Tried to load FB released pre-trained model on Instagram on PyTorch Hub. It works without passing the 'pretrained' parameter. But if the parameter is passed it raises error. I think this will cause inconsistency with loading of other models.
To Reproduce
Steps to reproduce the behavior:
model = torch.hub.load('facebookresearch/WSL-Images', 'resnext101_32x8d_wsl', pretrained=True)
TypeError: _resnext() got multiple values for argument 'pretrained'
Expected behavior
Load the architecture with random weights if the 'pretrained' parameter is False, and with the pretrained weights if it is True.
🐛 Bug
Tried to load FB released pre-trained model on Instagram on PyTorch Hub. It works without passing the 'pretrained' parameter. But if the parameter is passed it raises error. I think this will cause inconsistency with loading of other models.
To Reproduce
Steps to reproduce the behavior:
model = torch.hub.load('facebookresearch/WSL-Images', 'resnext101_32x8d_wsl', pretrained=True)
TypeError: _resnext() got multiple values for argument 'pretrained'
Expected behavior
Load the architecture with random weights if the 'pretrained' parameter is False, and with the pretrained weights if it is True.