open-llm-leaderboard-old/details_CHIH-HUNG__llama-2-13b-Open_Platypus_and_ccp_2.6w-3_epoch
收藏数据集概述
该数据集是在对模型 CHIH-HUNG/llama-2-13b-Open_Platypus_and_ccp_2.6w-3_epoch 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 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。
- 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_CHIH-HUNG__llama-2-13b-Open_Platypus_and_ccp_2.6w-3_epoch", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-29T07:23:25.462349 运行的最新结果: python { "all": { "em": 0.018141778523489933, "em_stderr": 0.0013667968592600756, "f1": 0.07543204697986576, "f1_stderr": 0.001821957230823281, "acc": 0.4184426148579471, "acc_stderr": 0.009471869145583567 }, "harness|drop|3": { "em": 0.018141778523489933, "em_stderr": 0.0013667968592600756, "f1": 0.07543204697986576, "f1_stderr": 0.001821957230823281 }, "harness|gsm8k|5": { "acc": 0.07050796057619409, "acc_stderr": 0.0070515438139836075 }, "harness|winogrande|5": { "acc": 0.7663772691397001, "acc_stderr": 0.011892194477183527 } }
配置详情
-
harness_arc_challenge_25
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|arc:challenge|25_2023-09-11T17-27-50.905630.parquet
-
harness_drop_3
- 拆分:2023_10_29T07_23_25.462349, latest
- 路径:
**/details_harness|drop|3_2023-10-29T07-23-25.462349.parquet
-
harness_gsm8k_5
- 拆分:2023_10_29T07_23_25.462349, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-29T07-23-25.462349.parquet
-
harness_hellaswag_10
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hellaswag|10_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_anatomy_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_astronomy_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_business_ethics_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_college_biology_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_college_chemistry_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_college_computer_science_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_college_mathematics_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_college_medicine_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_college_physics_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_computer_security_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_econometrics_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-09-11T17-27-50.905630.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 拆分:2023_09_11T17_27_50.905630, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-09-11T17-27-50.905630.parquet
以上是数据集的详细概述,包括数据集的组成、加载示例、最新结果以及各个配置的详细信息。



