按年被提到
20262026 · 3 次2026
强化学习与智能体›大模型智能体·arXiv:2609.37725 · cs.AI

Context Language Models

Rulin Shao, Shannon Zejiang Shen, Junjie Oscar Yin, Yuetai Li, Minheng Wang, Hamish Ivison, Radha Poovendran, Nathan Lambert, Teng Xiao, Mike Lewis, Wen-tau Yih, Luke Zettlemoyer 等 13 人 · 2026
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 3 人首次 2026-09-30 · 最近 2026-10-26
3不同的人
2活跃月份
3讨论串
0被引(OpenAlex)
arXiv 摘要(候选层不写导读)

We introduce Context Language Models (CLMs), language models that natively manage their own context. We implement this by treating the context as a file and allowing the model to make unrestricted updates to this file. This allows the model to learn what is most important to maintain in context, and naturally extends to multi-agent systems where multiple agent contexts coexist as files. Building CLMs zero-shot with existing models outperforms SOTA context management strategies across a variety of tasks: 11.4% higher accuracy with 21.5% fewer FLOPs on BrowseComp-Plus, 5% higher scores with 59% fewer FLOPs on 12-hour EdgeBench, and 65% greater improvement with the same compute on a 24-hour multi-repository agent-swarm task. Moreover, by shifting context management from external harness control to intrinsic model behavior, CLMs naturally enable both in-context and parametric learning of context-management strategies. We show that CLMs can be steered with natural-language instructions evolved through a standard skill-optimization loop, improving held-out accuracy by up to 35.9 points on a context-management task while reducing compute. We also introduce an online reinforcement learning method for CLMs, improving Qwen3.5-9B performance on BrowseComp-Plus by 47.6% while using 12% fewer FLOPs. Finally, we co-design Suffix Cache Reuse for CLM serving, further reducing server-side compute by 35% relative to standard SGLang at matched performance.

候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。

HN 书评

推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及
评论lurker137 · 2026-10

“This is why I think the next evolution will be giving models full control of their own context, where they can edit it just like a file. This idea was in a recent paper, where models could even be specifically trained for context management [1]. For the problem of the prefix cache being invalidated, they thought of implementing a suffix cache [1]: https:…”

The State of Context Engineering原评论 ↗
被提交为 story · 2 次