convert_to_caffe2_models.py: RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!

[rsamvelyan]

I'm using a laptop with Nvidia Cuda capable GPU.

C:\Users\rsamv\Documents\pytorch-ssd_DEC2020>python convert_to_caffe2_models.py mb1-ssd C:/Users/rsamv/Documents/pytorch-ssd_DEC2020/models/mb1-ssd-apple/mb1-ssd-Epoch-100-Loss-0.1923127407208085.pth C:/Users/rsamv/Documents/pytorch-ssd_DEC2020/models/mb1-ssd-apple/open-images-model-labels.txt
Traceback (most recent call last):
File "C:\Users\rsamv\Documents\pytorch-ssd_DEC2020\convert_to_caffe2_models.py", line 47, in
torch.onnx.export(net, dummy_input, model_path, verbose=False, output_names=['scores', 'boxes'])
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\onnx_init_.py", line 316, in export
return utils.export(model, args, f, export_params, verbose, training,
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\onnx\utils.py", line 107, in export
_export(model, args, f, export_params, verbose, training, input_names, output_names,
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\onnx\utils.py", line 724, in _export
_model_to_graph(model, args, verbose, input_names,
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\onnx\utils.py", line 493, in _model_to_graph
graph, params, torch_out, module = _create_jit_graph(model, args)
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\onnx\utils.py", line 437, in _create_jit_graph
graph, torch_out = _trace_and_get_graph_from_model(model, args)
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\onnx\utils.py", line 388, in _trace_and_get_graph_from_model
torch.jit._get_trace_graph(model, args, strict=False, _force_outplace=False, _return_inputs_states=True)
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\jit_trace.py", line 1166, in _get_trace_graph
outs = ONNXTracedModule(f, strict, _force_outplace, return_inputs, _return_inputs_states)(*args, **kwargs)
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\nn\modules\module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\jit_trace.py", line 127, in forward
graph, out = torch._C._create_graph_by_tracing(
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\jit_trace.py", line 118, in wrapper
outs.append(self.inner(*trace_inputs))
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\nn\modules\module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "C:\Users\rsamv\AppData\Roaming\Python\Python39\site-packages\torch\nn\modules\module.py", line 1090, in _slow_forward
result = self.forward(*input, **kwargs)
File "C:\Users\rsamv\Documents\pytorch-ssd_DEC2020\vision\ssd\ssd.py", line 93, in forward
boxes = box_utils.convert_locations_to_boxes(
File "C:\Users\rsamv\Documents\pytorch-ssd_DEC2020\vision\utils\box_utils.py", line 104, in convert_locations_to_boxes
locations[..., :2] * center_variance * priors[..., 2:] + priors[..., :2],
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!

[rsamvelyan]

I hardcoded CPU as follows in ssd/ssd.py and it worked, but looks like it only worked with the CPU.:

LINE 29

` # register layers in source_layer_indexes by adding them to a module list

    self.source_layer_add_ons = nn.ModuleList([t[1] for t in source_layer_indexes
    if isinstance(t, tuple) and not isinstance(t, GraphPath)])
    if device:
        self.device = "cpu" #device
    else:
        self.device = "cpu" #torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
    if is_test:
        self.config = config
        self.priors = config.priors.to(self.device)`

[yinggegit]

when I run convert_to_caffe2_models.p fond same error,
and I change (priors) to 'cpu' and run no error.
maybe is other way to fix this error.

--------------------------------------modify 104 line of box_utils.py --------------------------------------------------

#device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
device = 'cpu'
priors = priors.to(device)

return torch.cat([
    locations[..., :2] * center_variance * priors[..., 2:] + priors[..., :2],
    torch.exp(locations[..., 2:] * size_variance) * priors[..., 2:]
], dim=locations.dim() - 1)