Difference between 'efficientnet_b0' and 'tf_efficientnet_b0' ?

[pfeatherstone]

What is the difference between 'efficientnet_b0' and 'tf_efficientnet_b0' ?

[rwightman]

@pfeatherstone padding, batchnorm epsilon, and source of pretrained weights

[pfeatherstone]

@rwightman thanks! I've been porting some keras weights to pytorch. Even when using `tf_efficientnet_b0' i was getting a different prediction to keras. It turns out i wasn't setting the model to .eval(). After doing that, I get exactly the same prediction. So thank you for such a great repo!