Mathias Gruber

You should follow the notebook "Step4 - Imagenet Training.ipynb", and change the TRAIN_DIR, TEST_DIR and VAL_DIR as appropriate. Let me know if it doesn't work :)

I've updated the repo with a CLI, so you should now be able to do: python main.py \ --name MyDataset \ --train TRAINING_PATH \ --validation VALIDATION_PATH \ --test TEST_PATH \...

I've deployed my model on www.fixmyphoto.ai. With it, I get the following: 

I've ported the VGG16 model from pytorch to keras now, and also updated the PConv2d layer to use local normalization. Loss ranges should be significantly different now, eh?

This happened in my training as well. I'm not sure why, but it's on my todo list to dig a bit more into the batch normalization layer of keras and...

I've also been playing with multi-GPU implementation, but I've not been able to see any successful speedups. Seems like the VGG loss evaluations always happen on the first GPU, and...

Also facing this issue using tfp==0.13.0 and tf==2.5.0

I'm facing the same issue; I also want to filter entries based on content in a related one-to-one data model with planetscale

This is quite annoying when trying to enforce strict CSP. A workaround is to add `style={{ color: undefined }}` to the `` tag

> cc @MathiasGruber wanna test it? see description Absolutely, I'm sure I'm doing something wrong, but here is what I just quickly put together: https://github.com/MathiasGruber/TheNinjaRPG/pull/72/files It's complaining about an any...