pytorch转onnx

一个输入：

?

1 2 3 4 5 6 7 8

input_tensor = torch.randn([1, 3, 256, 512]) print ("Exporting to ONNX: ", onnx_save_name) torch_onnx_out = torch.onnx.export(model, input_tensor, onnx_save_name,                     export_params=True,                     verbose=True,                     input_names=['label'],                     output_names=["synthesized"],                     opset_version=11)

多个输入：

?

1 2 3 4 5 6 7 8 9

input_tensor = torch.randn([1, 3, 256, 512]) mask_tensor = torch.randn([1, 3, 256, 512]) print ("Exporting to ONNX: ", onnx_save_name) torch_onnx_out = torch.onnx.export(model, (input_tensor,mask_tensor), onnx_save_name,                     export_params=True,                     verbose=True,                     input_names=['label','mask'],                     output_names=["synthesized"],                     opset_version=11)