2 papers across 2 sessions
We show that diffusion policy can be made faster and more accurate without retraining and analyse why, we introduce a genetic algorithm to reduce to two denoising steps..