长上下文推理的注意力瓶颈终于有了轻量级解法——无训练、可更新、内存高效，做 LLM 推理优化或长文档应用的团队值得关注。
现代大语言模型依赖长前缀来控制推理行为，但前缀影响会随生成衰减，且注意力计算成本随前缀长度线性增长。现有方法要么压缩前缀但仍需注意力计算，要么通过梯度训练内化前缀但更新困难。本文提出 attention-state memory，一种无训练方法，将前缀与查询 token 的预计算注意力状态外化到轻量级查找表中。在 ManyICLBench 上，LLaMA-3.1-8B 在 1K-8K 内存预算下准确率超过上下文学习，注意力延迟降低 1.36 倍；在 NBA 基准上仅用 20% 内存就超越全注意力 RAG 性能。
Context Memorization for Efficient Long Context Generation
Modern large language model (LLM) applications increasingly rely on long conditioning prefixes to control model behavior at inference time. While prefix-augmented inference is effective, it incurs two structural limitations: i) the prefix's influence fades as generation proceeds, and ii) attention computation over the prefix scales linearly with its length. Existing approaches either keep the prefix in attention while compressing it, or internalize it into model parameters through gradient-based training. The former still attends to the prefix at inference, while the latter is training-intensive and ill-suited to prefix updates. To address these issues, we propose attention-state memory, a training-free approach that externalizes the prefix into a lightweight, lookup-based memory of precomputed attention states between prefix and query tokens. On ManyICLBench with LLaMA-3.1-8B, our method improves accuracy over in-context learning at 1K-8K memory budgets while reducing attention latency by 1.36x at 8K, and surpasses full-attention RAG performance on NBA benchmark using only 20% of its memory footprint.