Memory-augmented drafting is introduced for long-context SD, equipping a strong independent draft with compressed draft-side KV memory and incrementally updates this memory to retain distant information and exact recent context.
Abstract
Long-context LLM applications such as document summarization and multi-turn agents require generation from prefixes spanning tens of thousands of tokens, making decoding latency a major bottleneck. Speculative decoding (SD) reduces latency without changing model outputs, but its speedup depends on both accepted draft tokens and draft-step latency: Lightweight drafts are fast but lack the capacity to capture long-range dependencies, whereas strong independent drafts recover acceptance but incur growing KV-access cost at long prefixes. We introduce memory-augmented drafting for long-context SD, equipping a strong independent draft with compressed draft-side KV memory: A lightweight adaptor constructs and incrementally updates this memory to retain distant information and exact recent context. The target verifier retains its full KV cache and applies the standard accept/reject rule, preserving SD's lossless guarantee. Experiments on Llama~3.1-8B and 70B targets at prefix lengths up to 32K show that our method reduces draft-side memory by over 70%. It achieves speedups of up to 2.08x and 3.33x , respectively, over autoregressive decoding.
H-Spec is a hybrid Mamba-attention parallel drafter that consumes the two target-context sources through complementary modules, and consistently achieves higher throughput while maintaining lower KV cache utilization than baselines across evaluated concurrency levels.
Speculative decoding reduces sequential Target model calls by verifying multiple tokens from the Draft model in parallel. Yet KV Cache growth limits long-context serving under constrained GPU memory. Offloading KV to CPU memory relieves this pressure. However, existing offloading schemes restore the full KV history bef...
Fei Li, Song Liu, Shi-Qiang Nie et al.· arXiv.org· 0 citations
TierKV is presented, a mobile LLM inference framework built on Predictive Multi-Tier Cache Optimization (PMCO), which improves prefill throughput by up to 17.6x over existing mobile LLM frameworks, reduces RAM-resident KV cache by 12.5-34%, thereby enabling substantially longer contexts under the same memory budget, wh...
Zhi-Hao Shu, Md Musfiqur Rahman Sanim, Jie Hu et al.· arXiv.org· 0 citations
This work proposes ASPIRE, a non-synchronized batched self-speculative decoding framework built on three components, which achieves speedup in decoding throughput over autoregressive baselines and improves average speedup by approximately $27\% over the strongest prior self-speculative baselines.
Amir Ziashahabi, Hossein Entezari Zarch, Lei Gao et al.· 1 citation
This work introduces LongSpark, a block-diffusion drafter that achieves state-of-the-art end-to-end efficiency across multiple model scales and realistic serving conditions by extracting fixed-size, multiscale views from the target's verification pass, thereby eliminating the need for a growing persistent state.
Hao-Yuan He, Peng-Fei Liu, Si-Shi Shen et al.· 0 citations
Minima-KV, a retention-preserving hierarchy for mixed-format paged attention, is presented, a practical mixed-format path for compressing long-context state without evicting live-request KV pages.
S. Kozyrev, AI DavydMaiborodaMinima, Inc.· 0 citations
Related blog posts
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 13, 2026