智能体
x.com ·
@omarsar0
WFM：将 Agent 记忆存为链接式 Markdown Wiki 的论文
一篇论文提出 WFM（Wiki Foundation Model），将 LLM Wiki 转为图结构，并以查询为条件通过消息传递进行检索，使页面文本与链接结构共同影响结果。作者还构建了 GPU-to-GPU 训练协议，称训练速度提升 10.5 倍，并在五个 Agent 记忆与多跳推理基准上报告了较强结果。该方案面向以链接 Markdown 文件作为长期记忆的 Agent 场景。
AI推荐理由 该帖介绍了一篇关于 Agent 记忆的论文 WFM（Wiki Foundation Model），核心思路是将 LLM Wiki 转为图结构，通过以查询为条件的消息传递同时利用页面文本与链接结构进行检索，并提到 GPU-to-GPU 训练协议带来 10.5 倍加速及在五个 Agent 记忆与多跳推理基准上的结果。信息具体、指向可查论文链接，对 Agent 长期记忆方向有参考价值；但内容为个人转述，论文结果未经独立复现，训练加速与基准表现均为作者自报。
74
浏览 1,594
点赞 8
评论 4
转发 2
正文
Interesting paper on agent memory stored as a linked markdown wiki.
Lots of great ideas and insights if you work with LLM Wikis.
Wikis are useful for agents because each page holds dense text and the links between pages hold structure. WFM is a Wiki Foundation Model trained to use both at once.
It turns an LLM Wiki into a graph and retrieves from it with message passing conditioned on the query, so the text of each page and the link structure shape the result together.
The team also built a GPU-to-GPU training protocol that trains 10.5x faster, and reports strong results on five agent memory and multi-hop reasoning benchmarks.
If your agent's long-term memory is a folder of linked markdown files, WFM is designed for that format.
Paper: [arxiv.org/abs/2609.18182](https://arxiv.org/abs/2609.18182)
Chat with Paper: [academy.dair.ai/papers/wfm-wik…](https://academy.dair.ai/papers/wfm-wiki-foundation-model-for-complex-agentic-reasoning-2609.18182)
