I’ve recently got into score-based diffusion models. Their combination with methods for speeding up the reverse process seems very exciting.

Really looking forward to implementing them for computational microscopy tasks.