fredO13
Results
13
comments of
fredO13
I tried to change the ellipse loss function to "kullback-leibler", the loss remains at 1 at every iterration, but this is because the displacement term is infinite !
for this loss, it seems displacement_term is infinite because it overflows fp16. I tried to train using fp32, I don't get infinite values anymore, but still the exp(-displacement) is too...
all boxes are empty