It is failed when I try to train on Tesla V100GPU， what is your GPU type？

[EricShow]

[XavierCHEN34]

I use A100-80G to train the model. For small memories, you could find out whether there exist more tricks for Pytorch Lightening. Or modify run_train_anydoor.py to set batch size = 1 and use batch accumulation tech.

[EricShow]

I use A100-80G to train the model. For small memories, you could find out whether there exist more tricks for Pytorch Lightening. Or modify run_train_anydoor.py to set batch size = 1 and use batch accumulation tech.

Thank you，my batchsize already set 1，I see that your batchsize sets 16，so the 80G of memory can accommodate batch size 16？My V100 is 32G memory

[zz8615]

I use A100-80G to train the model. For small memories, you could find out whether there exist more tricks for Pytorch Lightening. Or modify run_train_anydoor.py to set batch size = 1 and use batch accumulation tech.

Thank you，my batchsize already set 1，I see that your batchsize sets 16，so the 80G of memory can accommodate batch size 16？My V100 is 32G memory

my model failed when i trained with 2 tesla V100 with batch size one.