five

open-llm-leaderboard-old/details_jzjiao__opt-1.3b-rlhf

收藏
Hugging Face2023-12-03 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_jzjiao__opt-1.3b-rlhf
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jzjiao/opt-1.3b-rlhf进行评估时自动生成的。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3次运行生成,每次运行的结果作为特定分割存储在配置中,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,用于计算和显示Open LLM Leaderboard上的聚合指标。

该数据集是在Open LLM Leaderboard上对模型jzjiao/opt-1.3b-rlhf进行评估时自动生成的。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3次运行生成,每次运行的结果作为特定分割存储在配置中,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,用于计算和显示Open LLM Leaderboard上的聚合指标。
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 jzjiao/opt-1.3b-rlhf 进行评估运行期间自动创建的,用于 Open LLM Leaderboard

数据集组成

  • 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 3 次运行中创建,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_jzjiao__opt-1.3b-rlhf", "harness_gsm8k_5", split="train")

最新结果

以下是 2023-12-03T18:51:18.497076 运行的最新结果: python { "all": { "acc": 0.004548900682335102, "acc_stderr": 0.0018535550440036204 }, "harness|gsm8k|5": { "acc": 0.004548900682335102, "acc_stderr": 0.0018535550440036204 } }

配置详情

  • harness_arc_challenge_25

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|arc:challenge|25_2023-07-19T14:36:48.435460.parquet
  • harness_drop_3

    • 分割:2023_10_19T11_35_35.582368, latest
    • 路径:**/details_harness|drop|3_2023-10-19T11-35-35.582368.parquet
  • harness_gsm8k_5

    • 分割:2023_10_19T11_35_35.582368, 2023_12_03T18_51_18.497076, latest
    • 路径:**/details_harness|gsm8k|5_2023-10-19T11-35-35.582368.parquet, **/details_harness|gsm8k|5_2023-12-03T18-51-18.497076.parquet
  • harness_hellaswag_10

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hellaswag|10_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:多个路径,如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_anatomy_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_astronomy_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_college_biology_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_college_physics_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_computer_security_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_econometrics_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T14:36:48.435460.parquet
  • harness_hendrycksTest_formal_logic_5

    • 分割:2023_07_19T14_36_48.435460, latest
    • 路径:**/details_harness|hendrycksTest-formal_logic|5_2023-07-19T14:36:48.435460.parquet

以上是数据集的详细概述,包括数据集的组成、加载示例、最新结果以及各个配置的详细信息。

二维码
社区交流群
二维码
科研交流群
商业服务