SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning Paper • 2602.13515 • Published Feb 13 • 44
VoxHammer: Training-Free Precise and Coherent 3D Editing in Native 3D Space Paper • 2508.19247 • Published Aug 26, 2025 • 43