An auxiliary project analysis of the characteristics of KV in DiT Attention.
(34 stars) (2 forks) (0 issues indexées) (0 good first issues ouvertes)
Dépôts
An auxiliary project analysis of the characteristics of KV in DiT Attention.
xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism