Closed nmakes closed 1 year ago
Hi, would you mind sharing your device info? It takes me a around an hour to generate a clip guided image, I am running on a single V100 machine.
Hi, At present, the question of how one can do universal guidance using a lesser number of steps is the open research question. The current work requires 500 steps and multiple self recurrences making it expensive.
Hi,
Thanks so much for the great work and the code release. Quick question - what is the expected inference time for a single image (say, for the style transfer based guidance)? For me, a single image takes about 20 minutes using the style_transfer.py script - is this expected?
Also, I'm curious to know if there is a way to infer using lesser number of diffusion steps (e.g. 50)?
Thanks!