five

open-llm-leaderboard/details_mosaicml__mpt-7b-chat

收藏
Hugging Face2023-10-17 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_mosaicml__mpt-7b-chat
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mosaicml/mpt-7b-chat在Open LLM Leaderboard上的评估运行期间自动创建的。数据集由64个配置组成,每个配置对应一个特定的评估任务。数据集是从2次运行中生成的,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。一个名为results的额外配置存储了所有运行的聚合结果,这些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`load_dataset`函数加载运行中的详细信息的示例。

This dataset was automatically created during the evaluation runs of the model mosaicml/mpt-7b-chat on the Open LLM Leaderboard. The dataset consists of 64 configurations, each corresponding to a specific evaluation task. The dataset is generated from two runs, where each run is represented as a specific split for each configuration, with the split name using the timestamp of the run. The train split always points to the most recent results. An additional configuration named "results" stores the aggregated results of all runs, which are used to calculate and display the aggregate metrics on the Open LLM Leaderboard. The README also provides examples of how to use the `load_dataset` function to load detailed information from the runs.
提供机构:
open-llm-leaderboard
原始信息汇总

数据集概述

数据集来源

该数据集是在评估模型 mosaicml/mpt-7b-chatOpen LLM Leaderboard 上的运行过程中自动创建的。

数据集结构

  • 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找到一个特定的分割,分割名称使用运行的时间戳。
  • "train" 分割始终指向最新的结果。
  • 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_mosaicml__mpt-7b-chat", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-17T09:38:22.163645 运行的最新结果: python { "all": { "em": 0.06952600671140939, "em_stderr": 0.002604746204517829, "f1": 0.12196937919463072, "f1_stderr": 0.002840521979064293, "acc": 0.3626168565432783, "acc_stderr": 0.009260585769647573 }, "harness|drop|3": { "em": 0.06952600671140939, "em_stderr": 0.002604746204517829, "f1": 0.12196937919463072, "f1_stderr": 0.002840521979064293 }, "harness|gsm8k|5": { "acc": 0.04094010614101592, "acc_stderr": 0.005458076796294338 }, "harness|winogrande|5": { "acc": 0.6842936069455406, "acc_stderr": 0.01306309474300081 } }

配置详情

  • harness_arc_challenge_25

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|arc:challenge|25_2023-07-20T10:00:41.356813.parquet
  • harness_drop_3

    • 分割:2023_10_17T09_38_22.163645, latest
    • 路径:**/details_harness|drop|3_2023-10-17T09-38-22.163645.parquet
  • harness_gsm8k_5

    • 分割:2023_10_17T09_38_22.163645, latest
    • 路径:**/details_harness|gsm8k|5_2023-10-17T09-38-22.163645.parquet
  • harness_hellaswag_10

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hellaswag|10_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_anatomy_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_astronomy_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_college_biology_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_college_physics_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_computer_security_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_econometrics_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-20T10:00:41.356813.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分割:2023_07_20T10_00_41.356813, latest
    • 路径:**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-20T10:00:41.356813.parquet

以上是数据集的详细概述,包括数据集的来源、结构、加载示例、最新结果以及各个配置的详细信息。

二维码
社区交流群
二维码
科研交流群
商业服务