unsqueeze version 13 is not implemented

[Saozy]

Describe the bug

Can't convert onnx model to TF frozen graph.

To Reproduce
code:

onnx_model = onnx.load("input_path")  
tf_rep = prepare(onnx_model)
tf_rep.export_graph(export_dir) 

log:

raise BackendIsNotSupposedToImplementIt("{} version {} is not implemented.".format(node.op_type, cls.SINCE_VERSION))
BackendIsNotSupposedToImplementIt: Unsqueeze version 13 is not implemented.

This section can be obtained by running get_version.py from util folder.

• Python version: 3.7.5
• ONNX version: 1.10.0
• ONNX-TF version: 1.9.0
• Tensorflow version: 2.6.0

Additional context

Please help me with this. Will it support unsqueeze in the next version~~~

[chinhuang007]

Thank for bringing this to our attention. Unfortunately ONNX spec changed axes from as an attribute to an input in version 13, which requires a major re-write of the handler. We hope some community members can step up and help out one this.

[Saozy]

thx~

[krishnanpooja]

Thank for bringing this to our attention. Unfortunately ONNX spec changed axes from as an attribute to an input in version 13, which requires a major re-write of the handler. We hope some community members can step up and help out one this.

I have an implementation for unsqueeze and squeeze that I have tested with unit tests as well as by exporting a fairseq transformer XL model (based on huggingface). Should I reopen this ticket and submit a request for review?

[chinhuang007]

Please go ahead to submit a PR. Feel free to reopen this issue too. Thanks!

[rbli-john]

Thank for bringing this to our attention. Unfortunately ONNX spec changed axes from as an attribute to an input in version 13, which requires a major re-write of the handler. We hope some community members can step up and help out one this.

I have an implementation for unsqueeze and squeeze that I have tested with unit tests as well as by exporting a fairseq transformer XL model (based on huggingface). Should I reopen this ticket and submit a request for review?

hi，I met the same problem. May I ask has the fix released? If yes, which onnx-tf version and which opset version I should use to apply this fix. thx

[krishnanNuance]

I couldn't reopen the ticket as I didn't create it. So I have opened another issue tracking both squeeze and unsqueeze- #1020.