PipeDiT: Accelerating Diffusion Transformers in Video Generation with Task Pipelining and Model Decoupling

Documenti analoghi

• DSV: Exploiting Dynamic Sparsity to Accelerate Large-Scale Video DiT Training
  di: Tan, Xin, et al.
  Pubblicazione: (2025)
• GraphLeap: Decoupling Graph Construction and Convolution for Vision GNN Acceleration on FPGA
  di: Ramachandran, Anvitha, et al.
  Pubblicazione: (2026)
• Partially Conditioned Patch Parallelism for Accelerated Diffusion Model Inference
  di: Zhang, XiuYu, et al.
  Pubblicazione: (2024)
• KnapFormer: An Online Load Balancer for Efficient Diffusion Transformers Training
  di: Zhang, Kai, et al.
  Pubblicazione: (2025)
• Real-Time Video Generation with Pyramid Attention Broadcast
  di: Zhao, Xuanlei, et al.
  Pubblicazione: (2024)