mlsanityData to 14 Oct 2026

Test-Time Alignment of Text-to-Image Diffusion Models via Null-Text Embedding Optimisation

arXiv 2511.20889 ↗First posted 25 Nov 2025cs.CV cs.AI cs.LGcited by 0

On the map: Diffusion models → Image generation, text-to-image and editing

Abstract (arXiv)

Test-time alignment (TTA) aims to adapt models to specific rewards during inference. However, existing methods tend to either under-optimise or over-optimise (reward hack) the target reward function. We propose Null-Text Test-Time Alignment (Null-TTA), which aligns diffusion models by optimising the unconditional embedding in classifier-free guidance, rather than manipulating latent or noise variables. Due to the structured semantic nature of the text embedding space, this ensures alignment occurs on a semantically coherent manifold and prevents reward hacking (exploiting non-semantic noise patterns to improve the reward). Since the unconditional embedding in classifier-free guidance serves as the anchor for the model's generative distribution, Null-TTA directly steers model's generative distribution towards the target reward rather than just adjusting the samples, even without updating model parameters. Thanks to these desirable properties, we show that Null-TTA achieves state-of-the-art target test-time alignment while maintaining strong cross-reward generalisation. This establishes semantic-space optimisation as an effective and principled novel paradigm for TTA.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Image generation, text-to-image and editing, one of 14 directions in Diffusion models. Builds on 35 of its 37 references that are in the map. First posted on arXiv 25 Nov 2025.

Builds on

35 of 37 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024InitNO: Boosting Text-to-Image Diffusion Models via Initial Noise Optimization6 Apr
  2. 2022Null-text Inversion for Editing Real Images using Guided Diffusion Models17 Nov
  3. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding23 May

Other references · 32

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  5. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  6. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,255
  7. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  8. 2022Classifier-Free Diffusion Guidancecited by 5,847
  9. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,356
  10. 2022Diffusion Posterior Sampling for General Noisy Inverse Problemscited by 1,416
  11. 2022Scaling Laws for Reward Model Overoptimizationcited by 1,095
  12. 2022Diffusion-LM Improves Controllable Text Generationcited by 1,027
20 more
  1. 2023Training Diffusion Models with Reinforcement Learningcited by 982
  2. 2023Pick-a-Pic: An Open Dataset of User Preferences for Text-to-Image Generationcited by 977
  3. 2023Diffusion Model Alignment Using Direct Preference Optimizationcited by 790
  4. 2023RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignmentcited by 686
  5. 2023DPOK: Reinforcement Learning for Fine-tuning Text-to-Image Diffusion Modelscited by 450
  6. 2023Directly Fine-Tuning Diffusion Models on Differentiable Rewardscited by 448
  7. 2023Aligning Text-to-Image Models using Human Feedbackcited by 422
  8. 2023Aligning Text-to-Image Diffusion Models with Reward Backpropagationcited by 288
  9. 2023Human Preference Score: Better Aligning Text-to-Image Models with Human Preferencecited by 287
  10. 2025Inference-Time Scaling for Diffusion Models beyond Scaling Denoising Stepscited by 234
  11. 2023Practical and Asymptotically Exact Conditional Sampling in Diffusion Modelscited by 232
  12. 2025A General Framework for Inference-time Scaling and Steering of Diffusion Modelscited by 224
  13. 2023Manifold Preserving Guided Diffusioncited by 157
  14. 2024Fine-Tuning of Continuous-Time Diffusion Models as Entropy-Regularized Controlcited by 143
  15. 2025Test-time Alignment of Diffusion Models without Reward Over-optimizationcited by 93
  16. 2024Inference-Time Alignment of Diffusion Models with Direct Noise Optimizationcited by 55
  17. 2025Dynamic Search for Inference-Time Alignment in Diffusion Modelscited by 37
  18. 2024Confronting Reward Overoptimization for Diffusion Models: A Perspective of Inductive and Primacy Biasescited by 30
  19. 2025Test-Time Scaling of Diffusion Models via Noise Trajectory Searchcited by 30
  20. 2024DyMO: Training-Free Diffusion Model Alignment with Dynamic Multi-Objective Schedulingcited by 27

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Alignment-Guided Score Matching for Text-to-Image Alignment in Diffusion Models
  2. 2025MIRA: Towards Mitigating Reward Hacking in Inference-Time Alignment of T2I Diffusion Models
  3. 2026HyperAlign: Hypernetwork for Efficient Test-Time Alignment of Diffusion Models
  4. 2025Inference-Time Alignment of Diffusion Models via Evolutionary Algorithms
  5. 2025CARINOX: Inference-time Scaling with Category-Aware Reward-based Initial Noise Optimization and Exploration
  6. 2026Alignment of Diffusion Model and Flow Matching for Text-to-Image Generation
  7. 2026TILT: Improving Compositional Generation in Diffusion Models with a Model-Intrinsic Reward
  8. 2025Aligning Text to Image in Diffusion Models is Easier Than You Think