open-llm-leaderboard/details_TheBloke__Llama-2-13B-fp16
收藏数据集概述
数据集简介
该数据集是在对模型 TheBloke/Llama-2-13B-fp16 进行评估运行期间自动创建的,用于 Open LLM Leaderboard。
数据集结构
- 配置数量:64个配置,每个配置对应一个评估任务。
- 运行次数:数据集由2次运行创建。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。
- 训练分割:"train" 分割始终指向最新的结果。
- 结果配置: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_TheBloke__Llama-2-13B-fp16", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-22T22:53:07.629534 运行 的最新结果: python { "all": { "em": 0.0014681208053691276, "em_stderr": 0.00039210421902982666, "f1": 0.0607822986577181, "f1_stderr": 0.0013583957676382913, "acc": 0.43739636770101, "acc_stderr": 0.010228023491905505 }, "harness|drop|3": { "em": 0.0014681208053691276, "em_stderr": 0.00039210421902982666, "f1": 0.0607822986577181, "f1_stderr": 0.0013583957676382913 }, "harness|gsm8k|5": { "acc": 0.10841546626231995, "acc_stderr": 0.008563852506627487 }, "harness|winogrande|5": { "acc": 0.7663772691397001, "acc_stderr": 0.011892194477183524 } }
配置详情
-
harness_arc_challenge_25
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|arc:challenge|25_2023-07-24T15:08:39.202746.parquet
-
harness_drop_3
- 分割:2023_10_22T22_53_07.629534, latest
- 路径:
**/details_harness|drop|3_2023-10-22T22-53-07.629534.parquet
-
harness_gsm8k_5
- 分割:2023_10_22T22_53_07.629534, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-22T22-53-07.629534.parquet
-
harness_hellaswag_10
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hellaswag|10_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_anatomy_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_astronomy_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_college_biology_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_college_physics_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_computer_security_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_econometrics_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-24T15:08:39.202746.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割:2023_07_24T15_08_39.202746, latest
- 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-24T15:08:39.202746.parquet
以上是数据集的详细概述,包括数据集的简介、结构、加载示例、最新结果以及各个配置的详细信息。



