# ToPO: Token-Conditioned Preference Routing for Attention-Based Latent Diffusion Models

arXiv: https://arxiv.org/abs/2609.03688 · HTML: https://mlsanity.com/paper/2609.03688/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models.
- Builds on 33 of its 42 references that are in the map.
- First posted on arXiv 3 Sep 2026.

## Placed in

- Diffusion models / [Diffusion and flow policies for robotics](https://mlsanity.com/diffusion/diffusion-and-flow-policies-for-robotics/)
- LLM alignment / RLHF / preference optimisation / [Diffusion and multimodal model alignment](https://mlsanity.com/alignment/diffusion-and-multimodal-model-alignment/)

## Builds on (most cited first, 30 of 33)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Prompt-to-Prompt Image Editing with Cross Attention Control](https://mlsanity.com/paper/2208.01626/) (2022)
- [Training Diffusion Models with Reinforcement Learning](https://mlsanity.com/paper/2305.13301/) (2023)
- [Pick-a-Pic: An Open Dataset of User Preferences for Text-to-Image Generation](https://mlsanity.com/paper/2305.01569/) (2023)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [Diffusion Model Alignment Using Direct Preference Optimization](https://mlsanity.com/paper/2311.12908/) (2023)
- [Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models](https://mlsanity.com/paper/2301.13826/) (2023)
- [Aligning Text-to-Image Diffusion Models with Reward Backpropagation](https://mlsanity.com/paper/2310.03739/) (2023)
- [KTO: Model Alignment as Prospect Theoretic Optimization](https://mlsanity.com/paper/2402.01306/) (2024)
- [Using Human Feedback to Fine-tune Diffusion Models without Any Reward Model](https://mlsanity.com/paper/2311.13231/) (2023)
- [Aligning Diffusion Models by Optimizing Human Utility](https://mlsanity.com/paper/2404.04465/) (2024)
- [Aesthetic Post-Training Diffusion Models from Generic Preferences with Step-by-step Preference Optimization](https://mlsanity.com/paper/2406.04314/) (2024)
- [Freestyle Layout-to-Image Synthesis](https://mlsanity.com/paper/2303.14412/) (2023)
- [A Dense Reward View on Aligning Text-to-Image Diffusion with Preference](https://mlsanity.com/paper/2402.08265/) (2024)
- [Margin-aware Preference Optimization for Aligning Diffusion Models without Reference](https://mlsanity.com/paper/2406.06424/) (2024)
- [Diffusion Model as a Noise-Aware Latent Reward Model for Step-Level Preference Optimization](https://mlsanity.com/paper/2502.01051/) (2025)
- [Curriculum Direct Preference Optimization for Diffusion and Consistency Models](https://mlsanity.com/paper/2405.13637/) (2024)
- [InPO: Inversion Preference Optimization with Reparametrized DDIM for Efficient Diffusion Model Alignment](https://mlsanity.com/paper/2503.18454/) (2025)
- [PatchDPO: Patch-level DPO for Finetuning-free Personalized Image Generation](https://mlsanity.com/paper/2412.03177/) (2024)
- [D-Fusion: Direct Preference Optimization for Aligning Diffusion Models with Visually Consistent Samples](https://mlsanity.com/paper/2505.22002/) (2025)
- [Seeing What Matters: Visual Preference Policy Optimization for Visual Generation](https://mlsanity.com/paper/2511.18719/) (2025)
- [Inversion-DPO: Precise and Efficient Post-Training for Diffusion Models](https://mlsanity.com/paper/2507.11554/) (2025)
- [Learning from Noisy Preferences: A Semi-Supervised Learning Approach to Direct Preference Optimization](https://mlsanity.com/paper/2604.24952/) (2026)
- [Rethinking Direct Preference Optimization in Diffusion Models](https://mlsanity.com/paper/2505.18736/) (2025)
- [ViPO: Visual Preference Optimization at Scale](https://mlsanity.com/paper/2604.24953/) (2026)
- [Compositional Text-to-Image Generation Via Region-aware Bimodal Direct Preference Optimization](https://mlsanity.com/paper/2605.28615/) (2026)
- [STAR: SpatioTemporal Adaptive Reward Allocation for Text-to-Image RL Post-Training](https://mlsanity.com/paper/2606.17979/) (2026)

## Abstract (arXiv)

Pairwise preference labels rank complete images, yet Diffusion-DPO applies their effect over many spatial and denoising-time coordinates. For attention-based, noise-prediction latent diffusion, ToPO (Token-Oriented Preference Optimization) constructs a per-minibatch, detached, separable spatial-temporal route from branchwise squared-residual contrast in a frozen reference denoiser. Preferred-branch cross-attention uses content tokens to modulate the spatial factor, and an auxiliary pixel-midpoint ordering term is added without local labels or a learned reward model. In matched three-seed retrainings with a shared update schedule, ToPO has higher endpoint estimates than Diffusion-DPO on all five reported SD-1.5 metrics and on HPSv2, ImageReward, and CLIP for SDXL. It also receives larger raw win shares in an aggregate blind SDXL A/B study. These findings are scoped to the reported equal-update U-Net protocols rather than an equal-compute comparison.