Diffusion Fine-Tuning via Reparameterized Policy Gradient of the Soft Q-Function
By: Hyeongyu Kang , Jaewoo Lee , Woocheol Shin and more
Potential Business Impact:
Makes AI art look better and more natural.
Diffusion models excel at generating high-likelihood samples but often require alignment with downstream objectives. Existing fine-tuning methods for diffusion models significantly suffer from reward over-optimization, resulting in high-reward but unnatural samples and degraded diversity. To mitigate over-optimization, we propose \textbf{Soft Q-based Diffusion Finetuning (SQDF)}, a novel KL-regularized RL method for diffusion alignment that applies a reparameterized policy gradient of a training-free, differentiable estimation of the soft Q-function. SQDF is further enhanced with three innovations: a discount factor for proper credit assignment in the denoising process, the integration of consistency models to refine Q-function estimates, and the use of an off-policy replay buffer to improve mode coverage and manage the reward-diversity trade-off. Our experiments demonstrate that SQDF achieves superior target rewards while preserving diversity in text-to-image alignment. Furthermore, in online black-box optimization, SQDF attains high sample efficiency while maintaining naturalness and diversity.
Similar Papers
Quantum Reinforcement Learning-Guided Diffusion Model for Image Synthesis via Hybrid Quantum-Classical Generative Model Architectures
Quantum Physics
Makes AI art look better by adjusting its settings.
Data-regularized Reinforcement Learning for Diffusion Models at Scale
Machine Learning (CS)
Makes AI create better videos that people like.
Fine-Tuning Diffusion-Based Recommender Systems via Reinforcement Learning with Reward Function Optimization
Information Retrieval
Makes movie suggestions better and faster.