Release v0.5
- Add
@torch.no_grad()to other threads in addition to the main thread to reduce memory consumption sincetorch.no_grad()is thread-local
@torch.no_grad() to other threads in addition to the main thread to reduce memory consumption since torch.no_grad() is thread-local