open-llm-leaderboard-old/details_rameshm__llama-2-13b-mathgpt-v4
收藏数据集概述
该数据集是在评估模型 rameshm/llama-2-13b-mathgpt-v4 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存在,分割名称使用运行的时间戳。
- 最新结果:"train" 分割总是指向最新的结果。
- 结果汇总: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_rameshm__llama-2-13b-mathgpt-v4", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-25T05:40:14.287010 运行的最新结果:
python { "all": { "em": 0.002202181208053691, "em_stderr": 0.0004800510816619372, "f1": 0.06708787751677872, "f1_stderr": 0.0015253339046219561, "acc": 0.41923906142571715, "acc_stderr": 0.011369111930643223 }, "harness|drop|3": { "em": 0.002202181208053691, "em_stderr": 0.0004800510816619372, "f1": 0.06708787751677872, "f1_stderr": 0.0015253339046219561 }, "harness|gsm8k|5": { "acc": 0.1470811220621683, "acc_stderr": 0.009756063660359863 }, "harness|winogrande|5": { "acc": 0.691397000789266, "acc_stderr": 0.012982160200926584 } }
配置详情
-
harness_arc_challenge_25
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|arc:challenge|25_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2023-09-13T04-13-58.726542.parquet
-
harness_drop_3
- 分割:2023_10_25T05_40_14.287010
- 路径:
**/details_harness|drop|3_2023-10-25T05-40-14.287010.parquet - 分割:latest
- 路径:
**/details_harness|drop|3_2023-10-25T05-40-14.287010.parquet
-
harness_gsm8k_5
- 分割:2023_10_25T05_40_14.287010
- 路径:
**/details_harness|gsm8k|5_2023-10-25T05-40-14.287010.parquet - 分割:latest
- 路径:
**/details_harness|gsm8k|5_2023-10-25T05-40-14.287010.parquet
-
harness_hellaswag_10
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hellaswag|10_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hellaswag|10_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T04-13-58.726542.parquet等 50 个文件 - 分割:latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T04-13-58.726542.parquet等 50 个文件
-
harness_hendrycksTest_abstract_algebra_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_anatomy_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_astronomy_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_college_biology_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_college_physics_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_computer_security_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_econometrics_5
- 分割:2023_09_13T04_13_58.726542
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-09-13T04-13-58.726542.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-09-13T04-13-58.726542.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割:2023_09_13T04_13_58.726542
- 路径:`**/details_harness|hendrycksTest-electrical_engineering|5_2023-09-13



