# Trainable Dynamic Mask Sparse Attention

arXiv: https://arxiv.org/abs/2508.02124 · HTML: https://mlsanity.com/paper/2508.02124/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 44 of its 59 references that are in the map.
- First posted on arXiv 4 Aug 2025.

## Builds on (most cited first, 30 of 44)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [Mamba: Linear-Time Sequence Modeling with Selective State Spaces](https://mlsanity.com/paper/2312.00752/) (2023)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://mlsanity.com/paper/1905.07830/) (2019)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [Longformer: The Long-Document Transformer](https://mlsanity.com/paper/2004.05150/) (2020)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://mlsanity.com/paper/1911.11641/) (2019)
- [The Pile: An 800GB Dataset of Diverse Text for Language Modeling](https://mlsanity.com/paper/2101.00027/) (2020)
- [Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering](https://mlsanity.com/paper/1809.02789/) (2018)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://mlsanity.com/paper/1901.02860/) (2019)
- [Decoupled Weight Decay Regularization](https://mlsanity.com/paper/1711.05101/) (2017)
- [Big Bird: Transformers for Longer Sequences](https://mlsanity.com/paper/2007.14062/) (2020)
- [Generating Long Sequences with Sparse Transformers](https://mlsanity.com/paper/1904.10509/) (2019)
- [LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding](https://mlsanity.com/paper/2308.14508/) (2023)
- [Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality](https://mlsanity.com/paper/2405.21060/) (2024)
- [The LAMBADA dataset: Word prediction requiring a broad discourse context](https://mlsanity.com/paper/1606.06031/) (2016)
- [In-context Learning and Induction Heads](https://mlsanity.com/paper/2209.11895/) (2022)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models](https://mlsanity.com/paper/2306.14048/) (2023)
- [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://mlsanity.com/paper/2204.06745/) (2022)
- [Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference](https://mlsanity.com/paper/2406.10774/) (2024)
- [Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs](https://mlsanity.com/paper/2310.01801/) (2023)
- [Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention](https://mlsanity.com/paper/2502.11089/) (2025)
- [From Softmax to Sparsemax: A Sparse Model of Attention and Multi-Label Classification](https://mlsanity.com/paper/1602.02068/) (2016)

## Built on by (most cited first, 6 of 6)

- [NOSA: Native and Offloadable Sparse Attention](https://mlsanity.com/paper/2510.13602/) (2025)
- [Alleviating Forgetfulness of Linear Attention by Hybrid Sparse Attention and Contextualized Learnable Token Eviction](https://mlsanity.com/paper/2510.20787/) (2025)
- [VMonarch: Efficient Video Diffusion Transformers with Structured Attention](https://mlsanity.com/paper/2601.22275/) (2026)
- [Elastic Threshold Attention: Learned Contextual Sparsity for Long-Context Decoding](https://mlsanity.com/paper/2609.20888/) (2026)
- [HierEdit: Region-Aware Hierarchical Diffusion for Efficient High-Resolution Editing](https://mlsanity.com/paper/2605.17294/) (2026)
- [A Self-Pruning Transformer: Extreme KV-Cache Compression with Universal Attention](https://mlsanity.com/paper/2610.09051/) (2026)

## Abstract (arXiv)

The increasing demand for long-context modeling in large language models (LLMs) is bottlenecked by the quadratic complexity of the standard self-attention mechanism. The community has proposed sparse attention to mitigate this issue. However, position-aware sparse attention methods rely on static sparse structures that lack adaptability to diverse query contexts, while content-aware sparse attention methods depend on heuristic key-value selection, hindering full differentiability. We introduce a trainable dynamic mask sparse attention mechanism, a method that merges the advantages of both position-aware and content-aware approaches. Dynamic Mask Attention (DMA) achieves this through three key innovations: First, it leverages value vector representations to generate content-aware dynamic masks, enabling the model to adaptively identify and attend to critical information. Second, it computes position-aware sparse weights in a hardware-friendly manner, efficiently skipping unnecessary computational regions. Finally, we demonstrate that the introduced dynamic mask and sparse weights do not obstruct gradients, supporting end-to-end training. We have validated the performance of DMA through comprehensive experiments. A large body of experimental evidence shows that DMA consistently holds a Pareto advantage over state-of-the-art sparse attention baselines in tasks including scaling laws, multi-query associative recall, standard benchmarks, and needle in a haystack tests, while also delivering up to a 10x overall speedup. These results highlight its ability to effectively balance model efficiency with long-context modeling capabilities. Our computational kernel code is now open-source at https://github.com/SmallDoges/flash-dmattn to encourage further research and application by the community.