L2P trains per-timestep linear weights on feature trajectories in about 20 seconds to enable aggressive caching in DiT models, delivering up to 4.55x FLOPs reduction with maintained visual quality.
U- net: Convolutional networks for biomedical image segmen- tation
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
FastSHADE is a new efficient hierarchical denoising network that achieves real-time mobile inference (<50 ms) and up to 37.94 dB PSNR on the MAI2021 benchmark via asymmetric frequency blocks and noise-shifting self-augmentation.
citing papers explorer
-
Beyond Fixed Formulas: Data-Driven Linear Predictor for Efficient Diffusion Models
L2P trains per-timestep linear weights on feature trajectories in about 20 seconds to enable aggressive caching in DiT models, delivering up to 4.55x FLOPs reduction with maintained visual quality.
-
FastSHADE: Fast Self-augmented Hierarchical Asymmetric Denoising for Efficient inference on mobile devices
FastSHADE is a new efficient hierarchical denoising network that achieves real-time mobile inference (<50 ms) and up to 37.94 dB PSNR on the MAI2021 benchmark via asymmetric frequency blocks and noise-shifting self-augmentation.