Great work, I was wondering if the latest 3D DIT's can load pixart-alpha weights or other open source text2image model weights? Or it only can be trained from scratch?
Thanks,Another question I'd like to ask, why is the 3d DIT's out_channels set to be the same as in_channels, instead of pred sigma as before, i.e. out_channels = 2*in_channels
Great work, I was wondering if the latest 3D DIT's can load pixart-alpha weights or other open source text2image model weights? Or it only can be trained from scratch?