open-llm-leaderboard-old/details_TFLai__Airboros2.1-Platypus2-13B-QLora-0.80-epoch
收藏数据集概述
该数据集是在对模型 TFLai/Airboros2.1-Platypus2-13B-QLora-0.80-epoch 进行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 2 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_TFLai__Airboros2.1-Platypus2-13B-QLora-0.80-epoch", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-22T07:28:22.251658 运行的最新结果：
python { "all": { "em": 0.30935402684563756, "em_stderr": 0.004733640588022148, "f1": 0.39316484899329085, "f1_stderr": 0.004587262243817744, "acc": 0.3756906077348066, "acc_stderr": 0.0060736573567015525 }, "harness|drop|3": { "em": 0.30935402684563756, "em_stderr": 0.004733640588022148, "f1": 0.39316484899329085, "f1_stderr": 0.004587262243817744 }, "harness|gsm8k|5": { "acc": 0.0, "acc_stderr": 0.0 }, "harness|winogrande|5": { "acc": 0.7513812154696132, "acc_stderr": 0.012147314713403105 } }
配置详情
-
harness_arc_challenge_25
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|arc:challenge|25_2023-08-30T11:40:47.771600.parquet
-
harness_drop_3
- 分割: 2023_10_22T07_28_22.251658, latest
- 路径: **/details_harness|drop|3_2023-10-22T07-28-22.251658.parquet
-
harness_gsm8k_5
- 分割: 2023_10_22T07_28_22.251658, latest
- 路径: **/details_harness|gsm8k|5_2023-10-22T07-28-22.251658.parquet
-
harness_hellaswag_10
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hellaswag|10_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: 包含多个主题的 parquet 文件路径
-
harness_hendrycksTest_abstract_algebra_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_anatomy_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-anatomy|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_astronomy_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-astronomy|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_college_biology_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-college_biology|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_college_physics_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-college_physics|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_computer_security_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-computer_security|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_econometrics_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-econometrics|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-08-30T11:40:47.771600.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割: 2023_08_30T11_40_47.771600, latest
- 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-30T11:40:47.771600.parquet
以上是数据集的详细概述，包括数据集的组成、加载示例、最新结果以及各个配置的详细信息。
