open-llm-leaderboard-old/details_ehartford__samantha-mistral-7b

该数据集是在评估模型ehartford/samantha-mistral-7b时自动生成的，包含64个配置，每个配置对应一个评估任务。数据集由2次运行生成，每次运行的结果存储在不同的分割中，分割名称使用运行的时间戳。train分割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，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 ehartford/samantha-mistral-7b 的过程中自动创建的，用于 Open LLM Leaderboard。

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。
• 数据来源：数据集从2次运行中创建，每次运行对应一个特定的时间戳。
• 数据分割：每个配置包含不同的时间戳分割，其中 "train" 分割指向最新的结果。
• 结果汇总：额外配置 "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。

数据加载示例

python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_ehartford__samantha-mistral-7b", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-29T01:10:37.829717 运行 的最新结果：

python { "all": { "em": 0.056732382550335574, "em_stderr": 0.0023690412638350568, "f1": 0.11221791107382512, "f1_stderr": 0.0025837884585332253, "acc": 0.4639627375502118, "acc_stderr": 0.010980763759790235 }, "harness|drop|3": { "em": 0.056732382550335574, "em_stderr": 0.0023690412638350568, "f1": 0.11221791107382512, "f1_stderr": 0.0025837884585332253 }, "harness|gsm8k|5": { "acc": 0.1599696739954511, "acc_stderr": 0.010097377827752538 }, "harness|winogrande|5": { "acc": 0.7679558011049724, "acc_stderr": 0.011864149691827934 } }

配置详情

• harness_arc_challenge_25

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|arc:challenge|25_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|arc:challenge|25_2023-10-09T12-11-18.939016.parquet

• harness_drop_3

  • 分割：2023_10_29T01_10_37.829717
  • 路径：**/details_harness|drop|3_2023-10-29T01-10-37.829717.parquet
  • 分割：latest
  • 路径：**/details_harness|drop|3_2023-10-29T01-10-37.829717.parquet

• harness_gsm8k_5

  • 分割：2023_10_29T01_10_37.829717
  • 路径：**/details_harness|gsm8k|5_2023-10-29T01-10-37.829717.parquet
  • 分割：latest
  • 路径：**/details_harness|gsm8k|5_2023-10-29T01-10-37.829717.parquet

• harness_hellaswag_10

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hellaswag|10_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hellaswag|10_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-09T12-11-18.939016.parquet 等
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-09T12-11-18.939016.parquet 等

• harness_hendrycksTest_abstract_algebra_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_anatomy_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-anatomy|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-anatomy|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_astronomy_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-astronomy|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-astronomy|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_business_ethics_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-business_ethics|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-business_ethics|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_clinical_knowledge_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_college_biology_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-college_biology|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_biology|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_college_chemistry_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-college_chemistry|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_chemistry|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_college_computer_science_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-college_computer_science|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_computer_science|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_college_mathematics_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-college_mathematics|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_mathematics|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_college_medicine_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-college_medicine|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_medicine|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_college_physics_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-college_physics|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-college_physics|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_computer_security_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-computer_security|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-computer_security|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_conceptual_physics_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_econometrics_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-econometrics|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：**/details_harness|hendrycksTest-econometrics|5_2023-10-09T12-11-18.939016.parquet

• harness_hendrycksTest_electrical_engineering_5

  • 分割：2023_10_09T12_11_18.939016
  • 路径：**/details_harness|hendrycksTest-electrical_engineering|5_2023-10-09T12-11-18.939016.parquet
  • 分割：latest
  • 路径：`**/details_harness|hendrycksTest-electrical_