{"id":9507,"date":"2026-06-22T14:53:48","date_gmt":"2026-06-22T06:53:48","guid":{"rendered":"https:\/\/infernews.com\/blog\/gatemem-benchmark\/"},"modified":"2026-06-22T14:54:12","modified_gmt":"2026-06-22T06:54:12","slug":"gatemem-benchmark","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/gatemem-benchmark\/","title":{"rendered":"GateMem\uff1a\u6e2c\u8a66 AI \u8a18\u61b6\u6709\u5187\u5206\u5bf8"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/06\/pasted-c14685cb4b35.jpg\" alt=\"GateMem logo\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u73fe\u6709\u8a18\u61b6\u57fa\u6e96\u591a\u6578\u96c6\u4e2d\u554f\u4e00\u4ef6\u4e8b\uff1a\u4ee3\u7406\u53ef\u5514\u53ef\u4ee5\u6b63\u78ba\u8a18\u4f4f\u8cc7\u6599\uff1bGateMem \u6539\u554f\u66f4\u63a5\u8fd1\u90e8\u7f72\u74b0\u5883\u7684\u554f\u984c\uff1a\u540c\u4e00\u500b shared memory \u4ffe\u591a\u500b principal \u5171\u7528\u6642\uff0c\u4ee3\u7406\u80fd\u5426\u6309\u89d2\u8272\u3001\u6388\u6b0a\u7bc4\u570d\u540c\u522a\u9664\u8981\u6c42\u53bb\u7ba1\u7406\u8cc7\u8a0a\u3002\u4f5c\u8005\u6279\u8a55\u820a\u7bc4\u5f0f\u504f\u5411 single-user recall\uff0c\u672a\u80fd\u53cd\u6620\u591a\u65b9\u5354\u4f5c\u5834\u666f\u5165\u9762\u6700\u5e38\u898b\u7684\u8d8a\u6b0a\u8b80\u53d6\u3001\u904e\u5ea6\u62ab\u9732\u540c\u522a\u9664\u5f8c\u91cd\u5efa\u8cc7\u8a0a\u98a8\u96aa\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">GateMem\u5c6c\u65bc<strong>Benchmark \/ Dataset \u6578\u64da\u96c6\u9805\u76ee<\/strong>\uff0c\u7528\u4f86\u8a55\u4f30 memory-augmented LLM agents \u5728 multi-principal shared-memory agents \u60c5\u5883\u4e0b\uff0c\u662f\u5426\u540c\u6642\u505a\u5230 Utility\u3001Access Control \u540c Active Forgetting\u3002\u5b83\u628a persistent memory \u8996\u70ba governed shared state\uff0c\u800c\u5514\u4fc2\u79c1\u4eba\u5feb\u53d6\uff0c\u9019\u500b framing \u4ee4\u6e2c\u8a66\u91cd\u9ede\u7531\u300c\u8a18\u5f97\u5e7e\u6e96\u300d\u8f49\u53bb\u300c\u5e7e\u6642\u61c9\u8a72\u7b54\u3001\u5e7e\u6642\u5514\u61c9\u8a72\u7b54\u300d\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u8cc7\u6599\u898f\u6a21\u5514\u7b97\u7d30\uff1a4 \u500b\u5834\u666f\u300191 \u500b long-form episodes\u30012,218 \u500b hidden checkpoints\uff0c\u6db5\u84cb Medical\u3001Office\u3001Education\u3001Household\u3002\u8a55\u5206\u6838\u5fc3\u6709\u4e00\u500b MGS \u6307\u6a19\uff1aMGS = U \u00b7 (1 \u2212 A) \u00b7 (1 \u2212 F)\uff0c\u5373\u4fc2\u6388\u6b0a\u4e0b\u8981\u6709\u7528\uff0c\u672a\u6388\u6b0a\u6642\u8981\u5c11\u6d29\u6f0f\uff0c\u522a\u9664\u5f8c\u4ea6\u5514\u53ef\u4ee5\u88ab\u78ba\u8a8d\u3001\u9084\u539f\u6216\u65c1\u6572\u5074\u64ca\u91cd\u5efa\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u8981\u7406\u89e3\u9ede\u6a23\u6e2c\uff0c\u91cd\u9ede\u4fc2\u7528\u5b83\u63d0\u4f9b\u7684 benchmark toolkit\u3001dataset \u540c leaderboard \u53bb\u8dd1\u4ee3\u7406\uff0c\u518d\u5c0d\u7167 hidden checkpoints \u7747\u8868\u73fe\u3002\u8f03\u53d7\u7528\u7684\u6703\u4fc2\u505a Agentic \u7cfb\u7d71\u3001\u9577\u671f\u8a18\u61b6\u4ee3\u7406\u3001\u4f01\u696d\u5167\u90e8\u52a9\u7406\u3001\u91ab\u7642\u6216\u6559\u80b2\u6d41\u7a0b\u81ea\u52d5\u5316\u7684\u5718\u968a\uff0c\u56e0\u70ba\u5462\u985e\u7cfb\u7d71\u6700\u6015\u7684\u901a\u5e38\u5514\u4fc2\u7b54\u932f\u4e00\u6b21\uff0c\u800c\u4fc2\u8a18\u5c0d\u5497\u4f46\u8b1b\u932f\u4eba\u807d\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u6838\u5fc3\u5dee\u7570\uff1a\u7531\u55ae\u4eba\u8a18\u61b6\u53ec\u56de\uff0c\u8f49\u6210\u591a\u89d2\u8272\u5171\u4eab\u8a18\u61b6\u6cbb\u7406<\/li>\n\n\n\n<li>\u4e09\u500b\u8a55\u6e2c\u9762\u5411\uff1a<strong>Utility\u3001Access Control\u3001Active Forgetting<\/strong><\/li>\n\n\n\n<li>\u5834\u666f\u8cbc\u8fd1\u6a5f\u69cb\u6d41\u7a0b\uff0c\u5305\u542b\u6388\u6b0a\u3001\u95dc\u4fc2\u8b8a\u5316\u3001\u522a\u9664\u8acb\u6c42<\/li>\n\n\n\n<li>\u76f8\u95dc\u6a21\u578b\u80cc\u666f\u5305\u62ec <strong>memory-augmented LLM agents<\/strong>\u3001persistent memory agents\uff0c\u540c\u9801\u9762\u4ea6\u63d0\u5230\u6e2c\u904e 6 backbone LLMs\u30017 memory baselines\uff0c\u4f46\u5177\u9ad4\u578b\u865f\u9700\u4ee5\u8ad6\u6587\u6216\u6392\u884c\u699c\u70ba\u6e96<\/li>\n\n\n\n<li>\u9650\u5236\u4fc2\u5b83\u4e3b\u8981\u8861\u91cf\u6cbb\u7406\u8868\u73fe\uff0c\u5514\u7b49\u65bc\u5b8c\u6574\u8986\u84cb\u6240\u6709\u771f\u5be6\u653f\u7b56\u3001\u6cd5\u898f\u6216\u7cfb\u7d71\u6574\u5408\u6210\u672c<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>GitHub\uff1a<\/strong> <a href=\"https:\/\/github.com\/rzhub\/GateMem\" rel=\"noopener noreferrer\">https:\/\/github.com\/rzhub\/GateMem<\/a><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>\u9805\u76ee\u4e3b\u9801\uff1a<\/strong> <a href=\"https:\/\/rzhub.github.io\/GateMem\/project.html\" rel=\"noopener noreferrer\">https:\/\/rzhub.github.io\/GateMem\/project.html<\/a><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Paper\uff1a<\/strong> <a href=\"https:\/\/arxiv.org\/pdf\/2606.18829\" rel=\"noopener noreferrer\">https:\/\/arxiv.org\/pdf\/2606.18829<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>GateMem \u5514\u4fc2\u8003\u4ee3\u7406\u8a18\u5514\u8a18\u5f97\uff0c\u800c\u4fc2\u8003\u4f62\u61c9\u5514\u61c9\u8a72\u8b1b\u3002\u5462\u500b\u57fa\u6e96\u76f4\u6307\u5171\u4eab\u8a18\u61b6\u4ee3\u7406\u6700\u96e3\u8655\u7406\u5605\u6b0a\u9650\u8207\u522a\u9664\u554f\u984c\u3002<\/p>\n","protected":false},"author":8,"featured_media":9506,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,116,144,170,76,197,199],"tags":[],"class_list":["post-9507","post","type-post","status-publish","format-standard","hentry","category-133","category-agentic","category-medical","category-170","category-76","category-framework","category-dataset-"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/9507","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/comments?post=9507"}],"version-history":[{"count":1,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/9507\/revisions"}],"predecessor-version":[{"id":9509,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/9507\/revisions\/9509"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/9506"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=9507"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=9507"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=9507"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}