Hi Tadas,

Thanks for the comment and feedback! I have not tried it yet as I work in PyTorch/FastAI and haven’t run any TF nets in some time.

Myself and another FastAI person are making progress in porting it over and will definitely try it out as an un-trained network once we do that.

In FastAI, we have a new XResNet architecture as of last month, and I’m very interested to compare EfficientNet vs the new XResNet. I expect EfficientNet will outperform it but at the same time, XResNet50 outperforms the old ResNet152, so looking forward to testing both!

Have a good weekend!