LeapAlign: Post-Training Flow Matching Models at Any Generation Step by Building Two-Step Trajectories
For practitioners fine-tuning diffusion/flow models, LeapAlign offers a more efficient and stable direct-gradient alignment method that updates early generation steps, which previous methods struggled with.
LeapAlign reduces the memory cost and gradient explosion in fine-tuning flow matching models by compressing the long ODE trajectory into two leaps, enabling direct reward gradient propagation to early generation steps. It outperforms GRPO-based and direct-gradient methods on Flux model fine-tuning, achieving superior image quality and alignment.
This paper focuses on the alignment of flow matching models with human preferences. A promising way is fine-tuning by directly backpropagating reward gradients through the differentiable generation process of flow matching. However, backpropagating through long trajectories results in prohibitive memory costs and gradient explosion. Therefore, direct-gradient methods struggle to update early generation steps, which are crucial for determining the global structure of the final image. To address this issue, we introduce LeapAlign, a fine-tuning method that reduces computational cost and enables direct gradient propagation from reward to early generation steps. Specifically, we shorten the long trajectory into only two steps by designing two consecutive leaps, each skipping multiple ODE sampling steps and predicting future latents in a single step. By randomizing the start and end timesteps of the leaps, LeapAlign leads to efficient and stable model updates at any generation step. To better use such shortened trajectories, we assign higher training weights to those that are more consistent with the long generation path. To further enhance gradient stability, we reduce the weights of gradient terms with large magnitude, instead of completely removing them as done in previous works. When fine-tuning the Flux model, LeapAlign consistently outperforms state-of-the-art GRPO-based and direct-gradient methods across various metrics, achieving superior image quality and image-text alignment.