2 papers
cs.CV2025
LEDiT: Your Length-Extrapolatable Diffusion Transformer without Positional Encoding
Shen Zhang, Siyuan Liang, Yaning Tan +9
Diffusion transformers (DiTs) struggle to generate images at resolutions higher than their training resolutions. The primary obstacle is that the explicit positional encodings(PE),…
cs.CV2025
Optimizing Knowledge Distillation in Transformers: Enabling Multi-Head Attention without Alignment Barriers
Zhaodong Bing, Linze Li, Jiajun Liang
Knowledge distillation (KD) in transformers often faces challenges due to misalignment in the number of attention heads between teacher and student models. Existing methods either…