five

open-llm-leaderboard/details_ausboss__llama-30b-supercot

收藏
Hugging Face2023-10-16 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_ausboss__llama-30b-supercot
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ausboss/llama-30b-supercot进行评估时自动创建的。数据集由64个配置组成,每个配置对应一个评估任务。数据集是从2次运行中创建的,每次运行在每个配置中表示为特定的分割,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,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`datasets`库中的`load_dataset`函数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

该数据集是在评估模型 ausboss/llama-30b-supercotOpen LLM Leaderboard 上的自动创建的。数据集包含 64 个配置,每个配置对应一个评估任务。

数据集结构

  • 配置数量: 64
  • 创建来源: 2 次运行
  • 数据分割: 每个配置包含特定运行的分割,分割名称使用运行的时间戳。"train" 分割始终指向最新结果。
  • 结果配置: 一个额外的 "results" 配置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_ausboss__llama-30b-supercot", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-16T01:12:25.087238 运行的最新结果

python { "all": { "em": 0.11619127516778524, "em_stderr": 0.0032817504432788346, "f1": 0.19070469798657516, "f1_stderr": 0.003401025373096678, "acc": 0.45967263712374484, "acc_stderr": 0.010077515646893735 }, "harness|drop|3": { "em": 0.11619127516778524, "em_stderr": 0.0032817504432788346, "f1": 0.19070469798657516, "f1_stderr": 0.003401025373096678 }, "harness|gsm8k|5": { "acc": 0.11902956785443518, "acc_stderr": 0.008919702911161629 }, "harness|winogrande|5": { "acc": 0.8003157063930545, "acc_stderr": 0.01123532838262584 } }

配置详情

  • harness_arc_challenge_25

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|arc:challenge|25_2023-07-19T22:24:52.456650.parquet
  • harness_drop_3

    • 分割: 2023_10_16T01_12_25.087238, latest
    • 路径: **/details_harness|drop|3_2023-10-16T01-12-25.087238.parquet
  • harness_gsm8k_5

    • 分割: 2023_10_16T01_12_25.087238, latest
    • 路径: **/details_harness|gsm8k|5_2023-10-16T01-12-25.087238.parquet
  • harness_hellaswag_10

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hellaswag|10_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: 多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_anatomy_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_astronomy_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_college_biology_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_college_physics_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_computer_security_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_econometrics_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T22:24:52.456650.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分割: 2023_07_19T22_24_52.456650, latest
    • 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T22:24:52.456650.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