train question

[cLEARLoVE00]

dear author,thanks your code. while I train LIP dataset,I occur a question.I use 4 GPU:2080ti,batch_size=8,num_worker=1,but found :"RuntimeError: CUDA out of memory. Tried to allocate 28.00 MiB (GPU 7; 10.76 GiB total capacity; 262.52 MiB already allocated; 22.12 MiB free; 312.00 MiB reserved in total by PyTorch)", can you give some advice to solve this question? Thanks

[planet-miemie]

我现在把batch-size改为8，输入图片改为256*256解决了这个错误，但是不得不说，LIP数据集太大了，还是很吃配置的