five

open-llm-leaderboard-old/details_gaodrew__gaodrew-llama-30b-instruct-2048-Open-Platypus-100steps

收藏
Hugging Face2023-10-01 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_gaodrew__gaodrew-llama-30b-instruct-2048-Open-Platypus-100steps
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 gaodrew/gaodrew-llama-30b-instruct-2048-Open-Platypus-100steps 在 Open LLM Leaderboard 上的评估过程中自动生成的。数据集包含 64 个配置,每个配置对应一个被评估的任务。它由 2 次运行生成,每次运行在每个配置中表示为特定的分割,分割名称由运行的时间戳命名。train 分割始终指向最新的结果。此外,还有一个 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。README 还提供了如何使用 Python 代码加载数据集详细信息的示例,并包含了特定运行的最新结果。

该数据集是在模型 gaodrew/gaodrew-llama-30b-instruct-2048-Open-Platypus-100steps 在 Open LLM Leaderboard 上的评估过程中自动生成的。数据集包含 64 个配置,每个配置对应一个被评估的任务。它由 2 次运行生成,每次运行在每个配置中表示为特定的分割,分割名称由运行的时间戳命名。train 分割始终指向最新的结果。此外,还有一个 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。README 还提供了如何使用 Python 代码加载数据集详细信息的示例,并包含了特定运行的最新结果。
提供机构:
open-llm-leaderboard-old
原始信息汇总

数据集概述

该数据集是在评估模型 gaodrew/gaodrew-llama-30b-instruct-2048-Open-Platypus-100stepsOpen LLM Leaderboard 上的自动创建的。

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2次运行中创建,每个运行可以在每个配置中作为一个特定的分割找到,分割名称使用运行的时间戳。
  • "train" 分割始终指向最新的结果。

额外配置

  • 一个额外的配置 "results" 存储所有运行的聚合结果,用于计算和显示在 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_gaodrew__gaodrew-llama-30b-instruct-2048-Open-Platypus-100steps", "harness_winogrande_5", split="train")

最新结果

这些是最新结果(来自2023-10-01T22:04:43.928284运行)的示例: python { "all": { "em": 0.0036703020134228187, "em_stderr": 0.0006192871806511078, "f1": 0.07745071308724842, "f1_stderr": 0.0016031429592015417, "acc": 0.4924286713583812, "acc_stderr": 0.01078503608525705 }, "harness|drop|3": { "em": 0.0036703020134228187, "em_stderr": 0.0006192871806511078, "f1": 0.07745071308724842, "f1_stderr": 0.0016031429592015417 }, "harness|gsm8k|5": { "acc": 0.17664897649734648, "acc_stderr": 0.01050486250585457 }, "harness|winogrande|5": { "acc": 0.8082083662194159, "acc_stderr": 0.011065209664659527 } }

配置详情

  • config_name: harness_arc_challenge_25

    • split: 2023_08_19T01_39_27.936729
      • path: **/details_harness|arc:challenge|25_2023-08-19T01:39:27.936729.parquet
    • split: latest
      • path: **/details_harness|arc:challenge|25_2023-08-19T01:39:27.936729.parquet
  • config_name: harness_drop_3

    • split: 2023_10_01T22_04_43.928284
      • path: **/details_harness|drop|3_2023-10-01T22-04-43.928284.parquet
    • split: latest
      • path: **/details_harness|drop|3_2023-10-01T22-04-43.928284.parquet
  • config_name: harness_gsm8k_5

    • split: 2023_10_01T22_04_43.928284
      • path: **/details_harness|gsm8k|5_2023-10-01T22-04-43.928284.parquet
    • split: latest
      • path: **/details_harness|gsm8k|5_2023-10-01T22-04-43.928284.parquet
  • config_name: harness_hellaswag_10

    • split: 2023_08_19T01_39_27.936729
      • path: **/details_harness|hellaswag|10_2023-08-19T01:39:27.936729.parquet
    • split: latest
      • path: **/details_harness|hellaswag|10_2023-08-19T01:39:27.936729.parquet
  • config_name: harness_hendrycksTest_5

    • split: 2023_08_19T01_39_27.936729
      • path:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_biology|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_chemistry|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_computer_science|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_european_history|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_geography|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_mathematics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_microeconomics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_physics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_psychology|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_statistics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_us_history|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-high_school_world_history|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-human_aging|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-human_sexuality|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-international_law|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-jurisprudence|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-logical_fallacies|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-machine_learning|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-management|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-marketing|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-medical_genetics|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-miscellaneous|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-moral_disputes|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-moral_scenarios|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-nutrition|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-philosophy|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-prehistory|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-professional_accounting|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-professional_law|5_2023-08-19T01:39:27.936729.parquet
        • **/details_harness|hendrycksTest-professional_medicine|5_2
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