{"id":6554,"date":"2025-11-28T10:17:10","date_gmt":"2025-11-28T02:17:10","guid":{"rendered":"https:\/\/nullthought.net\/?p=6554"},"modified":"2025-11-28T10:17:12","modified_gmt":"2025-11-28T02:17:12","slug":"agent0%ef%bc%9a%e4%b8%8d%e4%be%9d%e8%b5%96%e5%a4%96%e9%83%a8%e6%a0%87%e6%b3%a8%e6%95%b0%e6%8d%ae%ef%bc%8c%e5%9f%ba%e4%ba%8e%e5%8f%8c%e6%99%ba%e8%83%bd%e4%bd%93-%e5%b7%a5%e5%85%b7-%e5%bc%ba","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=6554","title":{"rendered":"Agent0\uff1a\u4e0d\u4f9d\u8d56\u5916\u90e8\u6807\u6ce8\u6570\u636e\uff0c\u57fa\u4e8e\u53cc\u667a\u80fd\u4f53 + \u5de5\u5177 + \u5f3a\u5316\u5b66\u4e60\uff0c\u8ba9\u57fa\u7840\u6a21\u578b\u5728\u63a8\u7406\u80fd\u529b\u4e0a\u6301\u7eed\u201c\u81ea\u6211\u5347\u7ea7\u201d"},"content":{"rendered":"\n<p>\u8bba\u6587<strong><a href=\"https:\/\/arxiv.org\/abs\/2511.16043\" target=\"_blank\" rel=\"noreferrer noopener\">Agent0: Unleashing Self-Evolving Agents from Zero Data via Tool-Integrated Reasoning<\/a><\/strong>\u4e3b\u8981\u63d0\u51fa\u4e86\u4e00\u4e2a\u53eb <strong>Agent0<\/strong> \u7684\u6846\u67b6\uff0c\u7528\u6765\u8ba9\u5927\u6a21\u578b\u201c\u81ea\u6211\u8fdb\u5316\u201d\uff0c\u800c\u4e14<strong>\u5b8c\u5168\u4e0d\u4f9d\u8d56\u4eba\u5de5\u6807\u6ce8\u6570\u636e<\/strong>\uff0c\u540c\u65f6\u53c8\u80fd\u5b66\u4f1a\u6bd4\u8f83\u590d\u6742\u7684\u5de5\u5177\u4f7f\u7528\u548c\u591a\u6b65\u63a8\u7406\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u4f5c\u8005\u4e3aPeng Xia, Kaide Zeng, Jiaqi Liu, Can Qin, Fang Wu, Yiyang Zhou, Caiming Xiong, Huaxiu Yao\uff0c\u6765\u81eaUNC-Chapel Hill, Salesforce Research, Stanford University\u3002<\/p>\n\n\n\n<p>\u4e00\u3001\u7814\u7a76\u80cc\u666f\uff1a\u4e3a\u4ec0\u4e48\u8981\u641e\u201c\u4ece\u96f6\u6570\u636e\u81ea\u8fdb\u5316\u7684 Agent\u201d<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u73b0\u6709\u505a\u6cd5\u7684\u74f6\u9888<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u73b0\u5728\u5f88\u591a\u201c\u667a\u80fd\u4f53\u578b\u201d\u5927\u6a21\u578b\uff08\u81ea\u52a8\u5199\u4ee3\u7801\u3001\u6df1\u5ea6\u68c0\u7d22\u3001\u590d\u6742\u63a8\u7406\u4e4b\u7c7b\uff09\u90fd\u662f\u9760 <strong>\u5f3a\u5316\u5b66\u4e60<\/strong> \u63d0\u5347\u80fd\u529b\uff0c\u6bd4\u5982\uff1a\n<ul class=\"wp-block-list\">\n<li>\u4ece\u4eba\u7c7b\u53cd\u9988\u6765\u7684 RL\uff08RLHF\uff09<\/li>\n\n\n\n<li>\u4ece\u53ef\u9a8c\u8bc1\u5956\u52b1\u6765\u7684 