[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-speccompact-agent-context-compaction":3,"topics-all":38,"news-related-821cbcc4-a8e0-41fe-90a8-284abd96f641":57},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"821cbcc4-a8e0-41fe-90a8-284abd96f641","Agent上下文压缩提速8.68倍:小模型起草大模型验收","SpecCompact 给 agent 上下文压缩提速:微调小模型并行起草摘要槽位,目标大模型逐行验证并修补被拒内容。AppWorld 与 τ³ 基准上压缩较 Hermes 式总结快 2.04 至 5.63 倍,Hermes 用户研究中位数提速 8.68 倍;论文提醒下游表现随模型与任务波动。","Agent 跑得越久,账本越厚。多轮交互留下的任务记录、工具返回和中间结论不断堆进上下文,token 一路上涨,最后都得靠「压缩」续命:让大模型把长历史总结成一份摘要,替换掉原文。问题是,这一步本身就是一次完整的生成任务——用你最贵的模型,把你最长的历史复述一遍。压缩触发的那个瞬间,界面看起来像卡住了,后台其实在整理记忆,而用户只能在旁边干等。\n\n## 压缩慢在哪\n\n上下文压缩(context compaction)的困难一直是道两难题:总结得太省,信息丢失,后续任务掉分;总结得太细,延迟飙升,交互体验先崩掉。理想状态是压缩足够勤、摘要足够准、等待足够短,但三件事互相牵制。论文对这一步的开销说得直白:用大模型总结长历史会引入可观的延迟,所以压缩必须在整个「信息保留」和「延迟」之间找平衡点。换句话说,压缩不是免费的管家,它自己就是账单上的一笔大支出。\n\n## 槽位并行,逐行验收\n\nSpecCompact 的起点是一个关键观察:结构化的摘要槽位(summary slots)如果各自独立生成,信息照样保得住。这句话把摘要内部的依赖链拆开了。于是整个框架变成一条流水线:先微调一个小模型担任 drafter,让它并行起草各个槽位;目标大模型不再亲自从头写摘要,只对草稿逐行验证,被拒绝的行才由它出手修补;后续的起草再拿更新过的槽位前缀当条件。系统层还配了一组优化,专门削减重复的上下文处理和重复的模型调用。这套「小模型起草、大模型验收」的结构,和投机解码在 token 层的逻辑同源:验证比生成便宜,那就把最贵的算力只花在把关上。\n\n## 提速多少,代价是什么\n\n数字层面:在 AppWorld 和 τ³ 两个基准、两个目标模型上,SpecCompact 的压缩环节比 Hermes 式总结提速 2.04 到 5.63 倍;在 Hermes 内做的用户研究里,压缩提速的中位数达到 8.68 倍。要说明的是,这些倍数比的是压缩环节本身的耗时,任务端到端成绩是另一回事,论文单独做了下游评估,结论是效果随模型和任务变化,存在权衡——速度拿到手了,质量端并没有一刀切的保证([论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.06876))。\n\n## 所以呢\n\n我的看法:这篇论文真正值得记住的不是那个 8.68 倍,而是「验证比生成便宜」这个不对称性正在被一路泛化——从 token 级的投机解码,一路搬到语义级的摘要验收。对做 agent 基建的人来说,压缩延迟从此要按交互路径上的一等公民来预算:它不再是后台批处理,而是用户能直接感知的卡顿。更实际的问题是选型:如果你的 agent 已经在用大模型做周期性总结,这篇论文给了一个把等待时间大幅压缩、又不必重训主模型的可参考配方。下次排查「我的 agent 为什么慢」,先看一眼它多久压缩一次历史。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.06876","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"6ad31a14-c0da-42df-81fd-564281f768db","agentic-ai",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"2d9c2fb0-2be5-4ad1-aedb-e9747addf355","compression",{"id":19,"name":20,"slug":20,"description":14,"color":14},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",{"id":22,"name":23,"slug":23,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"25e38ec2-6918-455c-8078-7e660b17c205","en","SpecCompact: 8.68x faster agent context compaction","A small drafter writes summary slots in parallel; the target LLM verifies each line. 2.04-5.63x compaction speedup, 8.68x median in a user study.","The longer an agent runs, the heavier its ledger. Task records, tool outputs and intermediate conclusions from many turns of interaction keep piling into the context, so compaction becomes survival gear: have the big model summarize the long history and swap the summary in for the original. The problem is that this step is itself a full generation pass — your most expensive model restating your longest history. The moment compaction triggers, the user just waits.\n\n## Why compaction is slow\n\nContext compaction has always been a dilemma: summarize too aggressively and information is lost, hurting later tasks; summarize too carefully and latency spikes, wrecking the interaction first. The paper is blunt about the cost: summarizing long histories with large models introduces substantial delays, so compaction must balance information retention against latency. In other words, compaction is not a free butler — it is a big line item on the bill itself.