open-llm-leaderboard/details_wannaphong__han-llm-7b-v3

该数据集是在模型 wannaphong/han-llm-7b-v3 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，每个配置对应一个评估任务。数据集是从 1 次运行中生成的，每次运行在每个配置中表示为特定的分割。train 分割始终指向最新结果。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 库加载运行中的详细信息的示例。还包括了 2024-03-12 运行的最新结果，显示了不同任务的各种准确率指标。

原始信息汇总

数据集概述

数据集名称

• pretty_name: Evaluation run of wannaphong/han-llm-7b-v3

数据集描述

• dataset_summary: 该数据集是在评估模型wannaphong/han-llm-7b-v3时自动创建的，用于Open LLM Leaderboard。

数据集组成

• 配置数量: 63个
• 每个配置对应一个评估任务
• 数据来源: 来自1次运行
• 数据分割: 每个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
• 额外配置: "results"存储所有运行的聚合结果，用于计算和显示聚合指标

数据集加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_wannaphong__han-llm-7b-v3", "harness_winogrande_5", split="train")

数据集详细配置

配置列表

1. config_name: harness_arc_challenge_25

   • data_files:
     • split: 2024_03_12T10_34_55.967829
       • path: /details_harness|arc:challenge|25_2024-03-12T10-34-55.967829.parquet
     • split: latest
       • path: /details_harness|arc:challenge|25_2024-03-12T10-34-55.967829.parquet

2. config_name: harness_gsm8k_5

   • data_files:
     • split: 2024_03_12T10_34_55.967829
       • path: /details_harness|gsm8k|5_2024-03-12T10-34-55.967829.parquet
     • split: latest
       • path: /details_harness|gsm8k|5_2024-03-12T10-34-55.967829.parquet

3. config_name: harness_hellaswag_10

   • data_files:
     • split: 2024_03_12T10_34_55.967829
       • path: /details_harness|hellaswag|10_2024-03-12T10-34-55.967829.parquet
     • split: latest
       • path: /details_harness|hellaswag|10_2024-03-12T10-34-55.967829.parquet

4. config_name: harness_hendrycksTest_5

   • data_files:
     • split: 2024_03_12T10_34_55.967829
       • path:
         • /details_harness|hendrycksTest-abstract_algebra|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-anatomy|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-astronomy|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-business_ethics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-clinical_knowledge|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_biology|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_chemistry|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_computer_science|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_mathematics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_medicine|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_physics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-computer_security|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-conceptual_physics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-econometrics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-electrical_engineering|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-elementary_mathematics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-formal_logic|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-global_facts|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-high_school_biology|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-high_school_chemistry|5_2024-03-12T10-34-55.967829.parquet
     • split: latest
       • path: /details_harness|hendrycksTest-abstract_algebra|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-anatomy|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-astronomy|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-business_ethics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-clinical_knowledge|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_biology|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_chemistry|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_computer_science|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_mathematics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_medicine|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-college_physics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-computer_security|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-conceptual_physics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-econometrics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-electrical_engineering|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-elementary_mathematics|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-formal_logic|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-global_facts|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-high_school_biology|5_2024-03-12T10-34-55.967829.parquet
         • /details_harness|hendrycksTest-high_school_chemistry|5_2024-03-12T10-34-55.967829.parquet

数据集使用方法

• 加载数据集: 使用load_dataset函数加载特定配置的数据，例如加载harness_winogrande_5配置的训练数据。
• 数据分割: 数据集支持按时间戳或最新数据进行分割，用户可根据需求选择不同的分割方式。