Float16 / half training in PyTorch
Tried to do it in the most obvious way + some hacks from here
Did anybody do it successfully on real models?
Hey, I have tried to launch the following code: from torchvision import models resnet = models.resnet18(pretrained=True).cpu() resnet.half() and have got an exception: libc++abi.dylib: terminating with uncaught exception of type std::invalid_argument: Unsupported tensor type Sounds like the halftensor type is not registered properly. But not sure why it’s the case. Using pytorch 0.2.0 py36_1cu75 soumith Any advice how to fix it?