open-llm-leaderboard-old/details_Faradaylab__ARIA-70B-V2
收藏数据集概述
该数据集是在对模型 Faradaylab/ARIA-70B-V2 进行评估运行期间自动创建的,评估运行在 Open LLM Leaderboard 上进行。
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_Faradaylab__ARIA-70B-V2", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-25T19:48:19.078343 运行的最新结果:
python { "all": { "em": 0.12174916107382551, "em_stderr": 0.0033487438315364985, "f1": 0.18035549496644265, "f1_stderr": 0.0034191831504093964, "acc": 0.552493667621485, "acc_stderr": 0.011672124185183144 }, "harness|drop|3": { "em": 0.12174916107382551, "em_stderr": 0.0033487438315364985, "f1": 0.18035549496644265, "f1_stderr": 0.0034191831504093964 }, "harness|gsm8k|5": { "acc": 0.2880970432145565, "acc_stderr": 0.012474469737197917 }, "harness|winogrande|5": { "acc": 0.8168902920284136, "acc_stderr": 0.01086977863316837 } }
配置详情
-
harness_arc_challenge_25
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|arc:challenge|25_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2023-09-14T05-14-04.383698.parquet
-
harness_drop_3
- 分割:2023_10_25T19_48_19.078343
- 路径:
**/details_harness|drop|3_2023-10-25T19-48-19.078343.parquet - 分割:latest
- 路径:
**/details_harness|drop|3_2023-10-25T19-48-19.078343.parquet
-
harness_gsm8k_5
- 分割:2023_10_25T19_48_19.078343
- 路径:
**/details_harness|gsm8k|5_2023-10-25T19-48-19.078343.parquet - 分割:latest
- 路径:
**/details_harness|gsm8k|5_2023-10-25T19-48-19.078343.parquet
-
harness_hellaswag_10
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hellaswag|10_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hellaswag|10_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-14T05-14-04.383698.parquet等 40 个文件 - 分割:latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-14T05-14-04.383698.parquet等 40 个文件
-
harness_hendrycksTest_abstract_algebra_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_anatomy_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_astronomy_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_college_biology_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_college_physics_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_computer_security_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_econometrics_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-09-14T05-14-04.383698.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割:2023_09_14T05_14_04.383698
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-09-14T05-14-04.383698.parquet - 分割:latest
- 路径:`**/details_



