open-llm-leaderboard/details_Charlie911__vicuna-7b-v1.5-lora-mixed-datasets-time-unit
收藏数据集概述
数据集创建背景
该数据集是在对模型 Charlie911/vicuna-7b-v1.5-lora-mixed-datasets-time-unit 进行评估运行期间自动创建的,用于 Open LLM Leaderboard。
数据集结构
- 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,每个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。
- "train" 分片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_Charlie911__vicuna-7b-v1.5-lora-mixed-datasets-time-unit", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-28T09:09:00.701109 运行的最新结果: python { "all": { "em": 0.004928691275167785, "em_stderr": 0.0007171872517059817, "f1": 0.06610213926174507, "f1_stderr": 0.001553905671666344, "acc": 0.4026417372707673, "acc_stderr": 0.009752392640502771 }, "harness|drop|3": { "em": 0.004928691275167785, "em_stderr": 0.0007171872517059817, "f1": 0.06610213926174507, "f1_stderr": 0.001553905671666344 }, "harness|gsm8k|5": { "acc": 0.0712661106899166, "acc_stderr": 0.007086462127954495 }, "harness|winogrande|5": { "acc": 0.734017363851618, "acc_stderr": 0.012418323153051046 } }
配置详情
-
harness_arc_challenge_25
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|arc:challenge|25_2023-10-08T16-13-20.175189.parquet
-
harness_drop_3
- 分片:2023_10_28T09_09_00.701109, latest
- 路径:
**/details_harness|drop|3_2023-10-28T09-09-00.701109.parquet
-
harness_gsm8k_5
- 分片:2023_10_28T09_09_00.701109, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-28T09-09-00.701109.parquet
-
harness_hellaswag_10
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hellaswag|10_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_anatomy_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_astronomy_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_business_ethics_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_college_biology_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_college_medicine_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_college_physics_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_computer_security_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_econometrics_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-10-08T16-13-20.175189.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分片:2023_10_08T16_13_20.175189, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-10-08T16-13-20.175189.parquet
以上是数据集的详细概述,包括数据集的创建背景、结构、加载示例、最新结果以及各个配置的详细信息。



