five

open-llm-leaderboard/details_TFLai__OrcaMini-Platypus2-13B-QLoRA-0.80-epoch

收藏
Hugging Face2023-10-18 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_TFLai__OrcaMini-Platypus2-13B-QLoRA-0.80-epoch
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在模型TFLai/OrcaMini-Platypus2-13B-QLoRA-0.80-epoch在Open LLM Leaderboard上的评估运行期间自动创建的。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,每次运行在每个配置中表示为特定的分割。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,这些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中的`datasets`库加载运行细节的示例。
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

数据集来源

该数据集是在对模型 TFLai/OrcaMini-Platypus2-13B-QLoRA-0.80-epoch 进行评估运行期间自动创建的,评估结果发布在 Open LLM Leaderboard 上。

数据集结构

  • 配置数量:数据集包含 64 个配置,每个配置对应一个评估任务。
  • 运行次数:数据集从 2 次运行中创建。每个运行在每个配置中作为一个特定的分片存在,分片名称使用运行的时间戳。
  • 训练分片:"train" 分片始终指向最新的结果。
  • 结果配置: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_TFLai__OrcaMini-Platypus2-13B-QLoRA-0.80-epoch", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-18T23:54:59.357050 运行的最新结果: python { "all": { "em": 0.40216023489932884, "em_stderr": 0.005021478569413354, "f1": 0.47240666946308846, "f1_stderr": 0.004780752235261512, "acc": 0.39100918935382517, "acc_stderr": 0.008061089924986945 }, "harness|drop|3": { "em": 0.40216023489932884, "em_stderr": 0.005021478569413354, "f1": 0.47240666946308846, "f1_stderr": 0.004780752235261512 }, "harness|gsm8k|5": { "acc": 0.022744503411675512, "acc_stderr": 0.004106620637749706 }, "harness|winogrande|5": { "acc": 0.7592738752959748, "acc_stderr": 0.012015559212224183 } }

配置详情

  • harness_arc_challenge_25

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|arc:challenge|25_2023-08-28T22:52:27.560095.parquet
  • harness_drop_3

    • 分片:2023_10_18T23_54_59.357050, latest
    • 路径:**/details_harness|drop|3_2023-10-18T23-54-59.357050.parquet
  • harness_gsm8k_5

    • 分片:2023_10_18T23_54_59.357050, latest
    • 路径:**/details_harness|gsm8k|5_2023-10-18T23-54-59.357050.parquet
  • harness_hellaswag_10

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hellaswag|10_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:多个文件路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_anatomy_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_astronomy_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_college_biology_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_college_physics_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_computer_security_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_econometrics_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-28T22:52:27.560095.parquet
  • harness_hendrycksTest_elementary_m

    • 分片:2023_08_28T22_52_27.560095, latest
    • 路径:**/details_harness|hendrycksTest-elementary_m|5_2023-08-28T22:52:27.560095.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