Noise Hypernetworks: Amortizing Test-Time Compute in Diffusion Models
read the original abstract
The new paradigm of test-time scaling has yielded remarkable breakthroughs in Large Language Models (LLMs) (e.g. reasoning models) and in generative vision models, allowing models to allocate additional computation during inference to effectively tackle increasingly complex problems. Despite the improvements of this approach, an important limitation emerges: the substantial increase in computation time makes the process slow and impractical for many applications. Given the success of this paradigm and its growing usage, we seek to preserve its benefits while eschewing the inference overhead. In this work we propose one solution to the critical problem of integrating test-time scaling knowledge into a model during post-training. Specifically, we replace reward guided test-time noise optimization in diffusion models with a Noise Hypernetwork that modulates initial input noise. We propose a theoretically grounded framework for learning this reward-tilted distribution for distilled generators, through a tractable noise-space objective that maintains fidelity to the base model while optimizing for desired characteristics. We show that our approach recovers a substantial portion of the quality gains from explicit test-time optimization at a fraction of the computational cost. Code is available at https://github.com/ExplainableML/HyperNoise
This paper has not been read by Pith yet.
Forward citations
Cited by 4 Pith papers
-
Don't Settle at the Mode! Mitigating Diversity Collapse in Pretrained Flow Models via Feature Self-Guidance
Feature self-guidance disperses internal features of flow models during batch generation and applies manifold regularization to increase output diversity while preserving condition alignment.
-
FASTER: Value-Guided Sampling for Fast RL
FASTER models multi-candidate denoising as an MDP and trains a value function to filter actions early, delivering the performance of full sampling at lower cost in diffusion RL policies.
-
Lagrangian Perturbation Diffusion Steering: Latent Reinforcement Learning for Generative Policies
LP-DS improves generative policies for imitation and RL by optimizing latent noise perturbations with a constrained Lagrangian objective, showing up to 25% better returns on manipulation and locomotion tasks.
-
Can We Predict The Human Preference For Text-to-Image Content Prior To Generation And Is It Even Useful To Do So?
Exploration of pre-generation prediction of human preference metrics (HPM) from noise seeds in diffusion models to improve output quality with negligible added cost.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.