open-llm-leaderboard/details_jondurbin__airoboros-33b-gpt4-m2.0

该数据集是在模型 jondurbin/airoboros-33b-gpt4-m2.0 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，每个配置对应一个被评估的任务。数据集是从 4 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。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 还提供了一个示例，展示了如何使用 `datasets` 库中的 `load_dataset` 函数加载运行中的详细信息。

This dataset was automatically created during the evaluation run of the model jondurbin/airoboros-33b-gpt4-m2.0 on the Open LLM Leaderboard. The dataset consists of 64 configurations, each corresponding to one evaluated task. The dataset is compiled from 4 runs, with each run denoted as a specific split under every configuration. The 'train' split always points to the most recent results. An additional configuration named 'results' stores the aggregated results across all runs, which are used to calculate and display the aggregate metrics on the Open LLM Leaderboard. The README also provides an example demonstrating how to use the `load_dataset` function from the `datasets` library to load detailed information from the runs.

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 jondurbin/airoboros-33b-gpt4-m2.0 进行评估运行期间自动创建的，用于 Open LLM Leaderboard。

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 4 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_jondurbin__airoboros-33b-gpt4-m2.0", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-22T05:59:09.159543 运行的最新结果： python { "all": { "em": 0.30432046979865773, "em_stderr": 0.004712049527083924, "f1": 0.37717596476510223, "f1_stderr": 0.00456045095000614, "acc": 0.4400130926002275, "acc_stderr": 0.009847939494812614 }, "harness|drop|3": { "em": 0.30432046979865773, "em_stderr": 0.004712049527083924, "f1": 0.37717596476510223, "f1_stderr": 0.00456045095000614 }, "harness|gsm8k|5": { "acc": 0.09628506444275967, "acc_stderr": 0.008125264128215886 }, "harness|winogrande|5": { "acc": 0.7837411207576953, "acc_stderr": 0.011570614861409345 } }

配置详情

• harness_arc_challenge_25

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|arc:challenge|25_2023-08-02T16:13:19.014173.parquet

• harness_drop_3

  • 分割：2023_10_19T10_51_22.664215, 2023_10_21T18_09_50.123692, 2023_10_22T05_59_09.159543, latest
  • 路径：**/details_harness|drop|3_2023-10-19T10-51-22.664215.parquet, **/details_harness|drop|3_2023-10-21T18-09-50.123692.parquet, **/details_harness|drop|3_2023-10-22T05-59-09.159543.parquet

• harness_gsm8k_5

  • 分割：2023_10_19T10_51_22.664215, 2023_10_21T18_09_50.123692, 2023_10_22T05_59_09.159543, latest
  • 路径：**/details_harness|gsm8k|5_2023-10-19T10-51-22.664215.parquet, **/details_harness|gsm8k|5_2023-10-21T18-09-50.123692.parquet, **/details_harness|gsm8k|5_2023-10-22T05-59-09.159543.parquet

• harness_hellaswag_10

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hellaswag|10_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：多个路径，详见原文档

• harness_hendrycksTest_abstract_algebra_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_anatomy_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-anatomy|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_astronomy_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-astronomy|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_business_ethics_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-business_ethics|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_clinical_knowledge_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_college_biology_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-college_biology|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_college_chemistry_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-college_chemistry|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_college_computer_science_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-college_computer_science|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_college_mathematics_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-college_mathematics|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_college_medicine_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-college_medicine|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_college_physics_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-college_physics|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_computer_security_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-computer_security|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_conceptual_physics_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-02T16:13:19.014173.parquet

• harness_hendrycksTest_econometrics_5

  • 分割：2023_08_02T16_13_19.014173, latest
  • 路径：**/details_harness|hendrycksTest-econometrics|5_2023-08-02T16:13:19.014173.parquet