[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.

[ICLR2025] Kolmogorov-Arnold Transformer

历年ICLR论文和开源项目合集，包含ICLR2021、ICLR2022、ICLR2023、ICLR2024、ICLR2025.

Efficient DiT architecture for text2any tasks, ICLR2025

[ICLR2025] Spatial-Mamba: Effective Visual State Space Models via Structure-Aware State Fusion

Official Implementation of paper accepted by ICLR2025-MoDGS: Dynamic Gaussian Splatting from Casually-captured Monocular Videos with Depth Priors