open-llm-leaderboard/details_TurkuNLP__gpt3-finnish-small

该数据集是在Open LLM Leaderboard上对TurkuNLP/gpt3-finnish-small模型进行评估时自动创建的。数据集由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，每次运行的结果作为特定配置中的一个split，split以运行的时间戳命名。train split始终指向最新的结果。此外，还有一个名为results的配置，存储了所有运行的聚合结果，并用于在Open LLM Leaderboard上计算和显示聚合指标。

原始信息汇总

数据集概述

该数据集是在评估模型 TurkuNLP/gpt3-finnish-small 的过程中自动创建的，用于 Open LLM Leaderboard 的评估。

数据集结构

• 配置数量：64个配置，每个配置对应一个评估任务。
• 创建来源：数据集来自2次运行，每次运行在每个配置中都有一个特定的分割，分割名称使用运行的时间戳。
• 最新结果："train" 分割总是指向最新的结果。
• 汇总结果：一个额外的配置 "results" 存储所有运行的汇总结果，用于计算和显示在 Open LLM Leaderboard 上的聚合指标。

加载数据示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_TurkuNLP__gpt3-finnish-small", "harness_winogrande_5", split="train")

最新结果

以下是来自2023-10-28T18:15:50.811191运行的最新结果：

python { "all": { "em": 0.0016778523489932886, "em_stderr": 0.0004191330178826759, "f1": 0.020169882550335586, "f1_stderr": 0.0009424790183928599, "acc": 0.24112075769534333, "acc_stderr": 0.007021809798087482 }, "harness|drop|3": { "em": 0.0016778523489932886, "em_stderr": 0.0004191330178826759, "f1": 0.020169882550335586, "f1_stderr": 0.0009424790183928599 }, "harness|gsm8k|5": { "acc": 0.0, "acc_stderr": 0.0 }, "harness|winogrande|5": { "acc": 0.48224151539068666, "acc_stderr": 0.014043619596174964 } }

配置详情

• harness_arc_challenge_25

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|arc:challenge|25_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|arc:challenge|25_2023-09-13T16-47-47.482079.parquet

• harness_drop_3

  • 分割：2023_10_28T18_15_50.811191
  • 路径：**/details_harness|drop|3_2023-10-28T18-15-50.811191.parquet
  • 分割：latest
  • 路径：**/details_harness|drop|3_2023-10-28T18-15-50.811191.parquet

• harness_gsm8k_5

  • 分割：2023_10_28T18_15_50.811191
  • 路径：**/details_harness|gsm8k|5_2023-10-28T18-15-50.811191.parquet
  • 分割：latest
  • 路径：**/details_harness|gsm8k|5_2023-10-28T18-15-50.811191.parquet

• harness_hellaswag_10

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hellaswag|10_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hellaswag|10_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：多个路径，例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：多个路径，例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_abstract_algebra_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_anatomy_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-anatomy|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-anatomy|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_astronomy_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-astronomy|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-astronomy|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_business_ethics_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-business_ethics|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-business_ethics|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_clinical_knowledge_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_college_biology_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-college_biology|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_biology|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_college_chemistry_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-college_chemistry|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_chemistry|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_college_computer_science_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-college_computer_science|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_computer_science|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_college_mathematics_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-college_mathematics|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_mathematics|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_college_medicine_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-college_medicine|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_medicine|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_college_physics_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-college_physics|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_physics|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_computer_security_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-computer_security|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-computer_security|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_conceptual_physics_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_econometrics_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-econometrics|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-econometrics|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_electrical_engineering_5

  • 分割：2023_09_13T16_47_47.482079
  • 路径：**/details_harness|hendrycksTest-electrical_engineering|5_2023-09-13T16-47-47.482079.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-electrical_engineering|5_2023-09-13T16-47-47.482079.parquet

• harness_hendrycksTest_elementary_mathematics_5

  • 分割：2023_09_13T16_47_47.482079