Fetching the paper…

Test-time Alignment of Diffusion Models without Reward Over-optimization · Around