
{"id":13146,"date":"2026-10-10T03:27:33","date_gmt":"2026-10-09T19:27:33","guid":{"rendered":"https:\/\/infernews.com\/blog\/ebg\/"},"modified":"2026-10-10T03:30:04","modified_gmt":"2026-10-09T19:30:04","slug":"ebg","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/ebg\/","title":{"rendered":"EBG \u7d00\u9304\u9577\u6d41\u7a0b AI \u4ee3\u7406\u7684\u6c7a\u7b56"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/10\/EBG-AI.jpg\" alt=\"Overview of the EBG method\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u9577\u6d41\u7a0b AI \u4ee3\u7406\u57f7\u884c\u8edf\u4ef6\u5de5\u7a0b\u5de5\u4f5c\u6642\uff0c\u4f7f\u7528\u8005\u5f80\u5f80\u96e3\u4ee5\u5224\u65b7\u54ea\u4e9b\u6c7a\u5b9a\u6703\u5f71\u97ff\u7d50\u679c\u3002Evidence-Grounded Behavior Graph\uff08EBG\uff09\u662f\u4e00\u5957 training-free \u76e3\u5bdf\u65b9\u6cd5\uff0c\u5c07\u4f86\u6e90\u9023\u7d50\u8b49\u64da\u6574\u7406\u6210\u884c\u70ba\u3001\u7bc4\u570d\u8207\u95dc\u4fc2\uff0c\u5354\u52a9\u76e3\u5bdf\u8005\u8ffd\u67e5\u4ee3\u7406\u505a\u904e\u751a\u9ebc\uff0c\u4ee5\u53ca\u76f8\u95dc\u8b49\u64da\u85cf\u5728\u54ea\u88cf\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">EBG \u4e0d\u53ea\u8a18\u9304\u64cd\u4f5c\u6b21\u5e8f\uff0c\u800c\u662f\u4ee5 task-oriented views \u628a\u8981\u6c42\u3001\u89c0\u5bdf\u5230\u7684\u884c\u70ba\u548c\u53ef\u80fd\u5f8c\u679c\u9023\u8d77\u4f86\uff0c\u8b93\u4eba\u96c6\u4e2d\u6838\u5be6\u5177\u5f71\u97ff\u529b\u7684\u6c7a\u5b9a\u3002\u9019\u6bd4\u9010\u6bb5\u7ffb\u67e5\u8a0a\u606f\u3001\u7a0b\u5f0f\u78bc\u3001\u5de5\u5177\u8f38\u51fa\u53ca\u4e2d\u9593\u7522\u7269\u66f4\u9069\u5408\u9577\u6d41\u7a0b\u5de5\u4f5c\uff0c\u4f46\u5b83\u4ecd\u9700\u8981\u53ef\u9760\u7684\u4f86\u6e90\u8b49\u64da\u548c\u76e3\u5bdf\u6a21\u578b\u4f5c\u5224\u65b7\uff0c\u4e26\u4e0d\u6703\u81ea\u52d5\u4fdd\u8b49\u7d50\u8ad6\u6b63\u78ba\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u9805\u76ee\u540c\u6642\u63d0\u4f9b AgentMonBench\uff0c\u6db5\u84cb SpecGAP\u3001SilentSwap \u548c FeedbackTrace \u4e09\u500b\u5b50\u96c6\uff0c\u5404\u6709 100 \u500b\u4f8b\u5b50\uff1b\u4e3b\u8981\u5be6\u9a57\u4f7f\u7528 FeedbackTrace Long \u7684 100 \u500b\u6a23\u672c\u3002\u57fa\u6e96\u5f9e\u5169\u500b\u65b9\u5411\u6e2c\u8a66\u4ee3\u7406\u76e3\u5bdf\u80fd\u529b\uff1a\u8981\u6c42\u8207\u884c\u70ba\u662f\u5426\u4e00\u81f4\uff0c\u4ee5\u53ca\u80fd\u5426\u627e\u51fa\u9700\u8981\u4f7f\u7528\u8005\u6838\u5be6\u7684\u91cd\u5927\u81ea\u4e3b\u6c7a\u5b9a\uff1b\u7db2\u7ad9\u4ea6\u5c55\u793a\u516b\u500b\u6a21\u578b\u7684\u7d50\u679c\uff0c\u4f46\u5132\u5b58\u5eab\u6c92\u6709\u5728\u6458\u8981\u4e2d\u5217\u51fa\u5b8c\u6574\u5206\u6578\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Python 3.11 \u6216\u4ee5\u4e0a\u53ef\u57f7\u884c EBG\uff0cFeedbackTrace \u5efa\u69cb\u9700\u8981 Python 