five

open-llm-leaderboard-old/details_ShenaoZhang__0.001_idpo_noreplacerej_iter_3

收藏
Hugging Face2024-04-08 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_ShenaoZhang__0.001_idpo_noreplacerej_iter_3
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ShenaoZhang/0.001_idpo_noreplacerej_iter_3时自动创建的,评估在Open LLM Leaderboard上进行。数据集由63个配置组成,每个配置对应一个评估任务。数据集是从1次运行中创建的,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。

该数据集是在评估模型ShenaoZhang/0.001_idpo_noreplacerej_iter_3时自动创建的,评估在Open LLM Leaderboard上进行。数据集由63个配置组成,每个配置对应一个评估任务。数据集是从1次运行中创建的,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。
提供机构:
open-llm-leaderboard-old
原始信息汇总

数据集概述

该数据集是在模型 ShenaoZhang/0.001_idpo_noreplacerej_iter_3 的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 63 个配置,每个配置对应一个评估任务。
  • 数据集从 1 次运行中创建,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。
  • 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_ShenaoZhang__0.001_idpo_noreplacerej_iter_3", "harness_winogrande_5", split="train")

最新结果

以下是 最新结果来自 run 2024-04-08T07:47:05.218059 的摘要:

python { "all": { "acc": 0.6041152593422084, "acc_stderr": 0.03322479712840997, "acc_norm": 0.6098183840000261, "acc_norm_stderr": 0.033910525157315, "mc1": 0.3843329253365973, "mc1_stderr": 0.0170287073012452, "mc2": 0.5430807001316682, "mc2_stderr": 0.015958299577086922 }, "harness|arc:challenge|25": { "acc": 0.6151877133105802, "acc_stderr": 0.014218371065251102, "acc_norm": 0.6501706484641638, "acc_norm_stderr": 0.013936809212158303 }, "harness|hellaswag|10": { "acc": 0.6686914957179845, "acc_stderr": 0.004697217912463008, "acc_norm": 0.8555068711412069, "acc_norm_stderr": 0.003508705076059111 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.32, "acc_stderr": 0.046882617226215034, "acc_norm": 0.32, "acc_norm_stderr": 0.046882617226215034 }, # 其他任务的结果省略... }

配置详情

  • config_name: harness_arc_challenge_25

    • split: 2024_04_08T07_47_05.218059
      • path: **/details_harness|arc:challenge|25_2024-04-08T07-47-05.218059.parquet
    • split: latest
      • path: **/details_harness|arc:challenge|25_2024-04-08T07-47-05.218059.parquet
  • config_name: harness_gsm8k_5

    • split: 2024_04_08T07_47_05.218059
      • path: **/details_harness|gsm8k|5_2024-04-08T07-47-05.218059.parquet
    • split: latest
      • path: **/details_harness|gsm8k|5_2024-04-08T07-47-05.218059.parquet
  • config_name: harness_hellaswag_10

    • split: 2024_04_08T07_47_05.218059
      • path: **/details_harness|hellaswag|10_2024-04-08T07-47-05.218059.parquet
    • split: latest
      • path: **/details_harness|hellaswag|10_2024-04-08T07-47-05.218059.parquet
  • config_name: harness_hendrycksTest_5

    • split: 2024_04_08T07_47_05.218059
      • path:
        • **/details_harness|hendrycksTest-abstract_algebra|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-anatomy|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-astronomy|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-business_ethics|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-clinical_knowledge|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-college_biology|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-college_chemistry|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-college_computer_science|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-college_mathematics|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-college_medicine|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-college_physics|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-computer_security|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-conceptual_physics|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-econometrics|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-electrical_engineering|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-elementary_mathematics|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-formal_logic|5_2024-04-08T07-47-05.218059.parquet
        • **/details_harness|hendrycksTest-global_facts|5_2024-04-08T07-47-05.218059.parquet

        其他路径省略...

}

二维码
社区交流群
二维码
科研交流群
商业服务