RuntimeError: Has not supported replacing the module: `PixelShuffle`

[dolphintear]

Describe the issue:

Environment:

• NNI version:Release 2.3 (github master branch)
• Training service (local|remote|pai|aml|etc):local
• Client OS:Ubuntu 18.04
• Server OS (for remote mode only):
• Python version: Python 3.7.0
• PyTorch/TensorFlow version: 1.7.1
• Is conda/virtualenv/venv used?: virutalenv
• Is running in Docker?: no

For super resolution models, nn.PixelShuffle(scale) is widely used. When I tried to compress and speed up a super resolution model with nn.PixelShuffle( ), an error was met: RuntimeError: Has not supported replacing the module: PixelShuffle . Is there a way that I can fix or work around it. Thanks very much!

[zheng-ningxin]

Thanks for the feedback! You can write a corresponding replace function in replace_modules.py or you can give a sample model (If it is convenient) that contains this op, so that our devs can support this op.

[dolphintear]

Thanks for the feedback! You can write a corresponding replace function in replace_modules.py or you can give a sample model (If it is convenient) that contains this op, so that our devs can support this op.

thanks for the reply! I added the sample model, will it take a long time for supporting this op? and there is another question about the compression of a model. For Super resolution models, BatchNorms are not used to boost the model results. What is the good choice for compressing a model with common convolutions, pointwise _conv and depthwsie_conv but without batchnorms? thanks very much!
pixelshuffle_opt.zip

[dolphintear]

Thanks for the feedback! You can write a corresponding replace function in replace_modules.py or you can give a sample model (If it is convenient) that contains this op, so that our devs can support this op.

I have uploaded the sample model with the op of nn.PixelShuffle. Could you help take a look at this problem? Thanks a lot!

[dolphintear]

Thanks for the feedback! You can write a corresponding replace function in replace_modules.py or you can give a sample model (If it is convenient) that contains this op, so that our devs can support this op.

sorry to bother you again! any updates on the not supporting issue of nn.PixelShuffle( )?