open-llm-leaderboard-old/details_TheBloke__stable-vicuna-13B-HF
收藏数据集概述
数据集名称
- Evaluation run of TheBloke/stable-vicuna-13B-HF
数据集来源
- 该数据集是在评估模型 TheBloke/stable-vicuna-13B-HF 在 Open LLM Leaderboard 上的自动创建的。
数据集结构
- 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作为一个特定的分片找到,分片名称使用运行的时间戳。
- "train" 分片总是指向最新的结果。
- 一个额外的配置 "results" 存储所有运行结果的聚合(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__stable-vicuna-13B-HF", "harness_winogrande_5", split="train")
最新结果
- 这些是最新结果,来自2023-10-22T23:53:33.704899的运行: python { "all": { "em": 0.005138422818791947, "em_stderr": 0.0007322104102794217, "f1": 0.07736682046979858, "f1_stderr": 0.001663035439531372, "acc": 0.3965552937966327, "acc_stderr": 0.008796231408079842 }, "harness|drop|3": { "em": 0.005138422818791947, "em_stderr": 0.0007322104102794217, "f1": 0.07736682046979858, "f1_stderr": 0.001663035439531372 }, "harness|gsm8k|5": { "acc": 0.04094010614101592, "acc_stderr": 0.005458076796294336 }, "harness|winogrande|5": { "acc": 0.7521704814522494, "acc_stderr": 0.012134386019865348 } }
配置详情
-
harness_arc_challenge_25
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|arc:challenge|25_2023-08-22T17:12:46.134347.parquet
-
harness_drop_3
- 分片: 2023_10_22T23_53_33.704899, latest
- 路径: **/details_harness|drop|3_2023-10-22T23-53-33.704899.parquet
-
harness_gsm8k_5
- 分片: 2023_10_22T23_53_33.704899, latest
- 路径: **/details_harness|gsm8k|5_2023-10-22T23-53-33.704899.parquet
-
harness_hellaswag_10
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hellaswag|10_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: 包含多个子任务的parquet文件路径
-
harness_hendrycksTest_abstract_algebra_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_anatomy_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-anatomy|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_astronomy_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-astronomy|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_business_ethics_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_college_biology_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-college_biology|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_college_medicine_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_college_physics_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-college_physics|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_computer_security_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-computer_security|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_econometrics_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-econometrics|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-08-22T17:12:46.134347.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分片: 2023_08_22T17_12_46.134347, latest
- 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-22T17:12:46.134347.parquet
以上是数据集的详细概述,包括数据集名称、来源、结构、加载示例、最新结果以及各个配置的详细信息。



