the light version performance badly

taki0112 / UGATIT

Official Tensorflow implementation of U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation (ICLR 2020)

MIT License

6.17k stars 1.05k forks source link

the light version performance badly #93

Open dragen1860 opened 4 years ago

dragen1860 commented 4 years ago

As i can tell , the full UGATIT model performance relatively good. However, due to its up to 1GB giant size, we can not adopt it conveniently. I tried the light version, it perform quite bad, a lot of anime crashed obvisouly. Could the author give some advice on how to keep the UGATIT model efficient while keeping SOTA? thank you.

tankfly2014 commented 3 years ago

please show demo image. I want to know the difference between the two models, is there a picture example?