RL\uff08RLVR\uff09<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u8fd9\u4e9b\u65b9\u6cd5\u9ad8\u5ea6\u4f9d\u8d56<strong>\u5927\u89c4\u6a21\u3001\u4eba\u7c7b\u7cbe\u5fc3\u6784\u9020\u7684\u8bad\u7ec3\u6570\u636e<\/strong>\uff1a\u5199\u6307\u4ee4\u3001\u6253\u5206\u3001\u6807\u6b63\u786e\u7b54\u6848\u2026\u2026<\/li>\n\n\n\n<li>\u8fd9\u6837\u5e26\u6765\u51e0\u4e2a\u95ee\u9898\uff1a\n<ul class=\"wp-block-list\">\n<li>\u6807\u6570\u636e\u975e\u5e38\u8017\u94b1\u3001\u8017\u65f6\u95f4\uff0c\u6269\u5c55\u6027\u5dee\uff1b<\/li>\n\n\n\n<li>\u6a21\u578b\u7684\u80fd\u529b\u88ab\u4eba\u7c7b\u77e5\u8bc6\u548c\u6807\u6ce8\u901f\u5ea6\u201c\u5361\u6b7b\u201d\uff0c\u5f88\u96be\u7a81\u7834\u201c\u4eba\u80fd\u60f3\u5230\u7684\u9898\u4e0e\u7b54\u6848\u201d\u7684\u4e0a\u9650\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ol start=\"2\" class=\"wp-block-list\">\n<li>\u81ea\u8fdb\u5316\u6846\u67b6\u7684\u73b0\u6709\u9650\u5236<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u6700\u8fd1\u5174\u8d77\u5f88\u591a\u201c\u81ea\u6211\u535a\u5f08\u3001\u81ea\u6211\u51fa\u9898\u3001\u81ea\u6211\u6311\u6218\u201d\u7684\u6846\u67b6\uff0c\u8ba9\u6a21\u578b\u81ea\u5df1\u751f\u6210\u8bad\u7ec3\u6570\u636e\u3001\u81ea\u6211\u8fdb\u5316\u3002<\/li>\n\n\n\n<li>\u4f46\u4f5c\u8005\u6307\u51fa\u4e24\u5927\u786c\u4f24\uff1a<br>1\uff09<strong>\u88ab\u57fa\u7840\u6a21\u578b\u80fd\u529b\u201c\u5c01\u9876\u201d<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li>\u6a21\u578b\u81ea\u5df1\u51fa\u9898\u3001\u81ea\u5df1\u505a\u9898\uff0c\u9898\u76ee\u5f80\u5f80\u96be\u5ea6\u4e0d\u4f1a\u660e\u663e\u8d85\u8fc7\u5b83\u5f53\u524d\u6c34\u5e73\uff1b<\/li>\n\n\n\n<li>\u7ed3\u679c\u5c31\u662f\u201c\u81ea\u5a31\u81ea\u4e50\u201d\uff1a\u9898\u8d8a\u51fa\u8d8a\u50cf\u81ea\u5df1\u4f1a\u505a\u7684\u90a3\u7c7b\uff0c\u96be\u5ea6\u5f88\u96be\u6301\u7eed\u4e0a\u5347\uff0c\u5b66\u4e60\u505c\u6ede\u3002<br>2\uff09<strong>\u5927\u591a\u53ea\u652f\u6301\u5355\u8f6e\u4ea4\u4e92<\/strong>\uff1a<\/li>\n\n\n\n<li>\u5f88\u591a\u6846\u67b6\u53ea\u5728\u5355\u8f6e\u95ee\u7b54\u4e2d\u73a9\u81ea\u535a\u5f08\uff1b<\/li>\n\n\n\n<li>\u800c\u771f\u5b9e\u4e16\u754c\u95ee\u9898\u5f80\u5f80\u662f\u591a\u8f6e\u3001\u591a\u6b65\u9aa4\u3001\u591a\u5de5\u5177\u8c03\u7528\u7684\u590d\u6742\u8fc7\u7a0b\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u8fd9\u610f\u5473\u7740\uff1a\n<ul