open-llm-leaderboard-old/details_EleutherAI__pythia-410m-deduped

该数据集是在Open LLM Leaderboard上对模型EleutherAI/pythia-410m-deduped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。数据集来自2次运行，每次运行在每个配置中表示为特定的分割，train分割始终指向最新结果。此外，还有一个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还包含一个Python代码片段，用于加载数据集，并提供了特定运行的最新结果。

原始信息汇总

数据集概述

数据集名称

Evaluation run of EleutherAI/pythia-410m-deduped

数据集描述

该数据集是在评估模型 EleutherAI/pythia-410m-deduped 在 Open LLM Leaderboard 上的自动创建的。

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，每个运行可以在每个配置中作为一个特定的分割找到，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_EleutherAI__pythia-410m-deduped", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-18T09:27:36.064128 运行 的最新结果： python { "all": { "em": 0.0014681208053691276, "em_stderr": 0.0003921042190298293, "f1": 0.042572357382550524, "f1_stderr": 0.0011637772390608397, "acc": 0.27341843124559817, "acc_stderr": 0.007756513586074438 }, "harness|drop|3": { "em": 0.0014681208053691276, "em_stderr": 0.0003921042190298293, "f1": 0.042572357382550524, "f1_stderr": 0.0011637772390608397 }, "harness|gsm8k|5": { "acc": 0.003032600454890068, "acc_stderr": 0.0015145735612245436 }, "harness|winogrande|5": { "acc": 0.5438042620363063, "acc_stderr": 0.013998453610924331 } }

配置详情

• harness_arc_challenge_25

  • 分割: 2023_07_19T14_23_02.980263
    • 路径: **/details_harness|arc:challenge|25_2023-07-19T14:23:02.980263.parquet
  • 分割: latest
    • 路径: **/details_harness|arc:challenge|25_2023-07-19T14:23:02.980263.parquet

• harness_drop_3

  • 分割: 2023_10_18T09_27_36.064128
    • 路径: **/details_harness|drop|3_2023-10-18T09-27-36.064128.parquet
  • 分割: latest
    • 路径: **/details_harness|drop|3_2023-10-18T09-27-36.064128.parquet

• harness_gsm8k_5

  • 分割: 2023_10_18T09_27_36.064128
    • 路径: **/details_harness|gsm8k|5_2023-10-18T09-27-36.064128.parquet
  • 分割: latest
    • 路径: **/details_harness|gsm8k|5_2023-10-18T09-27-36.064128.parquet

• harness_hellaswag_10

  • 分割: 2023_07_19T14_23_02.980263
    • 路径: **/details_harness|hellaswag|10_2023-07-19T14:23:02.980263.parquet
  • 分割: latest
    • 路径: **/details_harness|hellaswag|10_2023-07-19T14:23:02.980263.parquet

• harness_hendrycksTest_5

  • 分割: 2023_07_19T14_23_02.980263
    • 路径:
      • **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-anatomy|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-astronomy|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-business_ethics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-college_biology|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-college_medicine|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-college_physics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-computer_security|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-econometrics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-formal_logic|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-global_facts|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_biology|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_chemistry|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_computer_science|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_european_history|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_geography|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_mathematics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_microeconomics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_physics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_psychology|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_statistics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_us_history|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-high_school_world_history|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-human_aging|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-human_sexuality|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-international_law|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-jurisprudence|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-logical_fallacies|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-machine_learning|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-management|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-marketing|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-medical_genetics|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-miscellaneous|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-moral_disputes|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-moral_scenarios|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-nutrition|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-philosophy|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-prehistory|5_2023-07-19T14:23:02.980263.parquet
      • **/details_harness|hendrycksTest-professional_accounting|5_2023-07-19T14:23:02.980263.parquet
      • `**/details_harness|hendrycksTest-professional_law|5_2