{"id":7054,"date":"2025-07-09T05:06:00","date_gmt":"2025-07-08T20:06:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7054"},"modified":"2025-07-06T10:10:09","modified_gmt":"2025-07-06T01:10:09","slug":"memagent-reshaping-long-context-llm-with-multi-conv-rl-based-memory-agent","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7054","title":{"rendered":"MemAgent: Reshaping Long-Context LLM with Multi-Conv RL-based Memory Agent"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>MemAgent: Reshaping Long-Context LLM with Multi-Conv RL-based Memory Agent\u00a0<\/strong>[53.8]<br>\u6211\u3005\u306f,\u30bb\u30b0\u30e1\u30f3\u30c8\u5185\u306e\u30c6\u30ad\u30b9\u30c8\u3092\u8aad\u307f,\u4e0a\u66f8\u304d\u6226\u7565\u3092\u7528\u3044\u3066\u30e1\u30e2\u30ea\u3092\u66f4\u65b0\u3059\u308b\u65b0\u3057\u3044\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u30ef\u30fc\u30af\u30d5\u30ed\u30fc\u3067\u3042\u308bMemAgent\u3092\u7d39\u4ecb\u3057\u305f\u3002 MemAgent\u306f\u300132K\u30c6\u30ad\u30b9\u30c8\u3067\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3055\u308c\u305f8K\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u304b\u30893.5M QA\u30bf\u30b9\u30af\u3078\u306e\u5916\u633f\u304c\u53ef\u80fd\u3067\u3001\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u304c5%\u4f4e\u4e0b\u3057\u3001512K RULER\u30c6\u30b9\u30c8\u306795%\u4ee5\u4e0a\u3092\u5b9f\u73fe\u3057\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2507.02259v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2507.02259v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 03 Jul 2025 03:11:50 GMT)<\/li>\n\n\n\n<li>\u9577\u6587\u3092\u6271\u3046\u305f\u3081\u306eAgentic\u306a\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306e\u63d0\u6848\u3001\u4e0b\u8a18\u304c\u7279\u5fb4\u3068\u306e\u3053\u3068\uff08\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30b5\u30a4\u30c8\u3088\u308a\u5f15\u7528\uff09\n<ul class=\"wp-block-list\">\n<li>1 Novel memory mechanism: The agent reads text in segments and efficiently updates memory through an overwriting strategy. This design enables the model to process arbitrarily long inputs within a fixed context window, fundamentally overcoming the window length limitations of traditional Transformer architectures.<\/li>\n\n\n\n<li>2 O(n) complexity: By decoupling computation from text length, the complexity of processing long texts is transformed from quadratic growth to linear growth.<\/li>\n\n\n\n<li>3 RL-driven extrapolation: We enhance the DAPO algorithm to support multi-turn training over context-independent conversations. Based on this, the trained model exhibits unprecedented extrapolation performance.<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30b5\u30a4\u30c8\u306f<a href=\"https:\/\/memagent-sialab.github.io\/\">MemAgent: Reshaping Long-Context LLM with Multi-Conv RL based Memory Agent<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[244],"class_list":["post-7054","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-memory"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7054","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=7054"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7054\/revisions"}],"predecessor-version":[{"id":7055,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7054\/revisions\/7055"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7054"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7054"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7054"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}