five

open-llm-leaderboard/details_CHIH-HUNG__llama-2-13b-huangyt_FINETUNE2_3w-gate_up_down_proj

收藏
Hugging Face2023-10-19 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_CHIH-HUNG__llama-2-13b-huangyt_FINETUNE2_3w-gate_up_down_proj
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[CHIH-HUNG/llama-2-13b-huangyt_FINETUNE2_3w-gate_up_down_proj](https://huggingface.co/CHIH-HUNG/llama-2-13b-huangyt_FINETUNE2_3w-gate_up_down_proj)在[Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)上的评估运行过程中自动创建的。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次运行创建,每次运行的结果都存储为特定配置中的一个分割,分割名称使用运行的时间戳。train分割始终指向最新的结果。此外,还有一个名为results的配置存储了所有运行的聚合结果,用于计算和显示[Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)上的聚合指标。

本数据集是在模型[CHIH-HUNG/llama-2-13b-huangyt_FINETUNE2_3w-gate_up_down_proj](https://huggingface.co/CHIH-HUNG/llama-2-13b-huangyt_FINETUNE2_3w-gate_up_down_proj)于[开放大语言模型排行榜(Open LLM Leaderboard)](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)上开展评估运行的过程中自动生成的。本数据集共包含64个配置项,每个配置项对应一项独立的评估任务。本数据集通过两次评估运行生成,每次运行的结果均被存储为对应配置下的一个数据拆分,数据拆分的名称采用该次运行的时间戳。训练集拆分始终指向最新的评估结果。此外,还设有一个名为`results`的配置项,用于存储所有评估运行的聚合结果,以供开放大语言模型排行榜计算并展示其聚合评估指标。
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

数据集来源

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" 存储所有运行结果的聚合,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_CHIH-HUNG__llama-2-13b-huangyt_FINETUNE2_3w-gate_up_down_proj", "harness_winogrande_5", split="train")

最新结果

  • 最新结果来自 2023-10-19T01:12:50.738411 运行,详细结果如下: python { "all": { "em": 0.08137583892617449, "em_stderr": 0.0027999889835206245, "f1": 0.13315016778523447, "f1_stderr": 0.0029419319985989354, "acc": 0.4454347335673805, "acc_stderr": 0.010395126943573653 }, "harness|drop|3": { "em": 0.08137583892617449, "em_stderr": 0.0027999889835206245, "f1": 0.13315016778523447, "f1_stderr": 0.0029419319985989354 }, "harness|gsm8k|5": { "acc": 0.12054586808188021, "acc_stderr": 0.008968608285309067 }, "harness|winogrande|5": { "acc": 0.7703235990528808, "acc_stderr": 0.011821645601838238 } }

配置详情

  • harness_arc_challenge_25

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|arc:challenge|25_2023-09-02T04:13:58.324897.parquet
  • harness_drop_3

    • 分割: 2023_10_19T01_12_50.738411, latest
    • 路径: **/details_harness|drop|3_2023-10-19T01-12-50.738411.parquet
  • harness_gsm8k_5

    • 分割: 2023_10_19T01_12_50.738411, latest
    • 路径: **/details_harness|gsm8k|5_2023-10-19T01-12-50.738411.parquet
  • harness_hellaswag_10

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hellaswag|10_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: 多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_anatomy_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_astronomy_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_college_biology_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_college_physics_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_computer_security_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_econometrics_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-09-02T04:13:58.324897.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分割: 2023_09_02T04_13_58.324897, latest
    • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-09-02T04:13:58.324897.parquet
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