
{"id":11173,"date":"2026-08-18T16:23:38","date_gmt":"2026-08-18T08:23:38","guid":{"rendered":"https:\/\/infernews.com\/blog\/pace-bench\/"},"modified":"2026-08-18T16:23:38","modified_gmt":"2026-08-18T08:23:38","slug":"pace-bench","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/pace-bench\/","title":{"rendered":"PACE-Bench\uff1a\u5c08\u6e2c\u6a5f\u68b0\u74b0\u5883\u8b8a\u5316\u4e0b\u7684\u81ea\u6211\u6f14\u5316\u80fd\u529b"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/pasted-45b67d6e7bcb.jpg\" alt=\"S-01 Bridge Construction: source design passes, fails after mutation, self-evolves, and passes the target environment\"><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">PACE-Bench \u662f\u4e00\u500b\u7528\u4f86\u6e2c\u8a66\u81ea\u6211\u6f14\u5316 agents \u7684\u57fa\u6e96\uff0c\u6838\u5fc3\u554f\u984c\u662f\uff1a\u7576\u539f\u672c\u53ef\u884c\u7684 code-driven design \u56e0\u70ba\u74b0\u5883\u6539\u52d5\u800c\u5931\u6548\uff0c\u7cfb\u7d71\u80fd\u5426\u9760\u4e92\u52d5\u56de\u994b\u628a\u5b83\u6539\u56de\u53ef\u884c\u3002\u5b83\u66f4\u50cf\u662f\u5728\u9a57\u8b49\u9069\u61c9\u80fd\u529b\uff0c\u800c\u5514\u53ea\u662f\u770b\u4e00\u6b21\u6027\u89e3\u984c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u9019\u500b\u9805\u76ee\u628a 144 \u7d44 source-to-target \u914d\u5c0d\u653e\u5165 6 \u985e\u7269\u7406\u5834\u666f\uff0c\u8981\u6c42 agent \u5148\u5728 source \u74b0\u5883\u6210\u529f\uff0c\u518d\u5728 hidden mutation \u5f8c\u7684 target \u74b0\u5883\u91cd\u65b0\u4fee\u6b63\u3002\u6bcf\u6b21\u4fee\u8a02\u90fd\u8981\u7d93\u904e\u57f7\u884c\u9a57\u8b49\uff0cJSON \u6703\u4fdd\u5b58\uff0c<code>--save-gif<\/code> \u4ea6\u53ef\u8a18\u9304\u6bcf\u6b21\u901a\u904e\u9a57\u8b49\u7684\u5617\u8a66\u52d5\u756b\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u540c\u985e\u8a55\u6e2c\u591a\u6578\u56fa\u5b9a\u57f7\u884c\u689d\u4ef6\uff0cPACE-Bench \u76f4\u63a5\u628a\u8b8a\u5316\u653e\u9032\u4efb\u52d9\u6838\u5fc3\uff0c\u903c\u7cfb\u7d71\u9762\u5c0d\u5931\u6557\u5f8c\u7684\u518d\u8a2d\u8a08\u3002\u5b83\u7528 OpenAI-compatible endpoint \u8dd1\u6a21\u578b\uff0c<code>--base-url<\/code>\u3001<code>--model<\/code>\u3001<code>--api-key<\/code> \u8207 <code>--workers<\/code> \u9019\u4e9b\u53c3\u6578\u90fd\u5c0d\u61c9\u5be6\u969b\u90e8\u7f72\u9700\u8981\uff0c\u8f03\u9069\u5408\u62ff\u4f86\u63a5\u81ea\u5bb6\u63a8\u7406\u670d\u52d9\u505a\u6279\u91cf\u8a55\u4f30\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5f9e\u516c\u958b\u7d50\u679c\u770b\uff0cReflexion \u5728 Pass@2 \u4e0a\u9818\u5148\uff0c\u4f46 Tree-of-Thoughts \u5728\u6210\u672c\u6548\u7387\u66f4\u9ad8\uff0c\u8aaa\u660e\u591a\u505a\u81ea\u6211\u4fee\u6b63\u4e0d\u4e00\u5b9a\u63db\u5230\u66f4\u597d\u7684\u6027\u50f9\u6bd4\u3002\u9019\u985e\u57fa\u6e96\u7279\u5225\u9069\u5408\u505a\u5177\u8eab\u667a\u80fd\u3001\u4eff\u771f\u63a7\u5236\u3001\u7814\u7a76\u81ea\u6211\u4fee\u6b63\u6d41\u7a0b\u7684\u5718\u968a\uff0c\u7528\u4f86\u6aa2\u67e5\u6a21\u578b\u5728\u65b0\u689d\u4ef6\u4e0b\u662f\u5426\u771f\u80fd\u9760\u56de\u994b\u6539\u5c0d\uff0c\u800c\u5514\u4fc2\u53ea\u4fc2\u8a18\u4f4f\u820a\u89e3\u6cd5\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u91dd\u5c0d\u74b0\u5883\u7a81\u8b8a\u5f8c\u7684\u9069\u61c9\u80fd\u529b\uff0c\u4e0d\u662f\u55ae\u6b21\u89e3\u984c<\/li>\n<li>\u4ee5\u57f7\u884c\u9a57\u8b49\u9a45\u52d5\u6bcf\u6b21\u4fee\u8a02\uff0c\u7d50\u679c\u53ef\u91cd\u73fe<\/li>\n<li>\u652f\u63f4\u63a5\u5165 OpenAI-compatible \u670d\u52d9\u505a\u8a55\u6e2c<\/li>\n<li>\u7d50\u679c\u986f\u793a\u6e96\u78ba\u7387\u8207\u6210\u672c\u6548\u7387\u672a\u5fc5\u540c\u6b65\u63d0\u5347<\/li>\n<li>\u9069\u5408\u7814\u7a76 self-evolving agents\u3001\u63a7\u5236\u8207\u7269\u7406\u6a21\u64ec<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/thunlp.github.io\/PACE-Bench\/\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/thunlp\/PACE-Bench\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a> \u00b7 <a href=\"https:\/\/huggingface.co\/datasets\/YuhaoZhan\/PACE-Bench\" rel=\"noopener noreferrer\"><strong>\u6a21\u578b<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>PACE-Bench \u91dd\u5c0d\u74b0\u5883\u7a81\u8b8a\u5f8c\u7684\u9069\u61c9\u529b\u505a\u8a55\u6e2c\uff0c\u91cd\u9ede\u4e0d\u662f\u80fd\u5426\u7b2c\u4e00\u6b21\u505a\u5c0d\uff0c\u800c\u662f\u80fd\u5426\u628a\u539f\u672c\u6210\u529f\u7684\u8a2d\u8a08\u6539\u5230\u518d\u6b21\u901a\u95dc\u3002\u4f60\u53ef\u4ee5\u628a\u5b83\u7406\u89e3\u6210\u4e00\u500b\u7528\u7269\u7406\u6a21\u64ec\u9a57\u8b49\u81ea\u6211\u4fee\u6b63\u80fd\u529b\u7684\u57fa\u6e96\u3002<\/p>\n","protected":false},"author":8,"featured_media":11172,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,185,116,137,199],"tags":[],"class_list":["post-11173","post","type-post","status-publish","format-standard","hentry","category-133","category-qwen","category-agentic","category-api","category-dataset-"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11173","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=11173"}],"version-history":[{"count":0,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11173\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/11172"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=11173"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=11173"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=11173"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}