open-llm-leaderboard-old/details_yeontaek__llama-2-13b-Guanaco-QLoRA
收藏数据集概述
该数据集是在评估模型 yeontaek/llama-2-13b-Guanaco-QLoRA 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,每个运行在每个配置中作为一个特定的分割,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。
- 聚合结果: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_yeontaek__llama-2-13b-Guanaco-QLoRA", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-22T09:19:25.361261 运行的最新结果:
python { "all": { "em": 0.0018875838926174498, "em_stderr": 0.00044451099905589624, "f1": 0.06432256711409379, "f1_stderr": 0.0013911197722481863, "acc": 0.44091694875395904, "acc_stderr": 0.010204605702764498 }, "harness|drop|3": { "em": 0.0018875838926174498, "em_stderr": 0.00044451099905589624, "f1": 0.06432256711409379, "f1_stderr": 0.0013911197722481863 }, "harness|gsm8k|5": { "acc": 0.10993176648976498, "acc_stderr": 0.008616195587865406 }, "harness|winogrande|5": { "acc": 0.7719021310181531, "acc_stderr": 0.011793015817663592 } }
配置详情
-
harness_arc_challenge_25
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|arc:challenge|25_2023-08-09T21:23:10.081119.parquet
-
harness_drop_3
- 分割:2023_10_22T09_19_25.361261, latest
- 路径:
**/details_harness|drop|3_2023-10-22T09-19-25.361261.parquet
-
harness_gsm8k_5
- 分割:2023_10_22T09_19_25.361261, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-22T09-19-25.361261.parquet
-
harness_hellaswag_10
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hellaswag|10_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_anatomy_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_astronomy_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_college_biology_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_college_physics_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_computer_security_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_econometrics_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-09T21:23:10.081119.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割:2023_08_09T21_23_10.081119, latest
- 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-09T21:23:10.081119.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。



