five

open-llm-leaderboard-old/details_harborwater__dpo-test-hermes-open-llama-3b

收藏
Hugging Face2024-01-14 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_harborwater__dpo-test-hermes-open-llama-3b
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在Open LLM Leaderboard上对模型harborwater/dpo-test-hermes-open-llama-3b进行评估时自动创建的。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次运行创建,每次运行的结果作为特定配置中的一个分割,分割名称使用运行的时间戳。train分割始终指向最新结果。此外,results配置存储了所有运行的聚合结果,用于在Open LLM Leaderboard上计算和显示聚合指标。

该数据集是在Open LLM Leaderboard上对模型harborwater/dpo-test-hermes-open-llama-3b进行评估时自动创建的。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次运行创建,每次运行的结果作为特定配置中的一个分割,分割名称使用运行的时间戳。train分割始终指向最新结果。此外,results配置存储了所有运行的聚合结果,用于在Open LLM Leaderboard上计算和显示聚合指标。
提供机构:
open-llm-leaderboard-old
原始信息汇总

数据集概述

该数据集是在对模型 harborwater/dpo-test-hermes-open-llama-3b 进行评估运行期间自动创建的,用于 Open LLM Leaderboard。数据集包含 63 个配置,每个配置对应一个评估任务。

数据集结构

  • 配置数量:63 个配置
  • 创建来源:从 1 次运行中创建
  • 数据分割:每个配置包含特定运行的分割,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,用于计算和显示聚合指标。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_harborwater__dpo-test-hermes-open-llama-3b", "harness_winogrande_5", split="train")

最新结果

以下是 2024-01-14T04:56:07.071188 运行的最新结果

python { "all": { "acc": 0.2514093021467422, "acc_stderr": 0.03052650097964464, "acc_norm": 0.25202173312622367, "acc_norm_stderr": 0.03127688845727799, "mc1": 0.2484700122399021, "mc1_stderr": 0.015127427096520672, "mc2": 0.3980562710501165, "mc2_stderr": 0.014269053798319005 }, "harness|arc:challenge|25": { "acc": 0.36689419795221845, "acc_stderr": 0.014084133118104292, "acc_norm": 0.3924914675767918, "acc_norm_stderr": 0.014269634635670712 }, "harness|hellaswag|10": { "acc": 0.5091615216092412, "acc_stderr": 0.004988943721711217, "acc_norm": 0.6745668193586934, "acc_norm_stderr": 0.004675789156977649 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.22, "acc_stderr": 0.04163331998932269, "acc_norm": 0.22, "acc_norm_stderr": 0.04163331998932269 }, # 其他任务的结果... }

配置详情

  • config_name: harness_arc_challenge_25

    • data_files:
      • split: 2024_01_14T04_56_07.071188
        • **/details_harness|arc:challenge|25_2024-01-14T04-56-07.071188.parquet
      • split: latest
        • **/details_harness|arc:challenge|25_2024-01-14T04-56-07.071188.parquet
  • config_name: harness_gsm8k_5

    • data_files:
      • split: 2024_01_14T04_56_07.071188
        • **/details_harness|gsm8k|5_2024-01-14T04-56-07.071188.parquet
      • split: latest
        • **/details_harness|gsm8k|5_2024-01-14T04-56-07.071188.parquet
  • config_name: harness_hellaswag_10

    • data_files:
      • split: 2024_01_14T04_56_07.071188
        • **/details_harness|hellaswag|10_2024-01-14T04-56-07.071188.parquet
      • split: latest
        • **/details_harness|hellaswag|10_2024-01-14T04-56-07.071188.parquet
  • config_name: harness_hendrycksTest_5

    • data_files:
      • split: 2024_01_14T04_56_07.071188
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-01-14T04-56-07.071188.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-01-14T04-56-07.071188.parquet
        • `**/details_harness|hendrycksTest-global_facts|5_2024-01-14
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作