open-llm-leaderboard-old/details_Yukang__LongAlpaca-13B
收藏数据集概述
数据集简介
该数据集是在对模型 Yukang/LongAlpaca-13B 进行评估运行期间自动创建的,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:数据集由2次运行结果组成,每次运行的结果可以在每个配置中找到,以运行的时间戳命名的特定分片形式存储。
- 最新结果:"train" 分片始终指向最新结果。
- 汇总结果:一个额外的配置 "results" 存储所有运行结果的汇总,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_Yukang__LongAlpaca-13B", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-27T22:00:30.556276 运行的最新结果: python { "all": { "em": 0.17051174496644295, "em_stderr": 0.003851429222727117, "f1": 0.23656669463087293, "f1_stderr": 0.003934121554985558, "acc": 0.32044198895027626, "acc_stderr": 0.006741557601060113 }, "harness|drop|3": { "em": 0.17051174496644295, "em_stderr": 0.003851429222727117, "f1": 0.23656669463087293, "f1_stderr": 0.003934121554985558 }, "harness|gsm8k|5": { "acc": 0.0, "acc_stderr": 0.0 }, "harness|winogrande|5": { "acc": 0.6408839779005525, "acc_stderr": 0.013483115202120225 } }
配置详情
-
harness_arc_challenge_25
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|arc:challenge|25_2023-10-09T09-19-51.890196.parquet
-
harness_drop_3
- 分片:2023_10_27T22_00_30.556276, latest
- 路径:
**/details_harness|drop|3_2023-10-27T22-00-30.556276.parquet
-
harness_gsm8k_5
- 分片:2023_10_27T22_00_30.556276, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-27T22-00-30.556276.parquet
-
harness_hellaswag_10
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hellaswag|10_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_anatomy_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_astronomy_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_business_ethics_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_college_biology_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_college_medicine_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_college_physics_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_computer_security_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_econometrics_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-10-09T09-19-51.890196.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分片:2023_10_09T09_19_51.890196, latest
- 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-10-09T09-19-51.890196.parquet
以上是数据集的详细概述,包括数据集的简介、结构、加载示例、最新结果以及各个配置的详细信息。



