---
title: "SpecCompact: Accelerating LLM Agent Context Compaction with Slot-Parallel Semantic Speculative Verification"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.06876v1"
permalink: "https://feed.astralor.com/items/itm_0997e326d52881dc83ecc74c"
published: "2026-11-04T18:34:41.000Z"
score: 57
channel: "tech"
tags: ["LLM","AI Agent","推理效率","SpecCompact","Accelerating LLM Agent","Context Compaction","Slot-Parallel Semantic Speculative","Verification LLM"]
---

# SpecCompact: Accelerating LLM Agent Context Compaction with Slot-Parallel Semantic Speculative Verification

## 速览

LLM agents use context compaction to manage growing interaction histories, but summarizing long histories with large models can introduce substantial delays. Compaction therefore requires balancing information retention…

## 关键点

- LLM agents use context compaction to manage growing interaction histories, but summarizing long histories with large models can introduce substantial delays.
- Compaction therefore requires balancing information retention against latency.
- We observe that structured summary slots can retain information when generated independently.

## 全文

LLM agents use context compaction to manage growing interaction histories, but summarizing long histories with large models can introduce substantial delays. Compaction therefore requires balancing information retention against latency. We observe that structured summary slots can retain information when generated independently. Based on this observation, we present SpecCompact, a framework that accelerates compaction through slot-parallel semantic speculation. SpecCompact fine-tunes a small drafter and generates slots in parallel. The target model verifies each proposed line and repairs rejected content, while subsequent drafting uses the updated slot prefix. System optimizations further reduce repeated context processing and model calls. Across AppWorld and $τ^3$ benchmarks with two target models, SpecCompact achieves $2.04$--$5.63\times$ compaction speedups over Hermes-style summarization. Separate downstream evaluations reveal model- and task-dependent performance trade-offs. A user study within Hermes further shows an $8.68\times$ median compaction speedup.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_0997e326d52881dc83ecc74c