search

gabrielelanaro / ml-prototypes

A repository of machine learning prototypes
MIT License
10 stars 6 forks source link

Adjust the weights now that we have tv loss #47

Open gabrielelanaro opened 5 years ago

gabrielelanaro commented 5 years ago

This one should be loss_weights = LossWeights(style=c2s, content=1.0) according to what discussed here #38 Anyway, we might want to keep the incorrect implementation as it works a lot better in the end. Not sure.

Originally posted by @FraPochetti in https://github.com/gabrielelanaro/ml-prototypes/pull/43