five

open-llm-leaderboard/details_harborwater__open-llama-3b-everythingLM-2048

收藏
Hugging Face2023-10-24 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_harborwater__open-llama-3b-everythingLM-2048
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在模型 harborwater/open-llama-3b-everythingLM-2048 在 Open LLM Leaderboard 上评估期间自动创建的。数据集由 64 个配置组成,每个配置对应一个被评估的任务。数据集由 2 次运行创建,每次运行在每个配置中表示为特定的拆分。train 拆分始终指向最新的结果。此外,还有一个 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 上的指标。文件还提供了如何使用 Hugging Face datasets 库加载数据集的示例。

This dataset was automatically generated during the evaluation of the model harborwater/open-llama-3b-everythingLM-2048 on the Open LLM Leaderboard. It comprises 64 configurations, each corresponding to one of the evaluated tasks. The dataset is constructed from two runs, where each run is represented as a dedicated split within each configuration. The `train` split always references the most up-to-date results. Furthermore, a `results` configuration is included to store the aggregated results across all runs, which is utilized for computing and presenting the metrics displayed on the Open LLM Leaderboard. The accompanying files also provide examples demonstrating how to load the dataset using the Hugging Face Datasets library.
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

数据集摘要

该数据集是在对模型 harborwater/open-llama-3b-everythingLM-2048 进行评估时自动创建的,评估结果展示在 Open LLM Leaderboard 上。

数据集结构

  • 数据集包含 64 个配置,每个配置对应一个评估任务。
  • 数据集从 2 次运行中创建,每次运行在每个配置中作为一个特定的分割存在,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_harborwater__open-llama-3b-everythingLM-2048", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-24T01:01:11.414021 运行的最新结果: python { "all": { "em": 0.0014681208053691276, "em_stderr": 0.00039210421902986076, "f1": 0.053537122483221615, "f1_stderr": 0.0012879336042021898, "acc": 0.3390732138444075, "acc_stderr": 0.008325489359560807 }, "harness|drop|3": { "em": 0.0014681208053691276, "em_stderr": 0.00039210421902986076, "f1": 0.053537122483221615, "f1_stderr": 0.0012879336042021898 }, "harness|gsm8k|5": { "acc": 0.015163002274450341, "acc_stderr": 0.003366022949726365 }, "harness|winogrande|5": { "acc": 0.6629834254143646, "acc_stderr": 0.01328495576939525 } }

配置详情

以下是数据集的配置详情:

配置列表

  • harness_arc_challenge_25
  • harness_drop_3
  • harness_gsm8k_5
  • harness_hellaswag_10
  • harness_hendrycksTest_5
  • harness_hendrycksTest_abstract_algebra_5
  • harness_hendrycksTest_anatomy_5
  • harness_hendrycksTest_astronomy_5
  • harness_hendrycksTest_business_ethics_5
  • harness_hendrycksTest_clinical_knowledge_5
  • harness_hendrycksTest_college_biology_5
  • harness_hendrycksTest_college_chemistry_5
  • harness_hendrycksTest_college_computer_science_5
  • harness_hendrycksTest_college_mathematics_5
  • harness_hendrycksTest_college_medicine_5
  • harness_hendrycksTest_college_physics_5
  • harness_hendrycksTest_computer_security_5
  • harness_hendrycksTest_conceptual_physics_5
  • harness_hendrycksTest_econometrics_5
  • harness_hendrycksTest_electrical_engineering_5
  • harness_hendrycksTest_elementary_mat

数据文件路径

每个配置包含不同分割的数据文件路径,例如:

  • harness_arc_challenge_25 配置包含 2023_10_04T08_05_25.924210latest 分割的数据文件路径。
  • harness_drop_3 配置包含 2023_10_24T01_01_11.414021latest 分割的数据文件路径。

具体数据文件路径如下:

  • harness_arc_challenge_25:
    • 2023_10_04T08_05_25.924210: **/details_harness|arc:challenge|25_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|arc:challenge|25_2023-10-04T08-05-25.924210.parquet
  • harness_drop_3:
    • 2023_10_24T01_01_11.414021: **/details_harness|drop|3_2023-10-24T01-01-11.414021.parquet
    • latest: **/details_harness|drop|3_2023-10-24T01-01-11.414021.parquet
  • harness_gsm8k_5:
    • 2023_10_24T01_01_11.414021: **/details_harness|gsm8k|5_2023-10-24T01-01-11.414021.parquet
    • latest: **/details_harness|gsm8k|5_2023-10-24T01-01-11.414021.parquet
  • harness_hellaswag_10:
    • 2023_10_04T08_05_25.924210: **/details_harness|hellaswag|10_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hellaswag|10_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_5:
    • 2023_10_04T08_05_25.924210: 多个文件路径
    • latest: 多个文件路径
  • harness_hendrycksTest_abstract_algebra_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-abstract_algebra|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-abstract_algebra|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_anatomy_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-anatomy|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-anatomy|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_astronomy_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-astronomy|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-astronomy|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_business_ethics_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-business_ethics|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-business_ethics|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_clinical_knowledge_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_college_biology_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-college_biology|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-college_biology|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_college_chemistry_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-college_chemistry|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-college_chemistry|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_college_computer_science_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-college_computer_science|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-college_computer_science|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_college_mathematics_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-college_mathematics|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-college_mathematics|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_college_medicine_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-college_medicine|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-college_medicine|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_college_physics_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-college_physics|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-college_physics|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_computer_security_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-computer_security|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-computer_security|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_conceptual_physics_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-conceptual_physics|5_2023-10-04T08-05-25.924210.parquet
    • latest: **/details_harness|hendrycksTest-conceptual_physics|5_2023-10-04T08-05-25.924210.parquet
  • harness_hendrycksTest_econometrics_5:
    • 2023_10_04T08_05_25.924210: **/details_harness|hendrycksTest-econometrics|5_2023-10-04T08-05-25.924210.parquet
    • latest: `**/details_harness|hendrycksTest-econometrics|5_2023-10-04T08-05-25
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