open-llm-leaderboard/details_stabilityai__StableBeluga-13B
收藏数据集概述
该数据集是在评估模型 stabilityai/StableBeluga-13B 在 Open LLM Leaderboard 上的自动创建的。数据集包含 64 个配置,每个配置对应一个评估任务。
数据集结构
- 配置数量:64 个配置
- 数据来源:从 2 次运行中创建
- 数据分割:每个配置包含特定分割,分割名称使用运行的时间戳。"train" 分割始终指向最新结果。
- 结果汇总:额外配置 "results" 存储所有运行结果的汇总(用于计算和显示在 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_stabilityai__StableBeluga-13B", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-15T18:18:48.353426 运行 的最新结果:
python { "all": { "em": 0.13968120805369127, "em_stderr": 0.00355008169467152, "f1": 0.2125828439597308, "f1_stderr": 0.0036624757731315858, "acc": 0.4533641938925533, "acc_stderr": 0.010674908726298674 }, "harness|drop|3": { "em": 0.13968120805369127, "em_stderr": 0.00355008169467152, "f1": 0.2125828439597308, "f1_stderr": 0.0036624757731315858 }, "harness|gsm8k|5": { "acc": 0.1379833206974981, "acc_stderr": 0.009499777327746841 }, "harness|winogrande|5": { "acc": 0.7687450670876085, "acc_stderr": 0.011850040124850508 } }
配置详情
-
harness_arc_challenge_25
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|arc:challenge|25_2023-07-31T13:19:00.568083.parquet
-
harness_drop_3
- 分割:2023_10_15T18_18_48.353426, latest
- 路径:
**/details_harness|drop|3_2023-10-15T18-18-48.353426.parquet
-
harness_gsm8k_5
- 分割:2023_10_15T18_18_48.353426, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-15T18-18-48.353426.parquet
-
harness_hellaswag_10
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hellaswag|10_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_anatomy_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_astronomy_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_college_biology_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_college_physics_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_computer_security_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_econometrics_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-31T13:19:00.568083.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割:2023_07_31T13_19_00.568083, latest
- 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-31T13:19:00.568083.parquet
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。



