open-llm-leaderboard-old/details_circulus__Llama-2-13b-orca-v1
收藏数据集概述
数据集简介
该数据集是在对模型 circulus/Llama-2-13b-orca-v1 进行评估运行期间自动创建的，用于 Open LLM Leaderboard。
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。
- 创建来源：数据集由3次运行创建，每次运行在每个配置中作为一个特定的分片存在，分片名称使用运行的时间戳。
- 最新结果："train" 分片始终指向最新的结果。
- 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 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_circulus__Llama-2-13b-orca-v1", "harness_winogrande_5", split="train")
最新结果
以下是 2023-09-17T12:50:57.881579 运行的最新结果： python { "all": { "em": 0.1529991610738255, "em_stderr": 0.0036866006582882706, "f1": 0.2246581375838923, "f1_stderr": 0.003770616290655452, "acc": 0.44842066021890015, "acc_stderr": 0.010546865226614108 }, "harness|drop|3": { "em": 0.1529991610738255, "em_stderr": 0.0036866006582882706, "f1": 0.2246581375838923, "f1_stderr": 0.003770616290655452 }, "harness|gsm8k|5": { "acc": 0.1288855193328279, "acc_stderr": 0.009229580761400274 }, "harness|winogrande|5": { "acc": 0.7679558011049724, "acc_stderr": 0.011864149691827943 } }
配置详情
-
harness_arc_challenge_25
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|arc:challenge|25_2023-08-18T08:46:04.009114.parquet
-
harness_drop_3
- 分片：2023_09_17T07_53_18.900339, 2023_09_17T12_50_57.881579, latest
- 路径：
**/details_harness|drop|3_2023-09-17T07-53-18.900339.parquet,**/details_harness|drop|3_2023-09-17T12-50-57.881579.parquet
-
harness_gsm8k_5
- 分片：2023_09_17T07_53_18.900339, 2023_09_17T12_50_57.881579, latest
- 路径：
**/details_harness|gsm8k|5_2023-09-17T07-53-18.900339.parquet,**/details_harness|gsm8k|5_2023-09-17T12-50-57.881579.parquet
-
harness_hellaswag_10
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hellaswag|10_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：多个路径，详见原文
-
harness_hendrycksTest_abstract_algebra_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_anatomy_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-anatomy|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_astronomy_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-astronomy|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_business_ethics_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-business_ethics|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_college_biology_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-college_biology|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-college_chemistry|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-college_computer_science|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-college_mathematics|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_college_medicine_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-college_medicine|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_college_physics_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-college_physics|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_computer_security_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-computer_security|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_econometrics_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-econometrics|5_2023-08-18T08:46:04.009114.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分片：2023_08_18T08_46_04.009114, latest
- 路径：
**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-18T08:46:04.009114.parquet