class=\"wp-block-list\">\n<li>\u8bfe\u7a0b\u96be\u5ea6\u4e0a\u4e0d\u53bb\uff1b<\/li>\n\n\n\n<li>\u65e0\u6cd5\u771f\u6b63\u5b66\u4f1a\u590d\u6742\u7684\u5de5\u5177\u4f7f\u7528\u548c\u591a\u6b65\u63a8\u7406\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u4e8c\u3001\u6838\u5fc3\u601d\u8def\uff1a\u53cc\u667a\u80fd\u4f53 + \u5de5\u5177 + \u5171\u8fdb\u5316<\/p>\n\n\n\n<p>\u8bba\u6587\u63d0\u51fa\u7684 <strong>Agent0<\/strong> \u60f3\u89e3\u51b3\u7684\u5c31\u662f\u4e0a\u9762\u4e24\u4e2a\u95ee\u9898\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u4e24\u4e2a\u89d2\u8272\u3001\u540c\u4e00\u4e2a\u5e95\u6a21<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ece\u540c\u4e00\u4e2a\u57fa\u7840 LLM\uff08\u5982 Qwen3-8B-Base\uff09\u4e2d\u521d\u59cb\u5316\u4e24\u4e2a\u201c\u89d2\u8272\u201d\uff1a\n<ul class=\"wp-block-list\">\n<li><strong>Curriculum Agent\uff08\u8bfe\u7a0b Agent\uff09<\/strong>\uff1a\u8d1f\u8d23\u51fa\u9898\uff0c\u751f\u6210\u201c\u524d\u6cbf\u4efb\u52a1\u201d\u2014\u2014\u65e2\u4e0d\u8981\u8fc7\u4e8e\u7b80\u5355\uff0c\u4e5f\u4e0d\u8981\u5b8c\u5168\u505a\u4e0d\u51fa\u6765\uff1b<\/li>\n\n\n\n<li><strong>Executor Agent\uff08\u6267\u884c Agent\uff09<\/strong>\uff1a\u8d1f\u8d23\u89e3\u9898\uff0c\u901a\u8fc7\u5f3a\u5316\u5b66\u4e60\u4e0d\u65ad\u63d0\u5347\u89e3\u9898\u548c\u5de5\u5177\u4f7f\u7528\u80fd\u529b\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ol start=\"2\" class=\"wp-block-list\">\n<li>\u201c\u5171\u751f\u7ade\u4e89\u201d\u7684\u81ea\u8fdb\u5316\u95ed\u73af<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u6574\u4f53\u6d41\u7a0b\u8fed\u4ee3\u8fdb\u884c\uff0c\u6bcf\u4e00\u8f6e\u5206\u4e24\u6b65\uff1a<br>1\uff09<strong>\u8bfe\u7a0b\u8fdb\u5316<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li>\u5728\u5f53\u524d Executor \u6c34\u5e73\u7684\u57fa\u7840\u4e0a\uff0c\u8bad\u7ec3 Curriculum\uff0c\u8ba9\u5b83\u51fa\u66f4\u80fd\u201c\u5361\u4f4f\u201dExecutor \u7684\u9898\u3002<br>2\uff09<strong>\u6267\u884c\u8fdb\u5316<\/strong>\uff1a<\/li>\n\n\n\n<li>\u56fa\u5b9a\u4f4f\u5f53\u524d Curriculum\uff0c\u8ba9\u5b83\u6279\u91cf\u51fa\u9898\uff0c\u8fc7\u6ee4\u51fa\u201c\u6700\u6709\u953b\u70bc\u4ef7\u503c\u7684\u9898\u201d\uff1b<\/li>\n\n\n\n<li>\u518d\u7528\u8fd9\u4e9b\u9898\u6765\u8bad\u7ec3 Executor\uff0c\u8ba9\u5b83\u7684\u80fd\u529b\u5f80\u4e0a\u8d70\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u4e24\u8005\u4e4b\u95f4\u662f\u201c\u4f60\u5f3a\u6211\u66f4\u5f3a\u201d\u7684\u5bf9\u6297\u5171\u751f\u5173\u7cfb\uff1a\n<ul class=\"wp-block-list\">\n<li>Executor \u53d8\u5f3a \u2192 Curriculum \u5fc5\u987b\u51fa\u66f4\u96be\u3001\u66f4\u9700\u8981\u5de5\u5177\u7684\u9898\uff1b<\/li>\n\n\n\n<li>Curriculum \u51fa\u9898\u53d8\u96be \u2192 Executor \u88ab\u8feb\u5b66\u4f1a\u66f4\u590d\u6742\u7684\u63a8\u7406\u4e0e\u5de5\u5177\u4f7f\u7528\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ol start=\"3\" class=\"wp-block-list\">\n<li>\u5de5\u5177\u96c6\u6210\uff1a\u7528\u4ee3\u7801\u89e3\u91ca\u5668\u6253\u7834\u5929\u82b1\u677f<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ec5\u9760\u8bed\u8a00\u6a21\u578b\u672c\u8eab\uff0c\u9898\u76ee\u96be\u5ea6\u4f1a\u88ab\u5b83\u7684\u201c\u8111\u5bb9\u91cf\u201d\u5c01\u9876\u3002<\/li>\n\n\n\n<li>Agent0 \u7684\u5173\u952e\u662f\u7ed9 Executor\u3001\u540c\u65f6\u4e5f\u7ed9 Curriculum \u63a5\u5165 <strong>\u4ee3\u7801\u89e3\u91ca\u5668\u5de5\u5177\uff08python \u6c99\u76d2\uff09<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li>\u53ef\u4ee5\u5728\u63a8\u7406\u8fc7\u7a0b\u4e2d\u5199\u4ee3\u7801\u3001\u6267\u884c\u3001\u62ff\u7ed3\u679c\u3001\u518d\u7ee7\u7eed\u63a8\u7406\uff1b<\/li>\n\n\n\n<li>\u8fd9\u76f8\u5f53\u4e8e\u5f15\u5165\u4e86\u4e00\u4e2a\u201c\u5916\u90e8\u8ba1\u7b97\u5668\u548c\u73af\u5883\u201d\uff0c\u8ba9\u6a21\u578b\u53ef\u4ee5\u501f\u52a9\u5de5\u5177\u505a\u66f4\u590d\u6742\u7684\u8ba1\u7b97\u4e0e\u9a8c\u8bc1\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u6b63\u56e0\u4e3a\u6709\u5de5\u5177\u52a0\u6210\uff0cExecutor \u80fd\u529b\u88ab\u663e\u8457\u62ac\u9ad8\uff0c\u4ece\u800c\u903c\u7740 Curriculum \u751f\u6210\u771f\u6b63\u9700\u8981\u5de5\u5177\u3001\u771f\u6b63\u590d\u6742\u7684\u9898\u76ee\uff0c\u800c\u4e0d\u662f\u53ea\u5728\u8111\u8865\u5c42\u9762\u6253\u8f6c\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"4\" class=\"wp-block-list\">\n<li>\u652f\u6301\u591a\u8f6e\u4ea4\u4e92<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Agent0 \u4e0d\u662f\u7b80\u5355\u7684\u5355\u8f6e\u95ee\u7b54\uff0c\u800c\u662f\uff1a\n<ul class=\"wp-block-list\">\n<li>\u53ef\u4ee5\u751f\u6210\u591a\u8f6e\u3001\u5e26\u4e0a\u4e0b\u6587\u7684\u4efb\u52a1\uff1b<\/li>\n\n\n\n<li>Executor \u7684\u89e3\u9898\u8fc7\u7a0b\u662f\u201c\u601d\u8003 \u2192 \u89e6\u53d1\u5de5\u5177 \u2192 \u83b7\u5f97\u8f93\u51fa \u2192 \u7ee7\u7eed\u601d\u8003 \u2192 \u6700\u7ec8\u7b54\u6848\u201d\u7684\u591a\u6b65\u8f68\u8ff9\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u8fd9\u66f4\u63a5\u8fd1\u771f\u5b9e\u4e16\u754c\u4e2d\u4eba\u7c7b\u89e3\u51b3\u95ee\u9898\u7684\u65b9\u5f0f\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u4e09\u3001\u8bfe\u7a0b