"PyTorch convert function for op '_convolution_mode' not implemented" #1363
Labels
missing layer type
Unable to convert a layer type from the relevant framework
PyTorch (traced)
triaged
Reviewed and examined, release as been assigned if applicable (status)
馃尡 Describe your Feature Request
Pytorch has added 'same' padding for convolutions, but if you try to convert such models to coreml, you get the error "PyTorch convert function for op '_convolution_mode' not implemented".
There is a workaround discussed in recent comments on pytorch/pytorch#3867, but it would be great if we didn't have to change our model code for this, especially since tf->coreml supports this padding.
The text was updated successfully, but these errors were encountered: