open-llm-leaderboard/details_jordiclive__Llama-2-70b-oasst-1-200
收藏数据集概述
数据集摘要
该数据集是在对模型 jordiclive/Llama-2-70b-oasst-1-200 进行评估运行期间自动创建的。数据集由64个配置组成,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。
数据集结构
- 配置数量: 64个配置
- 运行次数: 2次运行
- 分割命名: 使用运行的时间戳命名
- 最新结果: "train" 分割指向最新结果
数据加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_jordiclive__Llama-2-70b-oasst-1-200", "harness_winogrande_5", split="train")
最新结果
以下是2023年9月17日01:27:48.135864运行的最新结果: python { "all": { "em": 0.0016778523489932886, "em_stderr": 0.00041913301788269036, "f1": 0.06728817114093961, "f1_stderr": 0.0013731781553029802, "acc": 0.5844391933091307, "acc_stderr": 0.011597519226727348 }, "harness|drop|3": { "em": 0.0016778523489932886, "em_stderr": 0.00041913301788269036, "f1": 0.06728817114093961, "f1_stderr": 0.0013731781553029802 }, "harness|gsm8k|5": { "acc": 0.32752084912812734, "acc_stderr": 0.012927102210426472 }, "harness|winogrande|5": { "acc": 0.8413575374901342, "acc_stderr": 0.010267936243028223 } }
配置详情
-
harness_arc_challenge_25
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|arc:challenge|25_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_drop_3
- 分割:
2023_09_17T01_27_48.135864,latest - 路径:
**/details_harness|drop|3_2023-09-17T01-27-48.135864.parquet
- 分割:
-
harness_gsm8k_5
- 分割:
2023_09_17T01_27_48.135864,latest - 路径:
**/details_harness|gsm8k|5_2023-09-17T01-27-48.135864.parquet
- 分割:
-
harness_hellaswag_10
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hellaswag|10_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径: 多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_abstract_algebra_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_anatomy_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_astronomy_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_business_ethics_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_college_biology_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_college_chemistry_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_college_computer_science_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_college_mathematics_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_college_medicine_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_college_physics_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_computer_security_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_conceptual_physics_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_econometrics_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_electrical_engineering_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-10T00:36:08.720698.parquet
- 分割:
-
harness_hendrycksTest_elementary_mathematics_5
- 分割:
2023_08_10T00_36_08.720698,latest - 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-10T00:36:08.720698.parquet
- 分割:
以上是数据集的详细概述,包括数据集的结构、加载示例、最新结果以及各个配置的详细信息。



