VMonarch: Efficient Video Diffusion Transformers with Structured Attention
cs.CV, cs.AI
Submitted: 2026-01-29
Updated: 2026-11-12
Terminology
Sources
- Longformer: The Long-Document Transformer
- Nemotron-H: A Family of Accurate and Efficient Hybrid Mamba-Transformer Models
- SANA-Video: Efficient Video Generation with Block Linear Diffusion Transformer
- Sparse-vDiT: Unleashing the Power of Sparse Attention to Accelerate Video Diffusion Transformers
- MagicPIG: LSH Sampling for Efficient LLM Generation
- Rethinking Attention with Performers
- FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning
- Kaleidoscope: An Efficient, Learnable Representation For All Structured Linear Maps
- Pixelated Butterfly: Simple and Efficient Sparse training for Neural Network Models
- Flex Attention: A Programming Model for Generating Optimized Attention Kernels
- Attention Surgery: An Efficient Recipe to Linearize Your Video Diffusion Transformer
- M4V: Multimodal Mamba for Efficient Text-to-Video Generation
- ARFlow: Autoregressive Flow with Hybrid Linear Attention
- Reformer: The Efficient Transformer
- HunyuanVideo: A Systematic Framework For Large Video Generative Models
- FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference
- MiniMax-01: Scaling Foundation Models with Lightning Attention
- Radial Attention: $O(n\log n)$ Sparse Attention with Energy Decay for Long Video Generation
- Jamba: A Hybrid Transformer-Mamba Language Model
- Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs