loading gpu trained models in cpu

Question

Currently I can't use GPU but I have to load several BERTopic pertained gpu models in CPU.I tried to add map.location=torch.device("cpu") as suggested but without results. I keep obtaining the same error : Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU. How can I resolve it?

Promod George · Accepted Answer · 2022-06-14 06:20:53Z

1

model = torch.load(path/to/the/gpu/model.pt, ,map_location=torch.device('cpu'))

torch.save(model, "GPUToCPU.pt")

In,
:~$ pip show torch
Name: torch
Version: 1.9.0

Then use as usual:
    model = torch.load('path/to/the/new/GPUToCPU.pt')['model'].cpu().float() 
    results = model(im)
    prob = F.softmax(results, dim=1)  # probabilities

answered Jun 14, 2022 at 6:20

Promod George

473 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

loading gpu trained models in cpu

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related