open-llm-leaderboard-old/details_Corianas__Quokka_590m
收藏数据集概述
该数据集是在评估模型 Corianas/Quokka_590m 在 Open LLM Leaderboard 上的自动创建的。数据集包含 64 个配置,每个配置对应一个评估任务。
数据集结构
- 配置数量:64 个配置
- 数据来源:从 2 次运行中创建,每次运行作为一个特定的分割,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。
- 聚合结果: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_Corianas__Quokka_590m", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-14T16:50:32.705897 运行的最新结果:
python { "all": { "em": 0.0036703020134228187, "em_stderr": 0.0006192871806511148, "f1": 0.03957634228187927, "f1_stderr": 0.0012672315965293443, "acc": 0.2509865824782952, "acc_stderr": 0.0070261881296128145 }, "harness|drop|3": { "em": 0.0036703020134228187, "em_stderr": 0.0006192871806511148, "f1": 0.03957634228187927, "f1_stderr": 0.0012672315965293443 }, "harness|gsm8k|5": { "acc": 0.0, "acc_stderr": 0.0 }, "harness|winogrande|5": { "acc": 0.5019731649565904, "acc_stderr": 0.014052376259225629 } }
配置详情
-
harness_arc_challenge_25
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|arc:challenge|25_2023-07-24T09:57:25.772408.parquet
-
harness_drop_3
- 分割:2023_10_14T16_50_32.705897, latest
- 路径:
**/details_harness|drop|3_2023-10-14T16-50-32.705897.parquet
-
harness_gsm8k_5
- 分割:2023_10_14T16_50_32.705897, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-14T16-50-32.705897.parquet
-
harness_hellaswag_10
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hellaswag|10_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_anatomy_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_astronomy_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_college_biology_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_college_physics_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_computer_security_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_econometrics_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-24T09:57:25.772408.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割:2023_07_24T09_57_25.772408, latest
- 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-24T09:57:25.772408.parquet
以上是数据集的详细概述,包括数据集的结构、加载示例、最新结果以及各个配置的详细信息。



