MATCH: Modulating Attention via In-Context Retrieval for Long-Context Transformers
For LLM practitioners, MATCH offers a way to enhance long-context recall without the quadratic cost of full attention, though the gains are incremental over existing sparse methods.
MATCH improves sparse-attention LLMs by adding in-context retrieval, achieving significant performance gains on long-context tasks while maintaining efficiency.
The quadratic computational cost of traditional attention mechanisms poses a major bottleneck to the scalability and practical deployment of large language models (LLMs), particularly in long-context scenarios. To improve efficiency, existing approaches often enforce rigid structural constraints such as local attention windows. However, these strategies typically lead to substantial performance degradation on tasks requiring precise long-range recall. In this work, we propose MATCH, a scalable and efficient framework that augments sparsified attention mechanisms with dynamically integrated in-context information through an efficient retrieval system. Empirical results show that MATCH significantly improves the performance of sparse-attention models on both synthetic and real-world natural-language tasks. These findings highlight the versatility of MATCH as a general approach for enhancing in-context retrieval capabilities while maintaining the efficiency benefits of sparse attention architectures.