\n\n## Slots in parallel, verified line by line\n\nSpecCompact starts from one observation: structured summary slots can retain information even when generated independently. That single sentence dissolves the dependency chain inside a summary. The framework then becomes a pipeline: fine-tune a small model as the drafter and let it draft the slots in parallel; the target large model no longer writes the whole summary from scratch — it verifies each proposed line, and only repairs the rejected ones; subsequent drafting conditions on the updated slot prefix. A set of system-level optimizations further cuts repeated context processing and repeated model calls. The \"small model drafts, big model verifies\" structure shares its logic with speculative decoding at the token level: verification is cheaper than generation, so spend the most expensive compute only on gatekeeping.\n\n## How much faster, at what cost\n\nOn the AppWorld and tau-3 benchmarks across two target models, SpecCompact's compaction step runs 2.04x to 5.63x faster than Hermes-style summarization; in a user study within Hermes, the median compaction speedup reached 8.68x. The paper does not oversell: separate downstream evaluations reveal model- and task-dependent performance trade-offs — you get the speed, with no across-the-board guarantee on quality ([paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.06876)).\n\n## So what\n\nMy take: the number worth remembering is not 8.68x but the asymmetry being generalized — verification is cheaper than generation, and that gap is now being exploited all the way from token-level speculative decoding up to semantic-level summary verification. For anyone building agent infrastructure, compaction latency is now a first-class line item on the interaction path: it is no longer a background batch job but a stall users can feel directly. Next time you debug \"why is my agent slow\", check how often it compacts its history.","speccompact-agent-context-compaction","2026-11-05T19:20:00Z","2026-11-05T19:06:32.661933Z","2026-11-05T19:06:32.661944Z",true,"agent",106,[39,48],{"slug":40,"tag_slug":40,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":58},[59,64,69,74,79,84],{"id":60,"title":61,"news_slug":62,"published_at":63},"f28966de-08b7-442e-ba3a-1d644131df26","工具输出不必是人话:清华潜token压缩最多省89%","coca-cola-latent-tool-context-compression","2026-11-07T23:05:00+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"851c9f58-ab76-414d-86b4-4b10dc782422","Memorilla:16个记忆token反超全上下文","memorilla-latent-semantic-memory-tokens","2026-11-03T07:03:31+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"a820bb5a-ac48-4aab-8d8b-34da340900c1","REMORY 让 LLM 用 5% 上下文继续跑 Agent","remory-agent-memory","2026-10-11T09:00:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"b8ca86ba-2401-441d-ba5c-06cb6ef09420","随机抽样几乎就够了:高误差风险再砍50.2%","bestrand-benchmark-compression-random-sampling","2026-11-06T15:40:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"53d00d00-e4fb-41b0-a36a-37161203b514","KV缓存拼接复用丢题可至100%,修补要盯答案块","kv-cache-repair-budget-target-chunk","2026-11-05T15:09:02+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"64088ca6-c244-474c-93ce-cdbfe55de6d6","投机解码没用满?agent循环把加速比挖到10.87倍","specpilot-agentic-speculative-decoding","2026-11-04T13:15:17+00:00"]