pytorch2keras icon indicating copy to clipboard operation
pytorch2keras copied to clipboard

Published 20 hours ago • verified publisher icon gmalivenko

Converted keras model has different parameters

Open mike-zyz opened this issue 3 years ago • 1 comments

The original PyTorch model has no bias (only weights) in certain Convo layers but the converted Keras model does have bias respectively. And the Keras model shows no BatchNorm2D layers compared to the PyTorch model. I am thinking that the code has conducted BN folding. Is there a way to switch off this function so it will be easier to match the weights?

mike-zyz avatar Mar 11 '21 09:03 mike-zyz

Have the same problem. Any updates?

Tailwhip avatar Jun 01 '21 08:06 Tailwhip