Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

No conversion function registered for op type BlockGrad yet #12682

Closed
zhengyi2015mi opened this issue Sep 27, 2018 · 7 comments · Fixed by #13821
Closed

No conversion function registered for op type BlockGrad yet #12682

zhengyi2015mi opened this issue Sep 27, 2018 · 7 comments · Fixed by #13821

Comments

@zhengyi2015mi
Copy link

When I transform my mxnet model to onnx model, I met the following error:
image
Here is my *-symbol.json file:
image

@vandanavk
Copy link
Contributor

@mxnet-label-bot [ONNX, Operator]

@vandanavk
Copy link
Contributor

Thank you for reporting this issue @zhengyi2015mi. Could you share the model you used for testing?

Are there any other operators that are giving such errors? Please share details of those operators too (if any).

@piyushghai
Copy link
Contributor

@zhengyi2015mi Can you provide a snippet of your model code in order to debug this issue ?
Thanks!

@vandanavk
Copy link
Contributor

PR #12946

@GokulNC
Copy link

GokulNC commented May 20, 2019

@vandanavk @Roshrini
I'm still facing the error AttributeError: No conversion function registered for op type BlockGrad yet.

I tried ONNX both from pip as well as installation from source as directed in the README, but the issue doesn't seem to be fixed.

For your reference, the model that I'm trying to convert to ONNX is here (input_shape = (1, 3, 112, 112)):
https://github.com/aidlearning/AidLearning-FrameWork/tree/master/examples/facencnn

And, the guide that I followed for converting MXNet model to ONNX is:
https://mxnet.incubator.apache.org/versions/master/tutorials/onnx/export_mxnet_to_onnx.html

@Roshrini
Copy link
Member

@GokulNC Can you check MXNet version you are using? Could you try latest master branch while installing from source? Or try to get latest nightly build through pip as follows:
pip install mxnet --pre

@GokulNC
Copy link

GokulNC commented May 21, 2019

Thanks, that helped! It seems the binaries in pip's stable line are not the latest ones.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
Development

Successfully merging a pull request may close this issue.

6 participants