arXiv:2605.30038v1 Announce Type: new Abstract: Diffusion models generate highly realistic images but often struggle with precise text-image alignment. While recent post-training methods improve alignment using external rewards or human preference signals, their performance heavily depends on reward quality and does not directly address alignment within the diffusion process itself. Recent reward-free approaches such as SoftREPA demonstrate that optimizing soft text tokens via contrastive learning can effectively improve text-image representation alignment, outperforming standard parameter-eff
Source: arXiv cs.LG — read the full report at the original publisher.