open-llm-leaderboard/details_TheBloke__Project-Baize-v2-7B-GPTQ
收藏数据集概述
数据集来源
该数据集是在对模型 TheBloke/Project-Baize-v2-7B-GPTQ 进行评估运行期间自动创建的,评估结果发布在 Open LLM Leaderboard 上。
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:数据集从2次运行中创建。每个运行在每个配置中作为一个特定的分割存在,分割名称使用运行的时间戳。
- 最新结果:"train" 分割始终指向最新结果。
- 汇总结果:一个额外的配置 "results" 存储所有运行的汇总结果,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_TheBloke__Project-Baize-v2-7B-GPTQ", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-22T21:24:57.179060 运行 的最新结果: python { "all": { "em": 0.001363255033557047, "em_stderr": 0.0003778609196460633, "f1": 0.05739828020134247, "f1_stderr": 0.001324280220685328, "acc": 0.36097040821028104, "acc_stderr": 0.00860938625459939 }, "harness|drop|3": { "em": 0.001363255033557047, "em_stderr": 0.0003778609196460633, "f1": 0.05739828020134247, "f1_stderr": 0.001324280220685328 }, "harness|gsm8k|5": { "acc": 0.025018953752843062, "acc_stderr": 0.0043020450465643045 }, "harness|winogrande|5": { "acc": 0.696921862667719, "acc_stderr": 0.012916727462634475 } }
配置详情
-
harness_arc_challenge_25
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|arc:challenge|25_2023-08-29T19:38:18.380876.parquet
-
harness_drop_3
- 分割: 2023_10_22T21_24_57.179060, latest
- 路径: **/details_harness|drop|3_2023-10-22T21-24-57.179060.parquet
-
harness_gsm8k_5
- 分割: 2023_10_22T21_24_57.179060, latest
- 路径: **/details_harness|gsm8k|5_2023-10-22T21-24-57.179060.parquet
-
harness_hellaswag_10
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hellaswag|10_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: 多个路径,包括 **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-29T19:38:18.380876.parquet 等
-
harness_hendrycksTest_abstract_algebra_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_anatomy_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-anatomy|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_astronomy_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-astronomy|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_college_biology_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-college_biology|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_college_physics_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-college_physics|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_computer_security_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-computer_security|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_econometrics_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-econometrics|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割: 2023_08_29T19_38_18.380876, latest
- 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-08-29T19:38:18.380876.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割: 2023_08_29T19_38_18.380876
- 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-29T19:38:18.380876.parquet
以上是数据集的详细概述,包括数据集的来源、结构、加载示例、最新结果以及各个配置的详细信息。



