Cannot force usage of CPU in flask server

[mulholio (James Mulholland)]

Got it!

This thread mentioned that I need to use learn.to and .cpu to set the learner and dataloaders to use the CPU post-training.

Moving learners off GPU

I trained a lot of models in a loop that I knew do not fit in GPU memory. So I called cpu = torch.device('cpu') learner.model.to(cpu) But still got a CUDA out of memory error. Does anyone know what’s going on? You can also call learner.dls.to(cpu or gpu) but I would assume that it only moves batches to gpu for training / inference.

learn.to('cpu')
learn.dls.cpu()
torch.save(learn.dls, 'models/dls-v0-0-1')