five

open-llm-leaderboard/details_vibhorag101__llama-2-7b-chat-hf-phr_mental_health-2048

收藏
Hugging Face2023-10-23 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_vibhorag101__llama-2-7b-chat-hf-phr_mental_health-2048
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是在评估模型vibhorag101/llama-2-7b-chat-hf-phr_mental_health-2048时自动生成的,包含了64个配置,每个配置对应一个评估任务。数据集由2次运行生成,每次运行的结果存储在不同的split中,最新的结果存储在train split中。此外,还有一个results配置用于存储所有运行的聚合结果。

该数据集是在评估模型vibhorag101/llama-2-7b-chat-hf-phr_mental_health-2048时自动生成的,包含了64个配置,每个配置对应一个评估任务。数据集由2次运行生成,每次运行的结果存储在不同的split中,最新的结果存储在train split中。此外,还有一个results配置用于存储所有运行的聚合结果。
提供机构:
open-llm-leaderboard
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 vibhorag101/llama-2-7b-chat-hf-phr_mental_health-2048Open LLM Leaderboard 上的运行过程中自动创建的。

数据集结构

  • 该数据集包含 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,每次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" 分片始终指向最新的结果。
  • 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_vibhorag101__llama-2-7b-chat-hf-phr_mental_health-2048", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-23T12:14:06.343612 运行的最新结果: python { "all": { "em": 0.06291946308724833, "em_stderr": 0.0024866840810898265, "f1": 0.1230484479865771, "f1_stderr": 0.002835124699649658, "acc": 0.3855268126040811, "acc_stderr": 0.00961257569233847 }, "harness|drop|3": { "em": 0.06291946308724833, "em_stderr": 0.0024866840810898265, "f1": 0.1230484479865771, "f1_stderr": 0.002835124699649658 }, "harness|gsm8k|5": { "acc": 0.05913570887035633, "acc_stderr": 0.006497266660428824 }, "harness|winogrande|5": { "acc": 0.7119179163378059, "acc_stderr": 0.012727884724248116 } }

配置详情

以下是数据集的配置详情:

  • harness_arc_challenge_25

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|arc:challenge|25_2023-09-16T15-28-08.035226.parquet
  • harness_drop_3

    • 分片: 2023_10_23T12_14_06.343612, latest
    • 路径: **/details_harness|drop|3_2023-10-23T12-14-06.343612.parquet
  • harness_gsm8k_5

    • 分片: 2023_10_23T12_14_06.343612, latest
    • 路径: **/details_harness|gsm8k|5_2023-10-23T12-14-06.343612.parquet
  • harness_hellaswag_10

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hellaswag|10_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: 多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_anatomy_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_astronomy_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_college_biology_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_college_physics_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_computer_security_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_econometrics_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-09-16T15-28-08.035226.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分片: 2023_09_16T15_28_08.035226, latest
    • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-09-16T15-28-08.035226.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、最新结果以及配置详情。

二维码
社区交流群
二维码
科研交流群
商业服务