LeapAlign fine-tunes flow matching models by constructing two consecutive leaps that skip multiple ODE steps with randomized timesteps and consistency weighting, enabling stable updates at any generation step.
Directly aligning the full diffusion trajectory with fine-grained human preference
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 3roles
background 1polarities
background 1representative citing papers
Real images contrasted with generated samples can supply effective preference signals for aligning diffusion models at performance levels comparable to standard preference-pair methods.
citing papers explorer
-
LeapAlign: Post-Training Flow Matching Models at Any Generation Step by Building Two-Step Trajectories
LeapAlign fine-tunes flow matching models by constructing two consecutive leaps that skip multiple ODE steps with randomized timesteps and consistency weighting, enabling stable updates at any generation step.
-
When Preference Labels Fall Short: Aligning Diffusion Models from Real Data
Real images contrasted with generated samples can supply effective preference signals for aligning diffusion models at performance levels comparable to standard preference-pair methods.
- HunyuanImage 3.0 Technical Report