{"id":7684,"date":"2025-11-04T05:30:00","date_gmt":"2025-11-03T20:30:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7684"},"modified":"2025-11-02T10:34:10","modified_gmt":"2025-11-02T01:34:10","slug":"evaluating-long-term-memory-for-long-context-question-answering","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7684","title":{"rendered":"Evaluating Long-Term Memory for Long-Context Question Answering"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Evaluating Long-Term Memory for Long-Context Question Answering\u00a0<\/strong>[100.1]<br>\u8cea\u554f\u5fdc\u7b54\u30bf\u30b9\u30af\u306b\u30a2\u30ce\u30c6\u30fc\u30c8\u3057\u305f\u5408\u6210\u9577\u6587\u5bfe\u8a71\u306e\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u3042\u308bLoCoMo\u3092\u7528\u3044\u3066,\u30e1\u30e2\u30ea\u62e1\u5f35\u624b\u6cd5\u306e\u4f53\u7cfb\u7684\u8a55\u4fa1\u3092\u884c\u3046\u3002 \u4ee5\u4e0a\u306e\u7d50\u679c\u304b\u3089,\u30e1\u30e2\u30ea\u62e1\u5f35\u30a2\u30d7\u30ed\u30fc\u30c1\u306b\u3088\u308a\u30c8\u30fc\u30af\u30f3\u4f7f\u7528\u7387\u304c90%\u4ee5\u4e0a\u524a\u6e1b\u3055\u308c,\u7af6\u4e89\u7cbe\u5ea6\u304c\u5411\u4e0a\u3057\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2510.23730v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2510.23730v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Mon, 27 Oct 2025 18:03:50 GMT)<\/li>\n\n\n\n<li>\u9577\u6587\u306b\u304a\u3051\u308bMemory\u306e\u6709\u52b9\u6027\u3001\u300cOur findings show that memory-augmented approaches re- duce token usage by over 90% while maintain- ing competitive accuracy. Memory architecture complexity should scale with model capability, with small foundation models benefitting most from RAG, and strong instruction-tuned rea- soning model gaining from episodic learning through reflections and more complex agentic semantic memory.\u300d\u3068\u30e2\u30c7\u30eb\u30b5\u30a4\u30ba\u306e\u5f71\u97ff\u3001\u30bf\u30b9\u30af\u306b\u3088\u308bFull context\u3068\u306e\u6027\u80fd\u5dee\u3082\u8208\u5473\u6df1\u3044\u3002<\/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-7684","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\/7684","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=7684"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7684\/revisions"}],"predecessor-version":[{"id":7685,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7684\/revisions\/7685"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7684"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7684"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7684"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}