[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.
cuda triton attention vit quantization video-generation mlsys inference-acceleration efficient-attention llm llm-infra video-generate
-
Updated
Jan 17, 2026 - Cuda