openai / DALL-E

PyTorch package for the discrete VAE used for DALL·E.
Other
10.78k stars 1.94k forks source link

How to derive the Equation 1? #66

Open AaronHeee opened 2 years ago

AaronHeee commented 2 years ago

Sorry I have to re-open an issue as the same as #20, where I posted my attempt to derive the equation 1 but cannot get the result. Could you let me know what I miss? Thanks!

geweihgg commented 2 years ago

Hi,I've tried to derive the equation 1, and get the same result as yours in https://github.com/openai/DALL-E/issues/20. Have you figured it out? Looking forward to your reply, thanks very much!

YUYUEinFDU commented 1 year ago

i have tried to derive the equation (1) using KL divergence between q(z|x) and p(z|x,y), which also does not work. hope the authors could give a clear and complete derive

finbarrtimbers commented 1 year ago

Hey folks- I was also confused by this, so I wrote up the derivation. I'd appreciate a second check over my math.

ahmdtaha commented 8 months ago

I have been looking into this equation for sometime. I wrote my thoughts in the attached main.pdf. I believe we need to convert a single-variable distribution $q{\phi}(z|x)$ into a multi-variable distribution $q{\phi}(y,z|x)$ to derive this equation. Yet, I am not sure if I made the right assumptions. I hope someone look into my derivation and share some feedback.