five

open-llm-leaderboard-old/details_haonan-li__bactrian-x-llama-13b-merged

收藏
Hugging Face2023-09-18 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_haonan-li__bactrian-x-llama-13b-merged
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haonan-li/bactrian-x-llama-13b-merged的评估运行中自动创建的。数据集由64个配置组成,每个配置对应一个评估任务。数据集是从2次运行中创建的,每次运行都可以在特定配置中找到,分割名称使用运行的时间戳。train分割始终指向最新的结果。此外,还有一个results配置,存储了所有运行的聚合结果,并用于计算和显示在Open LLM Leaderboard上的聚合指标。

该数据集是在模型haonan-li/bactrian-x-llama-13b-merged的评估运行中自动创建的。数据集由64个配置组成,每个配置对应一个评估任务。数据集是从2次运行中创建的,每次运行都可以在特定配置中找到,分割名称使用运行的时间戳。train分割始终指向最新的结果。此外,还有一个results配置,存储了所有运行的聚合结果,并用于计算和显示在Open LLM Leaderboard上的聚合指标。
提供机构:
open-llm-leaderboard-old
原始信息汇总

数据集概述

数据集简介

该数据集是在评估模型haonan-li/bactrian-x-llama-13b-mergedOpen LLM Leaderboard上的自动创建的。

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次运行结果组成,每次运行的结果作为特定分片存储,分片名称使用运行的时间戳。
  • 最新结果:"train"分片始终指向最新的结果。
  • 结果汇总: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_haonan-li__bactrian-x-llama-13b-merged", "harness_winogrande_5", split="train")

最新结果

以下是2023-09-18T01:46:24.914160的最新结果: python { "all": { "em": 0.2553481543624161, "em_stderr": 0.004465629087714431, "f1": 0.31091652684563814, "f1_stderr": 0.004443751758152442, "acc": 0.3974435920159074, "acc_stderr": 0.009316527734088942 }, "harness|drop|3": { "em": 0.2553481543624161, "em_stderr": 0.004465629087714431, "f1": 0.31091652684563814, "f1_stderr": 0.004443751758152442 }, "harness|gsm8k|5": { "acc": 0.05534495830174375, "acc_stderr": 0.006298221796179595 }, "harness|winogrande|5": { "acc": 0.739542225730071, "acc_stderr": 0.012334833671998289 } }

配置详情

  • harness_arc_challenge_25

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|arc:challenge|25_2023-07-19T19:16:59.483464.parquet
  • harness_drop_3

    • 分片:2023_09_18T01_46_24.914160, latest
    • 路径:**/details_harness|drop|3_2023-09-18T01-46-24.914160.parquet
  • harness_gsm8k_5

    • 分片:2023_09_18T01_46_24.914160, latest
    • 路径:**/details_harness|gsm8k|5_2023-09-18T01-46-24.914160.parquet
  • harness_hellaswag_10

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hellaswag|10_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:多个路径,例如**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_anatomy_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_astronomy_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_college_biology_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_college_physics_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_computer_security_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_econometrics_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T19:16:59.483464.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分片:2023_07_19T19_16_59.483464, latest
    • 路径:**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T19:16:59.483464.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。

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