Encoder-Decoder VQ-VAE - Githubissues

joh-fischer / PlantLDM

A latent diffusion model for visual synthesis of plant images.

MIT License

15 stars 1 forks source link

Closed joh-fischer closed 2 years ago

joh-fischer commented 2 years ago

In VQ-VAE

[x] make vector quantizer not compute the loss
[x] write encode function that returns $z_q$
[x] write decode function that gets $z_q$, quantizes it to $z_q$ and then decodes $z_q$ to $\hat{x}$
[x] compute loss in train_vqvae.py

joh-fischer commented 2 years ago

Need to test it and will then create a merge request.