PawelGlomski-Intel commented on issue #20659: URL: https://github.com/apache/incubator-mxnet/issues/20659#issuecomment-946883383
I assume you are using CPU and v1.x as you described [here](https://github.com/apache/incubator-mxnet/issues/11203#issuecomment-946223038). Does this code reproduce your issue? ``` import torch import numpy as np import mxnet as mx class TestDeconvTorch(torch.nn.Module): def __init__(self): super(TestDeconvTorch, self).__init__() self.deconv1 = torch.nn.ConvTranspose2d(1, 8, 4) self.deconv2 = torch.nn.ConvTranspose2d(8, 8, 4) self.deconv3 = torch.nn.ConvTranspose2d(8, 1, 4) def forward(self, x): return self.deconv3(self.deconv2(self.deconv1(x))) data = torch.linspace(0, 1, 64).reshape((1, 1, 8, 8)) tester = TestDeconvTorch() outt = tester(data) torch_onnx_export_output = torch.onnx.export( tester, data, 'onnx_test', verbose=True, keep_initializers_as_inputs=True, export_params=True, enable_onnx_checker=True, do_constant_folding=False, opset_version=11, ) sym_block = mx.contrib.onnx.import_to_gluon('onnx_test', ctx=mx.cpu()) outm = sym_block.forward(mx.nd.array(data.numpy())) assert np.all(outm.asnumpy() == outt.detach().numpy()) ``` For me, it runs just fine -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
