Flow-Anchored Consistency Models
This addresses a fundamental problem in efficient generative modeling for AI researchers and practitioners, offering a general recipe for high-performance, few-step models, though it is incremental as it builds on existing Consistency Models.
The paper tackles training instability in Consistency Models for few-step generation by introducing Flow-Anchored Consistency Models (FACM), which uses a Flow Matching task as an anchor during training, achieving a state-of-the-art FID of 1.32 with two steps and 1.76 with one step on ImageNet 256x256.
Continuous-time Consistency Models (CMs) promise efficient few-step generation but face significant challenges with training instability. We argue this instability stems from a fundamental conflict: by training a network to learn only a shortcut across a probability flow, the model loses its grasp on the instantaneous velocity field that defines the flow. Our solution is to explicitly anchor the model in the underlying flow during training. We introduce the Flow-Anchored Consistency Model (FACM), a simple but effective training strategy that uses a Flow Matching (FM) task as an anchor for the primary CM shortcut objective. This Flow-Anchoring approach requires no architectural modifications and is broadly compatible with standard model architectures. By distilling a pre-trained LightningDiT model, our method achieves a state-of-the-art FID of 1.32 with two steps (NFE=2) and 1.76 with just one step (NFE=1) on ImageNet 256x256, significantly outperforming previous methods. This provides a general and effective recipe for building high-performance, few-step generative models. Our code and pretrained models: https://github.com/ali-vilab/FACM.