five

open-llm-leaderboard/details_IkariDev__Athena-v4

收藏
Hugging Face2023-10-25 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_IkariDev__Athena-v4
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在Open LLM Leaderboard上对模型IkariDev/Athena-v4进行评估运行时自动创建的。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,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上的聚合指标。README还提供了如何使用`load_dataset`函数加载运行中的详细信息的示例。
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

该数据集是在对模型 IkariDev/Athena-v4 进行评估运行期间自动创建的,用于 Open LLM Leaderboard

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片找到,分片名称使用运行的时间戳。
  • "train" 分片总是指向最新的结果。
  • 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_IkariDev__Athena-v4", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-25T17:00:26.530924 运行的最新结果: python { "all": { "em": 0.05432046979865772, "em_stderr": 0.002321097609357669, "f1": 0.13087562919463042, "f1_stderr": 0.0026936499511124616, "acc": 0.44229322757129275, "acc_stderr": 0.010432110783601959 }, "harness|drop|3": { "em": 0.05432046979865772, "em_stderr": 0.002321097609357669, "f1": 0.13087562919463042, "f1_stderr": 0.0026936499511124616 }, "harness|gsm8k|5": { "acc": 0.1197877179681577, "acc_stderr": 0.008944213403553046 }, "harness|winogrande|5": { "acc": 0.7647987371744278, "acc_stderr": 0.011920008163650872 } }

配置详情

  • harness_arc_challenge_25

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|arc:challenge|25_2023-10-09T09-29-40.768179.parquet
  • harness_drop_3

    • 分片:2023_10_25T17_00_26.530924, latest
    • 路径:**/details_harness|drop|3_2023-10-25T17-00-26.530924.parquet
  • harness_gsm8k_5

    • 分片:2023_10_25T17_00_26.530924, latest
    • 路径:**/details_harness|gsm8k|5_2023-10-25T17-00-26.530924.parquet
  • harness_hellaswag_10

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hellaswag|10_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_anatomy_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_astronomy_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_college_biology_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_college_physics_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_computer_security_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_econometrics_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-10-09T09-29-40.768179.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分片:2023_10_09T09_29_40.768179, latest
    • 路径:**/details_harness|hendrycksTest-elementary_mathematics|5_2023-10-09T09-29-40.768179.parquet

以上是数据集的概述和详细配置信息。

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