open-llm-leaderboard-old/details_nthngdy__pythia-owt2-70m-50k

该数据集是在模型 nthngdy/pythia-owt2-70m-50k 在 Open LLM Leaderboard 上的评估运行期间自动创建的。它由 64 个配置组成，每个配置对应一个评估任务。数据集由 2 次运行生成，每次运行在每个配置中表示为特定的拆分，使用运行的时间戳命名。train 拆分始终指向最新结果。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。可以使用 datasets 库中的 load_dataset 函数加载数据集，指定数据集名称以及所需的配置和拆分。

原始信息汇总

数据集概述

数据集 Evaluation run of nthngdy/pythia-owt2-70m-50k 是在评估模型 nthngdy/pythia-owt2-70m-50k 在 Open LLM Leaderboard 上的自动创建的。

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，每个运行在每个配置中作为一个特定的分割存在，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_nthngdy__pythia-owt2-70m-50k", "harness_winogrande_5", split="train")

最新结果

以下是 2023-09-17T20:43:29.834346 运行的最新结果：

python { "all": { "em": 0.01164010067114094, "em_stderr": 0.0010984380734033166, "f1": 0.042847105704697984, "f1_stderr": 0.0015153408758494672, "acc": 0.26203630623520124, "acc_stderr": 0.007018094832697564 }, "harness|drop|3": { "em": 0.01164010067114094, "em_stderr": 0.0010984380734033166, "f1": 0.042847105704697984, "f1_stderr": 0.0015153408758494672 }, "harness|gsm8k|5": { "acc": 0.0, "acc_stderr": 0.0 }, "harness|winogrande|5": { "acc": 0.5240726124704025, "acc_stderr": 0.014036189665395127 } }

配置详情

• harness_arc_challenge_25

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|arc:challenge|25_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|arc:challenge|25_2023-07-19T13:38:15.957411.parquet

• harness_drop_3

  • 分割：2023_09_17T20_43_29.834346
  • 路径：**/details_harness|drop|3_2023-09-17T20-43-29.834346.parquet
  • 分割：latest
  • 路径：**/details_harness|drop|3_2023-09-17T20-43-29.834346.parquet

• harness_gsm8k_5

  • 分割：2023_09_17T20_43_29.834346
  • 路径：**/details_harness|gsm8k|5_2023-09-17T20-43-29.834346.parquet
  • 分割：latest
  • 路径：**/details_harness|gsm8k|5_2023-09-17T20-43-29.834346.parquet

• harness_hellaswag_10

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hellaswag|10_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hellaswag|10_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T13:38:15.957411.parquet 等 50 个文件
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T13:38:15.957411.parquet 等 50 个文件

• harness_hendrycksTest_abstract_algebra_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_anatomy_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-anatomy|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-anatomy|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_astronomy_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-astronomy|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-astronomy|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_business_ethics_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-business_ethics|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-business_ethics|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_clinical_knowledge_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_college_biology_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-college_biology|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_biology|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_college_chemistry_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_college_computer_science_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_college_mathematics_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_college_medicine_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-college_medicine|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_medicine|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_college_physics_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-college_physics|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_physics|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_computer_security_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-computer_security|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-computer_security|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_conceptual_physics_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_econometrics_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-econometrics|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-econometrics|5_2023-07-19T13:38:15.957411.parquet

• harness_hendrycksTest_electrical_engineering_5

  • 分割：2023_07_19T13_38_15.957411
  • 路径：**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T13:38:15.957411.parquet
  • 分割：latest