open-llm-leaderboard-old/details_dvruette__llama-13b-pretrained-sft-do2

该数据集是在模型 dvruette/llama-13b-pretrained-sft-do2 在 Open LLM Leaderboard 上进行评估时自动生成的。数据集包含 64 个配置，每个配置对应一个特定的评估任务。它包含 2 次运行的数据，每次运行在每个配置中表示为特定的分割，分割名称由运行的时间戳命名。train 分割始终指向最新的结果。此外，results 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 Python 中的 datasets 库加载数据集的示例，并包含了特定运行的最新结果。

原始信息汇总

数据集概述

该数据集是在对模型 dvruette/llama-13b-pretrained-sft-do2 进行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 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_dvruette__llama-13b-pretrained-sft-do2", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-21T22:43:14.661061 运行的最新结果：

python { "all": { "em": 0.2528313758389262, "em_stderr": 0.004451070247505258, "f1": 0.3196245805369137, "f1_stderr": 0.004416910326006887, "acc": 0.42391092962847055, "acc_stderr": 0.010031261264359954 }, "harness|drop|3": { "em": 0.2528313758389262, "em_stderr": 0.004451070247505258, "f1": 0.3196245805369137, "f1_stderr": 0.004416910326006887 }, "harness|gsm8k|5": { "acc": 0.09249431387414708, "acc_stderr": 0.007980396874560168 }, "harness|winogrande|5": { "acc": 0.755327545382794, "acc_stderr": 0.012082125654159738 } }

配置详情

• harness_arc_challenge_25

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|arc:challenge|25_2023-07-19T18:57:06.342295.parquet

• harness_drop_3

  • 分割: 2023_10_21T22_43_14.661061, latest
  • 路径: **/details_harness|drop|3_2023-10-21T22-43-14.661061.parquet

• harness_gsm8k_5

  • 分割: 2023_10_21T22_43_14.661061, latest
  • 路径: **/details_harness|gsm8k|5_2023-10-21T22-43-14.661061.parquet

• harness_hellaswag_10

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hellaswag|10_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: 多个路径，例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_abstract_algebra_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_anatomy_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-anatomy|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_astronomy_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-astronomy|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_business_ethics_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_clinical_knowledge_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_college_biology_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-college_biology|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_college_chemistry_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_college_computer_science_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_college_mathematics_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_college_medicine_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_college_physics_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-college_physics|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_computer_security_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-computer_security|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_conceptual_physics_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_econometrics_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-econometrics|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_electrical_engineering_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T18:57:06.342295.parquet

• harness_hendrycksTest_elementary_mathematics_5

  • 分割: 2023_07_19T18_57_06.342295, latest
  • 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T18:57:06.342295.parquet

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