five

open-llm-leaderboard-old/details_TheBloke__llama-2-70b-Guanaco-QLoRA-fp16

收藏
Hugging Face2023-10-22 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_TheBloke__llama-2-70b-Guanaco-QLoRA-fp16
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TheBloke/llama-2-70b-Guanaco-QLoRA-fp16的评估运行期间自动创建的,用于在Open LLM Leaderboard上进行评估。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次运行生成,每次运行可以在每个配置中找到特定的分割,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,还有一个名为"results"的配置存储了所有运行的聚合结果,并用于计算和显示Open LLM Leaderboard上的聚合指标。

该数据集是在模型TheBloke/llama-2-70b-Guanaco-QLoRA-fp16的评估运行期间自动创建的,用于在Open LLM Leaderboard上进行评估。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次运行生成,每次运行可以在每个配置中找到特定的分割,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,还有一个名为"results"的配置存储了所有运行的聚合结果,并用于计算和显示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

该数据集是在对模型 TheBloke/llama-2-70b-Guanaco-QLoRA-fp16 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。
  • 运行次数:数据集由2次运行创建。每次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。
  • 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" 存储所有运行结果的汇总,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_TheBloke__llama-2-70b-Guanaco-QLoRA-fp16", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-22T03:53:16.698758 运行 的最新结果: python { "all": { "em": 0.05620805369127517, "em_stderr": 0.0023587236332230886, "f1": 0.11980180369127513, "f1_stderr": 0.002592264922824749, "acc": 0.5688660001089055, "acc_stderr": 0.011453570865550992 }, "harness|drop|3": { "em": 0.05620805369127517, "em_stderr": 0.0023587236332230886, "f1": 0.11980180369127513, "f1_stderr": 0.002592264922824749 }, "harness|gsm8k|5": { "acc": 0.2979529946929492, "acc_stderr": 0.012597932232914513 }, "harness|winogrande|5": { "acc": 0.8397790055248618, "acc_stderr": 0.010309209498187472 } }

配置详情

以下是数据集的配置详情:

  • harness_arc_challenge_25

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|arc:challenge|25_2023-07-25T19:54:57.592623.parquet
  • harness_drop_3

    • 分片:2023_10_22T03_53_16.698758, latest
    • 路径:**/details_harness|drop|3_2023-10-22T03-53-16.698758.parquet
  • harness_gsm8k_5

    • 分片:2023_10_22T03_53_16.698758, latest
    • 路径:**/details_harness|gsm8k|5_2023-10-22T03-53-16.698758.parquet
  • harness_hellaswag_10

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hellaswag|10_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_anatomy_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_astronomy_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_college_biology_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_college_physics_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_computer_security_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_econometrics_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-25T19:54:57.592623.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分片:2023_07_25T19_54_57.592623, latest
    • 路径:**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-25T19:54:57.592623.parquet

以上是数据集的详细概述,包括数据集的结构、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