TL;DR
A dynamic hybrid attention (DHA) mechanism that integrates high-fidelity local modeling from self-attention with efficient global context aggregation from external attention, together with a decoupled knowledge distillation (DKD) framework for training the resulting heterogeneous architecture is proposed.
Stable Diffusion models deliver photorealistic image synthesis but face computational bottlenecks that hinder deployment in resource-constrained environments. However, most existing distillation methods for diffusion models rely on point-wise feature matching, which often fails to preserve global semantic structure, or...
Recent video diffusion models have achieved remarkable generation quality, but high-fidelity results still largely depend on closed-source systems or costly large-scale infrastructure. Test-time scaling (TTS) offers a training-free way to improve lightweight generators by spending additional inference compute, yet exis...
Swift-Image achieves leading aggregate performance among evaluated open-source models with only 6B parameters and 243K GPU training hours; the compressed 3B model incurs nearly no loss, while few-step distillation further improves aggregate editing performance with substantially fewer sampling steps.
Diffusion-based Vision-Language-Action (VLA) models achieve strong performance in embodied tasks, but their iterative sampling imposes heavy computational and memory-access cost, blocking real-time deployment on edge platforms. Existing acceleration methods either require expensive training (e.g., distillation, flow ma...
Unlike classical CNN approaches, the proposed model employs adversarial training to enhance realism and supports smooth stylization control ranging from content preservation to full stylization, and provides user-controllable interpolation and better inference time.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
