TypeError: pad() got multiple values for argument 'name'

[sayakpaul]

I am trying to convert the PyTorch checkpoints of tunit model to TensorFlow using ONNX. I was able to generate the ONNX files but I could not export the TensorFlow file from the ONNX files and ended up with the following error:

Here's the Colab Notebook that reproduces the issue. Please note that I used the animalFaces10_0_00 pre-trained checkpoints. I first copied the files from here to my personal Drive, created a folder called animalFaces10_0_00, and copied the files to that folder.

[chinhuang007]

There is a recent fix #650. Please install tensorflow 2.2 and do a source build for onnx-tf. The problem should be resolved.

[sayakpaul]

Thanks. That solved the issue.

[Zhoushanglin100]

Is there a way I can fix this problem when using tf 1.x? I need to use tensorflow 1.x version because of some other requirements. Thanks!

[chinhuang007]

Yes, we will backport the fix into the tf-1.x branch. A PR will be created shortly and a new onnx-tf release should be available in about a week.

[Guanbin-Huang]

Thanks. That solved the issue.

How did you solve the issue? I changed tf to tf1.x. But I got this error. (the major issue is at the last line). Can you tell me how to convert onnx to tensorflow, plz?

TypeError Traceback (most recent call last)
in ()
7
8 model = onnx.load('reidmodel.onnx')
----> 9 tf_rep = prepare(model)

8 frames
/usr/local/lib/python3.6/dist-packages/onnx_tf/handlers/backend_handler.py in _run_tf_func(cls, tf_func, inputs, attrs)
180 attrs = cls._process_attrs(attrs)
181 return tf_func(*inputs,
--> 182 **dict([(p, attrs[p]) for p in params if p in attrs]))

TypeError: pad() got multiple values for argument 'name'

[sayakpaul]

@HuangGuanbin I am in TensorFlow 2.

[Tristacheng]

Thanks. That solved the issue.

How did you solve the issue? I changed tf to tf1.x. But I got this error. (the major issue is at the last line). Can you tell me how to convert onnx to tensorflow, plz?

TypeError Traceback (most recent call last)
in ()
7
8 model = onnx.load('reidmodel.onnx')
----> 9 tf_rep = prepare(model)

8 frames
/usr/local/lib/python3.6/dist-packages/onnx_tf/handlers/backend_handler.py in _run_tf_func(cls, tf_func, inputs, attrs)
180 attrs = cls._process_attrs(attrs)
181 return tf_func(*inputs,
--> 182 **dict([(p, attrs[p]) for p in params if p in attrs]))

TypeError: pad() got multiple values for argument 'name'

have you solved it? I got same problem in tf1.15

[Tristacheng]

Thanks. That solved the issue.

How did you solve the issue? I changed tf to tf1.x. But I got this error. (the major issue is at the last line). Can you tell me how to convert onnx to tensorflow, plz?
TypeError Traceback (most recent call last)
in ()
7
8 model = onnx.load('reidmodel.onnx')
----> 9 tf_rep = prepare(model)
8 frames
/usr/local/lib/python3.6/dist-packages/onnx_tf/handlers/backend_handler.py in _run_tf_func(cls, tf_func, inputs, attrs)
180 attrs = cls._process_attrs(attrs)
181 return tf_func(*inputs,
--> 182 **dict([(p, attrs[p]) for p in params if p in attrs]))
TypeError: pad() got multiple values for argument 'name'

have you solved it? I got same problem in tf1.15

I solved it refers to 6685fd4