Luodian opened this issue a year ago · 1 comments
to support larger batchsize or more frames in video training data.
refere:
An example: https://github.com/Stability-AI/stablediffusion/blob/main/ldm/modules/attention.py