search

arobey1 / mbdg

MIT License
35 stars 4 forks source link

about generative model #2

Open AllenPu opened 2 years ago

AllenPu commented 2 years ago

Hi, I'm trying to replicate your results in the dataset : Rotated MNIST, however, I found that when I tried to use the code from package MUNIT/train_munit.py, I got two models, dist.pt and gen.pt, so I'm wondering referring to the Figure 3, Multi-modal image-to-image translation networks. Which one would represent the H+G as the generative model, or we only need to load gen.pt as generative model?

arobey1 commented 2 years ago

The gen.pt file is the H+G model (i.e. the multimodal image-to-image translation networks in Figure 3) described in the paper. The dis.pt are checkpoints for the discriminator -- to reproduce our results, you won't need this file.