open-llm-leaderboard-old/details_illuin__test-custom-llama
收藏数据集概述
数据集简介
该数据集是在评估模型 illuin/test-custom-llama 在 Open LLM Leaderboard 上的运行过程中自动创建的。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:2次运行结果
- 数据分割:每个配置包含不同时间戳的分割,其中 "train" 分割指向最新的结果。
额外配置
- 结果配置:名为 "results" 的配置存储所有运行结果的聚合数据,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_illuin__test-custom-llama", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-27T13:17:00.562267 运行 的最新结果: python { "all": { "em": 0.004089765100671141, "em_stderr": 0.0006535802669912844, "f1": 0.06308619966442945, "f1_stderr": 0.0014549394005291911, "acc": 0.38039089908704843, "acc_stderr": 0.009010133138187597 }, "harness|drop|3": { "em": 0.004089765100671141, "em_stderr": 0.0006535802669912844, "f1": 0.06308619966442945, "f1_stderr": 0.0014549394005291911 }, "harness|gsm8k|5": { "acc": 0.0401819560272934, "acc_stderr": 0.005409439736970511 }, "harness|winogrande|5": { "acc": 0.7205998421468035, "acc_stderr": 0.012610826539404684 } }
配置详情
以下是部分配置的详细信息:
-
harness_arc_challenge_25
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|arc:challenge|25_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2023-07-19T20:12:39.825467.parquet
-
harness_drop_3
- 分割:2023_10_27T13_17_00.562267
- 路径:
**/details_harness|drop|3_2023-10-27T13-17-00.562267.parquet - 分割:latest
- 路径:
**/details_harness|drop|3_2023-10-27T13-17-00.562267.parquet
-
harness_gsm8k_5
- 分割:2023_10_27T13_17_00.562267
- 路径:
**/details_harness|gsm8k|5_2023-10-27T13-17-00.562267.parquet - 分割:latest
- 路径:
**/details_harness|gsm8k|5_2023-10-27T13-17-00.562267.parquet
-
harness_hellaswag_10
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hellaswag|10_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hellaswag|10_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_5
- 分割:2023_07_19T20_12_39.825467
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_anatomy_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_astronomy_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_college_biology_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_college_physics_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_computer_security_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_econometrics_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-07-19T20:12:39.825467.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割:2023_07_19T20_12_39.825467
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T20:12:39.825467.parquet - 分割:latest
- 路径:`**/details_



