five

open-llm-leaderboard-old/details_h2oai__h2ogpt-gm-oasst1-en-1024-20b

收藏
Hugging Face2023-10-17 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_h2oai__h2ogpt-gm-oasst1-en-1024-20b
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`h2oai/h2ogpt-gm-oasst1-en-1024-20b`进行评估时自动生成的。数据集由64个配置组成,每个配置对应一个评估任务。数据集由2次运行生成,每次运行的结果存储在不同的split中,split以运行的时间戳命名。此外,还有一个名为`results`的配置,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`h2oai/h2ogpt-gm-oasst1-en-1024-20b`进行评估时自动生成的。数据集由64个配置组成,每个配置对应一个评估任务。数据集由2次运行生成,每次运行的结果存储在不同的split中,split以运行的时间戳命名。此外,还有一个名为`results`的配置,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

数据集简介

该数据集是在评估模型h2oai/h2ogpt-gm-oasst1-en-1024-20bOpen LLM Leaderboard上的运行过程中自动创建的。

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。
  • 数据集由2次运行创建,每次运行可以在每个配置中找到特定的分割,分割名称使用运行的时间戳。
  • "train"分割始终指向最新的结果。
  • 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Open LLM Leaderboard上的聚合指标。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_h2oai__h2ogpt-gm-oasst1-en-1024-20b", "harness_winogrande_5", split="train")

最新结果

以下是2023-10-17T11:30:13.056869的最新结果: python { "all": { "em": 0.007340604026845637, "em_stderr": 0.0008741896875346127, "f1": 0.059098154362416305, "f1_stderr": 0.001498238863999793, "acc": 0.3440011895836038, "acc_stderr": 0.008793500407979769 }, "harness|drop|3": { "em": 0.007340604026845637, "em_stderr": 0.0008741896875346127, "f1": 0.059098154362416305, "f1_stderr": 0.001498238863999793 }, "harness|gsm8k|5": { "acc": 0.025018953752843062, "acc_stderr": 0.004302045046564285 }, "harness|winogrande|5": { "acc": 0.6629834254143646, "acc_stderr": 0.013284955769395252 } }

配置详情

  • harness_arc_challenge_25

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|arc:challenge|25_2023-07-19T21:35:35.780060.parquet
  • harness_drop_3

    • 分割: 2023_10_17T11_30_13.056869, latest
    • 路径: **/details_harness|drop|3_2023-10-17T11-30-13.056869.parquet
  • harness_gsm8k_5

    • 分割: 2023_10_17T11_30_13.056869, latest
    • 路径: **/details_harness|gsm8k|5_2023-10-17T11-30-13.056869.parquet
  • harness_hellaswag_10

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hellaswag|10_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: 多个路径,例如**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_anatomy_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_astronomy_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_college_biology_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_college_physics_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_computer_security_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_econometrics_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T21:35:35.780060.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分割: 2023_07_19T21_35_35.780060, latest
    • 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T21:35:35.780060.parquet
二维码
社区交流群
二维码
科研交流群
商业服务