Unlike torch imagenet example where we load both pretrained model and optimState, in pytorch example we only load the model and not the optimState.
Is that intentional?
it’s not intentional. Just not implemented.
1 Like