Code and models for the paper "One Transformer Fits All Distributions in Multi-Modal Diffusion"
GNU Affero General Public License v3.0
1.37k
stars
86
forks
source link
Is it possible that use a convolution Unet to predict the joint distrubution? #22
Open
shencuifeng opened 1 year ago
like sampling time from the whole joint distribution, and adding the two embedding together as the time embedding?