Mean loss dropping very slowly

[Joseph_BEASSE]

Hello,

I don’t intend to get any solutions, maybe a hint would be enough !
I have implemented the training loop but the result is far from what I am supposed to get.
The mean loss should drop at around 320 for grading, but it isn’t near that at all
It drops to around 1800 hardly after 30min

If you have any clue or need more informations, feel free to ask !

[gent.spah]

Things to try checking again at this point should be implementation of encoder, decoder and vae models. The loss calculation and calculation of gradients.