Search papers, labs, and topics across Lattice.
5
0
6
9
DARS achieves superior performance in instruction-based image editing by transforming outcome-level feedback into actionable, localized supervision for both planning and rendering stages.
AdvFD not only boosts visual quality in generative models but also tackles the pitfalls of static feature spaces that lead to stagnation in performance.
NormGuard effectively preserves reward alignment in RL fine-tuning while significantly enhancing perceptual quality, challenging the notion that post-training improvements come at the cost of visual fidelity.
Randomly sampling token subsets during training can significantly enhance diffusion model performance by stabilizing representation alignment.
T2I models can now be diagnosed with unprecedented precision, linking defects directly to their semantic causes and importance for image quality.
