run of rizla/rizla-17
• 数据集来源: 自动创建于模型 rizla/rizla-17 在 Open LLM Leaderboard 的评估运行中。
• 数据集组成: 包含63个配置，每个配置对应一个评估任务。
• 数据集创建: 从1次运行中创建，每次运行作为一个特定的分割存储，分割名称使用运行的时间戳。"train" 分割始终指向最新的结果。
• 额外配置: "results" 配置存储所有运行的聚合结果，用于计算和显示聚合指标在 Open LLM Leaderboard 上。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_rizla__rizla-17", "harness_winogrande_5", split="train")

最新结果

• 最新结果来源: 2024-02-02T23:58:31.137464
• 结果示例: python { "all": { "acc": 0.6499955281921569, "acc_stderr": 0.0321248371661185, "acc_norm": 0.6498192760869008, "acc_norm_stderr": 0.032802028367607344, "mc1": 0.6132190942472461, "mc1_stderr": 0.017048857010515103, "mc2": 0.7692598698604171, "mc2_stderr": 0.013946986686441027 }, "harness|arc:challenge|25": { "acc": 0.712457337883959, "acc_stderr": 0.013226719056266129, "acc_norm": 0.7363481228668942, "acc_norm_stderr": 0.01287592915129705 }, "harness|hellaswag|10": { "acc": 0.7381995618402709, "acc_stderr": 0.0043871612030879446, "acc_norm": 0.8972316271659032, "acc_norm_stderr": 0.0030303552466563882 }, ... }

配置信息

• 配置名称: harness_arc_challenge_25

  • 数据文件:
    • 分割: 2024_02_02T23_58_31.137464
      • 路径: **/details_harness|arc:challenge|25_2024-02-02T23-58-31.137464.parquet
    • 分割: latest
      • 路径: **/details_harness|arc:challenge|25_2024-02-02T23-58-31.137464.parquet

• 配置名称: harness_gsm8k_5

  • 数据文件:
    • 分割: 2024_02_02T23_58_31.137464
      • 路径: **/details_harness|gsm8k|5_2024-02-02T23-58-31.137464.parquet
    • 分割: latest
      • 路径: **/details_harness|gsm8k|5_2024-02-02T23-58-31.137464.parquet

• 配置名称: harness_hellaswag_10

  • 数据文件:
    • 分割: 2024_02_02T23_58_31.137464
      • 路径: **/details_harness|hellaswag|10_2024-02-02T23-58-31.137464.parquet
    • 分割: latest
      • 路径: **/details_harness|hellaswag|10_2024-02-02T23-58-31.137464.parquet

• 配置名称: harness_hendrycksTest_5

  • 数据文件:
    • 分割: 2024_02_02T23_58_31.137464
      • 路径:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2024-02-02T23-58-31.137464.parquet
        • **/details_harness|hendrycksTest-high_school_biology|5_2024-02-02T23-58-31.137464.parquet
        • ...