five

open-llm-leaderboard/details_lgaalves__llama-2-13b-hf-platypus

收藏
Hugging Face2023-10-28 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_lgaalves__llama-2-13b-hf-platypus
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在评估模型[lgaalves/llama-2-13b-hf-platypus](https://huggingface.co/lgaalves/llama-2-13b-hf-platypus)时自动创建的,评估是在[Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)上进行的。数据集由64个配置组成,每个配置对应一个评估任务。数据集是从3次运行中创建的,每次运行都可以在每个配置的特定分割中找到,分割名称使用运行的时间戳命名。"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,并用于计算和显示[Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)上的聚合指标。
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

数据集来源

该数据集是在评估模型 lgaalves/llama-2-13b-hf-platypusOpen LLM Leaderboard 上的自动创建的。

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_lgaalves__llama-2-13b-hf-platypus", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-28T02:33:59.939371 运行的最新结果: python { "all": { "em": 0.0017827181208053692, "em_stderr": 0.00043200973460388544, "f1": 0.05985213926174496, "f1_stderr": 0.0013641672120704657, "acc": 0.4325617395685546, "acc_stderr": 0.009923090021448928 }, "harness|drop|3": { "em": 0.0017827181208053692, "em_stderr": 0.00043200973460388544, "f1": 0.05985213926174496, "f1_stderr": 0.0013641672120704657 }, "harness|gsm8k|5": { "acc": 0.09401061410159212, "acc_stderr": 0.00803881981887246 }, "harness|winogrande|5": { "acc": 0.771112865035517, "acc_stderr": 0.011807360224025398 } }

配置详情

  • harness_arc_challenge_25

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|arc:challenge|25_2023-09-18T14-15-46.670153.parquet
  • harness_drop_3

    • 分割: 2023_10_28T00_17_42.072889, 2023_10_28T02_33_59.939371, latest
    • 路径: **/details_harness|drop|3_2023-10-28T00-17-42.072889.parquet, **/details_harness|drop|3_2023-10-28T02-33-59.939371.parquet
  • harness_gsm8k_5

    • 分割: 2023_10_28T00_17_42.072889, 2023_10_28T02_33_59.939371, latest
    • 路径: **/details_harness|gsm8k|5_2023-10-28T00-17-42.072889.parquet, **/details_harness|gsm8k|5_2023-10-28T02-33-59.939371.parquet
  • harness_hellaswag_10

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hellaswag|10_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: 多个路径,详见原文
  • harness_hendrycksTest_abstract_algebra_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_anatomy_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_astronomy_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_college_biology_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_college_physics_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_computer_security_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_econometrics_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-09-18T14-15-46.670153.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分割: 2023_09_18T14_15_46.670153, latest
    • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-09-18T14-15-46.670153.parquet
二维码
社区交流群
二维码
科研交流群
商业服务