3.12\uff0cSilentSwap \u53e6\u9700 Docker\u3002<\/li>\n\n\n\n<li><code>scripts.demo<\/code> \u53ef\u96e2\u7dda\u5efa\u7acb\u5408\u6210\u5716\uff0c\u4e0d\u9700\u6191\u8b49\u6216\u57fa\u6e96\u8cc7\u6599\uff0c\u6e2c\u8a66\u7d50\u679c\u6703\u5132\u5b58\u81f3 <code>outputs\/demo\/<\/code>\u3002<\/li>\n\n\n\n<li>\u516c\u958b\u8cc7\u6599\u5305\u62ec\u5b8c\u6574\u8a55\u4f30\u8f38\u5165\u3001Gold annotations \u548c\u6a23\u672c\u6e05\u55ae\uff1b\u9810\u6e2c\u53ca judging \u5247\u9700\u81ea\u884c\u63d0\u4f9b endpoint\u3001\u6a21\u578b ID \u548c\u6191\u8b49\u3002<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u5c0d\u7814\u7a76AI \u5b89\u5168\u76e3\u5bdf\u53ca\u8edf\u4ef6\u5de5\u7a0b\u5718\u968a\u4f86\u8aaa\uff0cEBG \u7684\u628a\u300c\u4ee3\u7406\u505a\u4e86\u751a\u9ebc\u300d\u8f49\u5316\u6210\u53ef\u8ffd\u6eaf\u7684\u8b49\u64da\u7d50\u69cb\u3002\u5b83\u76ee\u524d\u4ecd\u662f under review at ICLR 2027 \u7684\u7814\u7a76\u9805\u76ee\uff0c\u8f03\u9069\u5408\u4f5c\u70ba\u53ef\u91cd\u73fe\u7684\u7814\u7a76\u57fa\u790e\u53ca\u76e3\u5bdf\u6d41\u7a0b\u539f\u578b\uff0c\u800c\u975e\u5373\u63d2\u5373\u7528\u7684\u751f\u7522\u7cfb\u7d71\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/zhk-lab.github.io\/agentmonbench-ebg\/\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/zhk-lab\/EBG\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u4ee3\u7406\u505a\u5f97\u8d8a\u591a\uff0c\u76e3\u5bdf\u8d8a\u96e3\u3002EBG \u5c07\u5206\u6563\u5728\u8a0a\u606f\u3001\u7a0b\u5f0f\u53ca\u5de5\u5177\u7d50\u679c\u4e2d\u7684\u8b49\u64da\uff0c\u6574\u7406\u6210\u53ef\u6838\u67e5\u7684\u884c\u70ba\u95dc\u4fc2\u5716\u3002<\/p>\n","protected":false},"author":8,"featured_media":13145,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"wpai_generated_summary":"","wpai_meta_description":"","footnotes":""},"categories":[133,76,127,137,24,38,152,150,170,189,199],"tags":[],"class_list":["post-13146","post","type-post","status-publish","format-standard","hentry","category-133","category-76","category-127","category-api","category-aisoftware","category-38","category-python","category-150","category-170","category-189","category-dataset-"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13146","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=13146"}],"version-history":[{"count":2,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13146\/revisions"}],"predecessor-version":[{"id":13150,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13146\/revisions\/13150"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/13145"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=13146"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=13146"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=13146"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}