open-llm-leaderboard/details_oobabooga__CodeBooga-34B-v0.1

该数据集是在模型 oobabooga/CodeBooga-34B-v0.1 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。数据集包含一次运行的结果，每次运行都作为一个特定的分割，分割名称由运行的时间戳命名。train 分割始终指向最新的结果。一个名为 results 的额外配置存储了所有运行的聚合结果，这些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 Hugging Face datasets 库加载数据集的示例。

原始信息汇总

数据集概述

数据集名称

• pretty_name: Evaluation run of oobabooga/CodeBooga-34B-v0.1

数据集描述

• dataset_summary: 该数据集是在评估模型oobabooga/CodeBooga-34B-v0.1的过程中自动创建的，该模型在Open LLM Leaderboard上进行评估。

数据集组成

• 包含63个配置，每个配置对应一个评估任务。
• 数据集由1次运行创建，每次运行作为一个特定的分割，分割名称使用运行的时间戳命名。
• “train”分割始终指向最新结果。
• 额外的“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。

数据集加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_oobabooga__CodeBooga-34B-v0.1", "harness_winogrande_5", split="train")

最新结果

• 最新结果来自2024-04-17T08:23:48.373546的运行，包含多个任务的评估数据。

数据集配置详情

配置列表

• config_name: harness_arc_challenge_25

  • data_files:
    • split: 2024_04_17T08_23_48.373546
      • path: **/details_harness|arc:challenge|25_2024-04-17T08-23-48.373546.parquet
    • split: latest
      • path: **/details_harness|arc:challenge|25_2024-04-17T08-23-48.373546.parquet

• config_name: harness_gsm8k_5

  • data_files:
    • split: 2024_04_17T08_23_48.373546
      • path: **/details_harness|gsm8k|5_2024-04-17T08-23-48.373546.parquet
    • split: latest
      • path: **/details_harness|gsm8k|5_2024-04-17T08-23-48.373546.parquet

• config_name: harness_hellaswag_10

  • data_files:
    • split: 2024_04_17T08_23_48.373546
      • path: **/details_harness|hellaswag|10_2024-04-17T08-23-48.373546.parquet
    • split: latest
      • path: **/details_harness|hellaswag|10_2024-04-17T08-23-48.373546.parquet

• config_name: harness_hendrycksTest_5

  • data_files:
    • split: 2024_04_17T08_23_48.373546
      • path:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2024-04-17T08-23-48.373546.parquet
        • **/details_harness|hendrycksTest-high_school_biology|5_2024-04-17T08-23

搜集汇总

背景与挑战

背景概述

该数据集是模型oobabooga/CodeBooga-34B-v0.1在Open LLM Leaderboard评估过程中自动生成的，包含63个任务配置和一次运行结果，以时间戳分割存储，train分割指向最新结果，并有一个results配置用于聚合指标计算，便于排行榜展示和数据分析。