thu-ml/SLA · AI Repo StatsTHthu-ml/SLA
SLA: Beyond Sparsity in Diffusion Transformers via Fine-Tunable Sparse–Linear Attention
Tạo videoPython Apache-2.0 23tạo 2025-09-25 #ai-infra#diffusion-transformer#inference-acceleration#linear-attention#mlsys#sparse-attention#sparse-linear-attention#train-acceleration#transformer#video-generation