open-llm-leaderboard/details_cyberagent__open-calm-large
收藏数据集概述
数据集来源
该数据集是在评估模型 cyberagent/open-calm-large 在 Open LLM Leaderboard 上的表现时自动创建的。
数据集结构
- 该数据集包含 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一个特定的分割,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_cyberagent__open-calm-large", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-27T06:03:22.906817 运行的最新结果: python { "all": { "em": 0.0030411073825503355, "em_stderr": 0.0005638896908753121, "f1": 0.0342187500000001, "f1_stderr": 0.001160719751306324, "acc": 0.25610125343097334, "acc_stderr": 0.007403477156790928 }, "harness|drop|3": { "em": 0.0030411073825503355, "em_stderr": 0.0005638896908753121, "f1": 0.0342187500000001, "f1_stderr": 0.001160719751306324 }, "harness|gsm8k|5": { "acc": 0.000758150113722517, "acc_stderr": 0.0007581501137225266 }, "harness|winogrande|5": { "acc": 0.5114443567482242, "acc_stderr": 0.014048804199859329 } }
配置详情
-
harness_arc_challenge_25
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|arc:challenge|25_2023-09-09T15-42-03.677218.parquet
-
harness_drop_3
- 分割: 2023_10_27T06_03_22.906817, latest
- 路径: **/details_harness|drop|3_2023-10-27T06-03-22.906817.parquet
-
harness_gsm8k_5
- 分割: 2023_10_27T06_03_22.906817, latest
- 路径: **/details_harness|gsm8k|5_2023-10-27T06-03-22.906817.parquet
-
harness_hellaswag_10
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hellaswag|10_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: 多个路径,包括 **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-09T15-42-03.677218.parquet 等
-
harness_hendrycksTest_abstract_algebra_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_anatomy_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-anatomy|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_astronomy_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-astronomy|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_college_biology_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-college_biology|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_college_physics_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-college_physics|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_computer_security_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-computer_security|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_econometrics_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-econometrics|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-09-09T15-42-03.677218.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割: 2023_09_09T15_42_03.677218, latest
- 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-09-09T15-42-03.677218.parquet
以上是数据集的概述和详细配置信息。



