open-llm-leaderboard/details_AlpinDale__pygmalion-instruct
收藏数据集概述
该数据集是在模型 AlpinDale/pygmalion-instruct 的评估运行期间自动创建的,用于 Open LLM Leaderboard。
数据集组成
- 数据集包含 64 个配置,每个配置对应一个评估任务。
- 数据集从 3 次运行中创建,每次运行可以在每个配置中找到特定的分割,分割名称使用运行的时间戳。
- "train" 分割始终指向最新的结果。
- 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 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_AlpinDale__pygmalion-instruct", "harness_gsm8k_5", split="train")
最新结果
以下是 2023-12-03T16:32:30.526837 运行的最新结果: python { "all": { "acc": 0.09855951478392722, "acc_stderr": 0.008210320350946347 }, "harness|gsm8k|5": { "acc": 0.09855951478392722, "acc_stderr": 0.008210320350946347 } }
配置详情
-
harness_arc_challenge_25
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|arc:challenge|25_2023-08-17T11:20:15.687659.parquet
-
harness_drop_3
- 分割: 2023_09_09T13_40_12.695061, latest
- 路径:
**/details_harness|drop|3_2023-09-09T13-40-12.695061.parquet
-
harness_gsm8k_5
- 分割: 2023_09_09T13_40_12.695061, 2023_12_03T16_32_30.526837, latest
- 路径:
**/details_harness|gsm8k|5_2023-09-09T13-40-12.695061.parquet,**/details_harness|gsm8k|5_2023-12-03T16-32-30.526837.parquet
-
harness_hellaswag_10
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hellaswag|10_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径: 多个路径,详见原文
-
harness_hendrycksTest_abstract_algebra_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_anatomy_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_astronomy_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_college_biology_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_college_physics_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_computer_security_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_econometrics_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-17T11:20:15.687659.parquet
-
harness_hendrycksTest_formal_logic_5
- 分割: 2023_08_17T11_20_15.687659, latest
- 路径:
**/details_harness|hendrycksTest-formal_logic|5_2023-08-17T11:20:15.687659.parquet



