open-llm-leaderboard/details_garage-bAInd__Camel-Platypus2-13B

该数据集是在模型 garage-bAInd/Camel-Platypus2-13B 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，每个配置对应一个特定的评估任务。数据集是从 2 次运行中生成的，每次运行在每个配置中表示为特定的分割，分割名称使用运行的时间戳。train 分割始终指向最新的结果。一个名为 results 的额外配置存储了运行的所有聚合结果，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 garage-bAInd/Camel-Platypus2-13B on the Open LLM Leaderboard. The dataset consists of 64 configurations, each corresponding to a specific evaluation task. The dataset is generated from 2 runs, where each run is represented as a specific split for each configuration, with the split name using the timestamp of the run. The "train" split always points to the most recent results. An additional configuration named "results" stores all aggregated results from the 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 the detailed information from the runs.

原始信息汇总

数据集概述

数据集简介

该数据集是在评估模型garage-bAInd/Camel-Platypus2-13B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_garage-bAInd__Camel-Platypus2-13B", "harness_winogrande_5", split="train")

最新结果

以下是2023-10-13T04:35:13.977731运行的最新结果： python { "all": { "em": 0.3248741610738255, "em_stderr": 0.004796115152921962, "f1": 0.38906250000000175, "f1_stderr": 0.004663274154133875, "acc": 0.37725358176562207, "acc_stderr": 0.006433257710580032 }, "harness|drop|3": { "em": 0.3248741610738255, "em_stderr": 0.004796115152921962, "f1": 0.38906250000000175, "f1_stderr": 0.004663274154133875 }, "harness|gsm8k|5": { "acc": 0.000758150113722517, "acc_stderr": 0.0007581501137225365 }, "harness|winogrande|5": { "acc": 0.7537490134175217, "acc_stderr": 0.012108365307437528 } }

配置详情

以下是数据集的配置详情：

配置列表

• harness_arc_challenge_25

  • 分割: 2023_08_09T16_10_57.360881
    • 路径: **/details_harness|arc:challenge|25_2023-08-09T16:10:57.360881.parquet
  • 分割: latest
    • 路径: **/details_harness|arc:challenge|25_2023-08-09T16:10:57.360881.parquet

• harness_drop_3

  • 分割: 2023_10_13T04_35_13.977731
    • 路径: **/details_harness|drop|3_2023-10-13T04-35-13.977731.parquet
  • 分割: latest
    • 路径: **/details_harness|drop|3_2023-10-13T04-35-13.977731.parquet

• harness_gsm8k_5

  • 分割: 2023_10_13T04_35_13.977731
    • 路径: **/details_harness|gsm8k|5_2023-10-13T04-35-13.977731.parquet
  • 分割: latest
    • 路径: **/details_harness|gsm8k|5_2023-10-13T04-35-13.977731.parquet

• harness_hellaswag_10

  • 分割: 2023_08_09T16_10_57.360881
    • 路径: **/details_harness|hellaswag|10_2023-08-09T16:10:57.360881.parquet
  • 分割: latest
    • 路径: **/details_harness|hellaswag|10_2023-08-09T16:10:57.360881.parquet

• harness_hendrycksTest_5

  • 分割: 2023_08_09T16_10_57.360881
    • 路径:
      • **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-anatomy|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-astronomy|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-business_ethics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-college_biology|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-college_chemistry|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-college_computer_science|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-college_mathematics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-college_medicine|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-college_physics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-computer_security|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-conceptual_physics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-econometrics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-electrical_engineering|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-formal_logic|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-global_facts|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_biology|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_chemistry|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_computer_science|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_european_history|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_geography|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_mathematics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_microeconomics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_physics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_psychology|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_statistics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_us_history|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-high_school_world_history|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-human_aging|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-human_sexuality|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-international_law|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-jurisprudence|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-logical_fallacies|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-machine_learning|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-management|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-marketing|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-medical_genetics|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-miscellaneous|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-moral_disputes|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-moral_scenarios|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-nutrition|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-philosophy|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-prehistory|5_2023-08-09T16:10:57.360881.parquet
      • **/details_harness|hendrycksTest-professional_accounting|5_2023-08-09T16:10:57.360881.parquet
      • `**/details_harness|hendrycksTest-professional_law|5_2023-08-09T16:10:57.36088

搜集汇总

背景与挑战

背景概述

该数据集是为评估模型'garage-bAInd/Camel-Platypus2-13B'在Open LLM Leaderboard上的性能而自动创建的，包含64个配置对应不同的评估任务，并通过2次运行生成分割数据，其中train分割提供最新结果，同时包含聚合结果的配置以支持排行榜指标计算。数据集主要用于模型性能的详细分析和比较，便于用户通过标准工具加载和使用。