Accelerating Inference of Masked Image Generators via Reinforcement Learning
By: Pranav Subbaraman , Shufan Li , Siyan Zhao and more
Potential Business Impact:
Makes AI draw better pictures much faster.
Masked Generative Models (MGM)s demonstrate strong capabilities in generating high-fidelity images. However, they need many sampling steps to create high-quality generations, resulting in slow inference speed. In this work, we propose Speed-RL, a novel paradigm for accelerating a pretrained MGMs to generate high-quality images in fewer steps. Unlike conventional distillation methods which formulate the acceleration problem as a distribution matching problem, where a few-step student model is trained to match the distribution generated by a many-step teacher model, we consider this problem as a reinforcement learning problem. Since the goal of acceleration is to generate high quality images in fewer steps, we can combine a quality reward with a speed reward and finetune the base model using reinforcement learning with the combined reward as the optimization target. Through extensive experiments, we show that the proposed method was able to accelerate the base model by a factor of 3x while maintaining comparable image quality.
Similar Papers
MaskFocus: Focusing Policy Optimization on Critical Steps for Masked Image Generation
CV and Pattern Recognition
Focuses on important steps to make AI images better.
Masked Auto-Regressive Variational Acceleration: Fast Inference Makes Practical Reinforcement Learning
Machine Learning (CS)
Makes AI create pictures much faster and better.
Generation then Reconstruction: Accelerating Masked Autoregressive Models via Two-Stage Sampling
CV and Pattern Recognition
Makes image creation much faster and better.