trainset_tasks/:训练任务数据
testset_tasks/:测试任务数据
description.tar: 任务的描述
pipeline.py:处理数据的脚本
json文件夹中放着处理完和划分好比例的json文件
其他子目录:相关数据文件
处理后的文件结构如下:
dataroot/
Clips/
trainset/
testset/
json/
All/
cot100.json
target100.json
...
Exp1/
cot50.json
target5.json
target10.json
target15.json
target20.json
target25.json
target30.json
target35.jsontarget40.json
target45.json
target50.json
Exp2/
cot50.json
cot60.json
cot70.json… See the full description on the dataset page:
https://huggingface.co/datasets/xuejingyang/SpatialLogic-Processed.