five

open-llm-leaderboard/details_TFLai__Nous-Hermes-Platypus2-13B-QLoRA-0.80-epoch

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_TFLai__Nous-Hermes-Platypus2-13B-QLoRA-0.80-epoch
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TFLai/Nous-Hermes-Platypus2-13B-QLoRA-0.80-epoch时自动生成的,主要用于Open LLM Leaderboard的评估任务。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次运行生成,每次运行的结果作为一个特定的分割,分割名称使用运行的时间戳。此外,数据集还包含一个名为“results”的配置,用于存储所有运行的聚合结果,并在Open LLM Leaderboard上显示聚合指标。
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

数据集简介

该数据集是在对模型 TFLai/Nous-Hermes-Platypus2-13B-QLoRA-0.80-epoch 进行评估运行期间自动创建的,用于 Open LLM Leaderboard

数据集结构

  • 配置数量:64个配置,每个配置对应一个评估任务。
  • 数据来源:数据集由2次运行创建,每次运行的结果可以在每个配置中找到,以运行的时间戳命名的特定分片形式存储。
  • 最新结果:"train" 分片始终指向最新的结果。
  • 汇总结果:一个额外的配置 "results" 存储所有运行的汇总结果,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_TFLai__Nous-Hermes-Platypus2-13B-QLoRA-0.80-epoch", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-19T04:16:45.714438 运行 的最新结果: python { "all": { "em": 0.345008389261745, "em_stderr": 0.004868244118482663, "f1": 0.4264691694630892, "f1_stderr": 0.004672170372384348, "acc": 0.3832876668064886, "acc_stderr": 0.007708220968501149 }, "harness|drop|3": { "em": 0.345008389261745, "em_stderr": 0.004868244118482663, "f1": 0.4264691694630892, "f1_stderr": 0.004672170372384348 }, "harness|gsm8k|5": { "acc": 0.014404852160727824, "acc_stderr": 0.003282055917136951 }, "harness|winogrande|5": { "acc": 0.7521704814522494, "acc_stderr": 0.012134386019865348 } }

配置详情

  • harness_arc_challenge_25

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|arc:challenge|25_2023-08-28T22:44:43.350947.parquet
  • harness_drop_3

    • 分片:2023_10_19T04_16_45.714438, latest
    • 路径:**/details_harness|drop|3_2023-10-19T04-16-45.714438.parquet
  • harness_gsm8k_5

    • 分片:2023_10_19T04_16_45.714438, latest
    • 路径:**/details_harness|gsm8k|5_2023-10-19T04-16-45.714438.parquet
  • harness_hellaswag_10

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hellaswag|10_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:多个文件路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_anatomy_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_astronomy_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_college_biology_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_college_physics_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_computer_security_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_econometrics_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-08-28T22:44:43.350947.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分片:2023_08_28T22_44_43.350947, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-28T22:44:43.350947.parquet

以上是数据集的详细概述,包括数据集的简介、结构、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