I have resolved the issues. It’s related to 1686
state_dict = torch.load(weight_path)
from collections import OrderedDict
new_state_dict = OrderedDict()
for k, v in state_dict.items():
name = k[7:] # remove 'module.' of DataParallel/DistributedDataParallel
new_state_dict[name] = v
m.load_state_dict(new_state_dict)