Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference

TL;DR

Flux Attention introduces a dynamic, context-aware hybrid attention mechanism that adaptively balances full and sparse attention at the layer level, significantly improving inference speed for long-context LLMs.

Contribution

The paper proposes a lightweight Layer Router that enables layer-wise adaptive attention routing, enhancing efficiency without sacrificing retrieval quality in pretrained LLMs.

Findings

Achieves up to 2.8x speedup in prefill and 2.0x in decoding stages.

Requires only 12 hours of training on 8×A800 GPUs.

Demonstrates superior performance-speed trade-offs on multiple benchmarks.

Abstract

The quadratic computational complexity of standard attention mechanisms presents a severe scalability bottleneck for LLMs in long-context scenarios. While hybrid attention mechanisms combining Full Attention (FA) and Sparse Attention (SA) offer a potential solution, existing methods typically rely on static allocation ratios that fail to accommodate the variable retrieval demands of different tasks. Furthermore, head-level dynamic sparsity often introduces severe computational load imbalance and synchronization long-tails, which hinder hardware acceleration during autoregressive decoding. To bridge this gap, we introduce Flux Attention, a context-aware framework that dynamically optimizes attention computation at the layer level. By integrating a lightweight Layer Router into frozen pretrained LLMs, the proposed method adaptively routes each layer to FA or SA based on the input context. This layer-wise routing preserves high-fidelity information retrieval while ensuring contiguous memory access, translating theoretical computational reductions into practical wall-clock speedups. As a parameter-efficient approach, our framework requires only 12 hours of training on 8$\times$A800 GPUs. Extensive experiments across multiple long-context and mathematical reasoning benchmarks demonstrate that Flux Attention achieves a superior trade-off between performance and inference speed compared with baseline models, with speed improvements of up to $2.8\times$ and $2.0\times$ in the prefill and decode stages.