open-llm-leaderboard-old/details_AI-Sweden-Models__gpt-sw3-356m
收藏数据集概述
数据集摘要
该数据集是在对模型 AI-Sweden-Models/gpt-sw3-356m 进行评估运行期间自动创建的,用于 Open LLM Leaderboard。
数据集组成
- 数据集包含 63 个配置,每个配置对应一个评估任务。
- 数据集从 1 次运行中创建,每个运行可以在每个配置中找到一个特定的拆分,拆分名称使用运行的时间戳。
- "train" 拆分始终指向最新的结果。
- 额外的 "results" 配置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_AI-Sweden-Models__gpt-sw3-356m", "harness_winogrande_5", split="train")
最新结果
这些是最新的结果,来自 2023-12-06T17:48:56.014901 的运行: python { "all": { "acc": 0.25979396469353283, "acc_stderr": 0.030934747009693227, "acc_norm": 0.26083135270282387, "acc_norm_stderr": 0.03173400312275678, "mc1": 0.24112607099143207, "mc1_stderr": 0.014974827279752325, "mc2": 0.4254820499111399, "mc2_stderr": 0.014746028552389436 }, "harness|arc:challenge|25": { "acc": 0.21245733788395904, "acc_stderr": 0.011953482906582947, "acc_norm": 0.2363481228668942, "acc_norm_stderr": 0.012414960524301836 }, "harness|hellaswag|10": { "acc": 0.3201553475403306, "acc_stderr": 0.0046558259808920175, "acc_norm": 0.37054371639115713, "acc_norm_stderr": 0.004819633668832553 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.22, "acc_stderr": 0.04163331998932268, "acc_norm": 0.22, "acc_norm_stderr": 0.04163331998932268 }, # 其他任务的结果... }
配置详情
-
config_name: harness_arc_challenge_25
- split: 2023_12_06T17_48_56.014901
- path: **/details_harness|arc:challenge|25_2023-12-06T17-48-56.014901.parquet
- split: latest
- path: **/details_harness|arc:challenge|25_2023-12-06T17-48-56.014901.parquet
- split: 2023_12_06T17_48_56.014901
-
config_name: harness_gsm8k_5
- split: 2023_12_06T17_48_56.014901
- path: **/details_harness|gsm8k|5_2023-12-06T17-48-56.014901.parquet
- split: latest
- path: **/details_harness|gsm8k|5_2023-12-06T17-48-56.014901.parquet
- split: 2023_12_06T17_48_56.014901
-
config_name: harness_hellaswag_10
- split: 2023_12_06T17_48_56.014901
- path: **/details_harness|hellaswag|10_2023-12-06T17-48-56.014901.parquet
- split: latest
- path: **/details_harness|hellaswag|10_2023-12-06T17-48-56.014901.parquet
- split: 2023_12_06T17_48_56.014901
-
config_name: harness_hendrycksTest_5
- split: 2023_12_06T17_48_56.014901
- path:
- **/details_harness|hendrycksTest-abstract_algebra|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-anatomy|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-astronomy|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-business_ethics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-clinical_knowledge|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_biology|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_chemistry|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_computer_science|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_mathematics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_medicine|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_physics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-computer_security|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-conceptual_physics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-econometrics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-electrical_engineering|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-elementary_mathematics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-formal_logic|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-global_facts|5_2023-12-06T17-48-56.014901.parquet
其他路径...
- path:
- split: latest
- path:
- **/details_harness|hendrycksTest-abstract_algebra|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-anatomy|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-astronomy|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-business_ethics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-clinical_knowledge|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_biology|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_chemistry|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_computer_science|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_mathematics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_medicine|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-college_physics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-computer_security|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-conceptual_physics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-econometrics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-electrical_engineering|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-elementary_mathematics|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-formal_logic|5_2023-12-06T17-48-56.014901.parquet
- **/details_harness|hendrycksTest-global_facts|5_2023-12-06T17-48-56.014901.parquet
其他路径...
- path:
- split: 2023_12_06T17_48_56.014901



