ToPO (Token-Oriented Preference Optimization) is a method for attention-based latent diffusion models that constructs a per‑minibatch, detached, separable spatial‑temporal route using branchwise squared‑residual contrast from a frozen reference denoiser. It employs preferred‑branch cross‑attention with content tokens to modulate spatial factors and adds an auxiliary pixel‑midpoint ordering term, all without local labels or a learned reward model. In experiments with shared update schedules, ToPO outperforms Diffusion‑DPO on five SD‑1.5 metrics and on HPSv2, ImageReward, and CLIP for SDXL, and achieves higher raw win shares in a blind SDXL A/B study.
arXiv:2609.03688v2 Announce Type: replace Abstract: Diffusion models have achieved remarkable success in text-to-image generation, yet aligning their outputs with human preferences remains challengin...
The paper introduces HetA-DiT, a heterogeneous attention mechanism for video diffusion models that allocates computation based on token difficulty. A lightweight uncertainty branch predicts denoising difficulty, routing uncertain tokens through dense global attention while applying efficient local attention to reliable tokens. This adaptive routing retains global context where needed, offers a single parameter to balance quality and efficiency, and achieves competitive generation quality while only about 20% of tokens use dense attention.
The paper introduces Diffusion LAIR, a listwise preference optimization technique that leverages continuous reward scores instead of binary pairwise comparisons to align text‑to‑image diffusion models. LAIR transforms reward scores into centered advantage weights and optimizes an advantage‑weighted regression objective on an implicit reward defined by denoising‑loss improvement over a reference model, with a quadratic penalty to regulate reward magnitude. Experiments demonstrate that Diffusion LAIR surpasses strong baseline methods on SD1.5 and SDXL across generation, compositional, and editing tasks.
arXiv:2603.02531v3 Announce Type: replace-cross Abstract: Classifier-Free Guidance (CFG) improves sample quality in diffusion models, but its dual-pass inference and null-condition training limit its...
arXiv:2606. 17979v1 Announce Type: new Abstract: Existing RL post-training methods for text-to-image generation usually convert the final-image reward into a single scalar advantage and apply it with the same strength to the entire generative trajectory.
ReState is a training‑free method for object removal in few‑step diffusion models that corrects the masked clean prediction toward background evidence using a noise‑level‑aware weight that fades out at the final forward pass. It complements attention rerouting, requiring no additional denoiser evaluations, and outperforms existing training‑free techniques on two paired benchmarks and two distilled backbones, improving PSNR by an average of 3.17 dB. On the OBER‑Test set, ReState also surpasses a 50‑step attention‑forcing reference while running 6.7× faster.
arXiv:2606. 10829v1 Announce Type: cross Abstract: Masked diffusion language models can reduce inference steps by revealing multiple tokens per denoising iteration, but this parallelism is fragile: positions that are individually confident may be unsafe to commit together when their predictions are coupled.
MAST (Mask‑Guided Attention Control for Training‑Free Regional‑Multi Style Transfer) is a framework that enables diffusion models to apply multiple reference styles to user‑specified regions of a content image without any training or optimization. It introduces logit‑level attention mass allocation, sharpness‑aware temperature scaling, and discrepancy‑aware detail injection to address mass allocation, selectivity, and detail loss problems in regional‑multi style transfer. Experiments with two to five styles show that MAST outperforms baselines in ArtFID, FID, and R‑FID, achieving high regional style fidelity, content preservation, and scalability.
arXiv:2603. 08026v2 Announce Type: replace-cross Abstract: Masked diffusion language models enable parallel token decoding, providing a promising alternative to the sequential nature of autoregressive generation.
arXiv:2610.36671v1 Announce Type: cross Abstract: Diffusion vision-language models (DVLMs) process their entire input at every denoising step. Since visual tokens make up 89-97% of this input on the...
The paper investigates attention sinks—tokens or positions that attract disproportionate attention—in text-to-image transformers, comparing bidirectional diffusion models and causal autoregressive models. It finds that both model types exhibit attention sinks, but with distinct patterns: autoregressive models focus on the first and global tokens, while diffusion models develop sinks near modality boundaries, within text regions, and at spatial corners. Perturbation experiments show that sink-related heads influence generation more than non-sink heads, and introducing register tokens can redistribute attention without harming performance.
arXiv:2608.21229v1 Announce Type: new Abstract: Omnimodal generation is central to a wide range of content creation and editing applications. In-context conditioning is essential to this paradigm. It...