Add to device here

[ZihaoZHENG]

HugsVision/hugsvision/inference/TorchVisionClassifierInference.py

Line 46 in d93cdd1

 pred = self.model(torch.unsqueeze(img, 0)) 

[Labrak Yanis]

RuntimeError: module must have its parameters and buffers on device cuda:0 (device_ids[0]) but found one of them on device: cpu

[Labrak Yanis]

The issue is related to the DataParallel layer at the end of the initialization block for the TorchVision models. Thus, we need to remove the layer to make it working on CPU devices.

Note: Will be merged in the version 0.72

Code edited:

• HugsVision/hugsvision/nnet/TorchVisionClassifierTrainer.py

  Line 117 in d93cdd1

   self.model.features = torch.nn.DataParallel(self.model.features) 

• HugsVision/hugsvision/nnet/TorchVisionClassifierTrainer.py

  Line 146 in d93cdd1

   self.model = torch.nn.DataParallel(self.model).to(self.device) 

Sources:

• https://discuss.pytorch.org/t/does-dataparallel-matters-in-cpu-mode/7587

[Labrak Yanis]

Close with the push: 8529361