Pytorch-cyclegan-and-pix2pix: Problem converting to ONNX

[PyJedi]

Hey,

I've been trying to convert the winter2summer checkpoint to an onnx model, so i've tried running test.py with the following additions:

dummy_input = torch.randn(1, 3, 256, 256)
torch.onnx.export(model, dummy_input, "./cycleGAN.onnx")

I got the following error:

Traceback (most recent call last): File "test.py", line 69, in <module> torch.onnx.export(model, dummy_input, "./cycle.onnx") File "/usr/local/lib/python3.6/dist-packages/torch/onnx/__init__.py", line 25, in export return utils.export(*args, **kwargs) File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 131, in export strip_doc_string=strip_doc_string) File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 363, in _export _retain_param_name, do_constant_folding) File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 266, in _model_to_graph graph, torch_out = _trace_and_get_graph_from_model(model, args, training) File "/usr/local/lib/python3.6/dist-packages/torch/onnx/utils.py", line 217, in _trace_and_get_graph_from_model orig_state_dict_keys = _unique_state_dict(model).keys() File "/usr/local/lib/python3.6/dist-packages/torch/jit/__init__.py", line 238, in _unique_state_dict state_dict = module.state_dict(keep_vars=True) AttributeError: 'TestModel' object has no attribute 'state_dict'
Any and all help is much appreciated
Thanks!

[PyJedi]

Looks like I was feeding it the wrong model instance, the issue has been fixed after inputting the generator network

This worked for me:

dummy_input = torch.randn(1, 3, 256, 256)
torch.onnx.export(model.netG, dummy_input, "./cycleGAN.onnx")

I'll be testing the model now.