Closed DaHaiHuha closed 4 years ago
This shouldn't be a problem. AFAIK pytorch made some code changes between 0.4 and 1.0 that cause this warning to be raised when loading a model in 1.0 that was serialized with 0.4. If you want to get rid of this warning, you can resave the models with some code like:
model = torch.load(old_model_path)
torch.save(model, new_model_path)
Loading the resaved models won't give the warning.
@tbepler Thanks for your kind reply, I'll close this issue.
I used the code & model weights in original repo to reproduce the performance in the paper, using python3.7 and torch1.0.0, and I got following warnings.
However, I do not know whether this would cause a bad consequence, anyone has any ideas? Very appreciate!