Agent\uff1a\u5982\u4f55\u201c\u51fa\u597d\u9898\u201d<\/p>\n\n\n\n<p>\u8bfe\u7a0b Agent \u7684\u76ee\u6807\uff1a<strong>\u751f\u6210\u5bf9\u5f53\u524d Executor \u6765\u8bf4\u201c\u521a\u597d\u6709\u6311\u6218\u6027\u201d\u7684\u4efb\u52a1<\/strong>\uff0c\u7528\u4e00\u4e2a\u7efc\u5408\u5956\u52b1\u6765\u8861\u91cf\u9898\u76ee\u7684\u201c\u597d\u574f\u201d\u3002\u6838\u5fc3\u51e0\u4e2a\u4fe1\u53f7\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u5229\u7528\u6267\u884c Agent \u7684\u201c\u4e0d\u786e\u5b9a\u6027\u201d<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5bf9\u6bcf\u4e2a\u5019\u9009\u4efb\u52a1\uff0cExecutor \u4f1a\u5c1d\u8bd5\u591a\u6b21\u4f5c\u7b54\uff08\u6bd4\u5982\u91c7\u6837 10 \u6b21\u7b54\u6848\uff09\u3002<\/li>\n\n\n\n<li>\u7528\u201c\u591a\u6b21\u4f5c\u7b54\u7684\u4e00\u81f4\u7a0b\u5ea6\u201d\u6765\u8861\u91cf\uff1a\n<ul class=\"wp-block-list\">\n<li>\u5982\u679c\u5927\u90e8\u5206\u7b54\u6848\u90fd\u4e00\u6837 \u2192 \u4efb\u52a1\u592a\u7b80\u5355\uff1b<\/li>\n\n\n\n<li>\u5982\u679c\u5b8c\u5168\u4e71\u4e03\u516b\u7cdf\u3001\u6ca1\u6709\u660e\u663e\u591a\u6570\u7b54\u6848 \u2192 \u4efb\u52a1\u53ef\u80fd\u592a\u96be\u6216\u592a\u6df7\u4e71\uff1b<\/li>\n\n\n\n<li><strong>\u6700\u4f73\u7684\u60c5\u51b5\u662f\uff1a\u6709\u4e00\u5b9a\u5206\u6b67\uff0c\u4f46\u80fd\u5f62\u6210\u4e00\u4e2a\u660e\u663e\u7684\u591a\u6570\u6d3e<\/strong>\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u8bba\u6587\u628a\u8fd9\u4e2a\u201c\u4e00\u81f4\u7a0b\u5ea6\u201d\u6620\u5c04\u6210\u4e00\u4e2a\u5956\u52b1\uff1a\n<ul class=\"wp-block-list\">\n<li>\u5f53\u201c\u4e00\u81f4\u5ea6\u201d\u63a5\u8fd1 50% \u65f6\uff08\u5373 Executor \u975e\u5e38\u7ea0\u7ed3\uff09\u5956\u52b1\u6700\u9ad8\uff1b<\/li>\n\n\n\n<li>\u4e00\u81f4\u5ea6\u592a\u9ad8\u6216\u592a\u4f4e\u90fd\u4f1a\u88ab\u60e9\u7f5a\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ol start=\"2\" class=\"wp-block-list\">\n<li>\u660e\u786e\u5956\u52b1\u201c\u7528\u5de5\u5177\u7684\u4efb\u52a1\u201d<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4e3a\u4e86\u8ba9\u8bfe\u7a0b\u91cc\u591a\u51fa\u73b0\u9700\u8981\u5de5\u5177\u7684\u9898\uff0c\u4f5c\u8005\u76f4\u63a5\u52a0\u4e86\u4e00\u4e2a <strong>\u5de5\u5177\u4f7f\u7528\u5956\u52b1<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li>\u7edf\u8ba1 Executor \u5728\u89e3\u8fd9\u9053\u9898\u65f6\u8c03\u7528\u5de5\u5177\u7684\u6b21\u6570\uff08\u6bd4\u5982 python \u6267\u884c\u5757\u6570\uff09\uff1b<\/li>\n\n\n\n<li>\u6b21\u6570\u8d8a\u591a\uff0c\u5956\u52b1\u8d8a\u9ad8\uff0c\u4f46\u8bbe\u4e00\u4e2a\u4e0a\u9650\uff0c\u9632\u6b62\u6a21\u578b\u65e0\u8111\u591a\u6b21\u8c03\u7528\u5de5\u5177\u5237\u5206\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u8fd9\u6837 Curriculum \u5c31\u5b66\u4f1a\u4e86\uff1a\u201c\u51fa\u90a3\u4e9b\u903c Executor \u7528\u5de5\u5177\u624d\u80fd\u505a\u597d\u7684\u9898\uff0c\u5956\u52b1\u66f4\u5927\u201d\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"3\" class=\"wp-block-list\">\n<li>\u591a\u6837\u6027\u7ea6\u675f\uff1a\u907f\u514d\u4e00\u7a9d\u8702\u51fa\u540c\u4e00\u7c7b\u9898<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5982\u679c\u4e0d\u7ba1\uff0cCurriculum \u53ef\u80fd\u4f1a\u56f4\u7ed5\u67d0\u79cd\u201c\u597d\u62ff\u5206\u201d\u7684\u9898\u578b\u75af\u72c2\u590d\u5236\u3002<\/li>\n\n\n\n<li>\u8bba\u6587\u5bf9\u540c\u4e00\u6279\u4efb\u52a1\u505a\u7c7b\u4f3c\u201c\u805a\u7c7b\u201d\uff1a\n<ul class=\"wp-block-list\">\n<li>\u7528\u6587\u672c\u76f8\u4f3c\u5ea6\uff08\u4f8b\u5982\u57fa\u4e8e BLEU\uff09\u7684\u65b9\u5f0f\uff0c\u5224\u65ad\u54ea\u4e9b\u9898\u592a\u50cf\uff1b<\/li>\n\n\n\n<li>\u5bf9\u5c5e\u4e8e\u5927\u7c07\u3001\u91cd\u590d\u5ea6\u9ad8\u7684\u9898\u52a0\u60e9\u7f5a\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u8fd9\u8ba9\u8bfe\u7a0b\u4e2d\u7684\u4efb\u52a1\u7c7b\u578b\u66f4\u4e30\u5bcc\uff1a\u4e0d\u4ec5\u96be\uff0c\u800c\u4e14\u591a\u6837\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"4\" class=\"wp-block-list\">\n<li>\u5f62\u5f0f\u5408\u6cd5\u6027\u68c0\u67e5<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5728\u7efc\u5408\u5956\u52b1\u4e4b\u524d\uff0c\u8fd8\u4f1a\u505a <strong>\u683c\u5f0f\u68c0\u67e5<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li>\u9898\u76ee\u662f\u5426\u7b26\u5408\u9884\u671f\u6a21\u677f\u3001\u662f\u5426\u5305\u542b\u8981\u6c42\u7684\u5b57\u6bb5\u3001\u662f\u5426\u53ef\u89e3\u6790\u7b49\uff1b<\/li>\n\n\n\n<li>\u4e0d\u5408\u6cd5\u7684\u4efb\u52a1\u76f4\u63a5\u4e0d\u7ed9\u5956\u52b1\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ol start=\"5\" class=\"wp-block-list\">\n<li>\u7528\u5f3a\u5316\u5b66\u4e60\u8bad\u7ec3\u51fa\u9898\u7b56\u7565<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u6709\u4e86\u4e0a\u8ff0\u7efc\u5408\u5956\u52b1\u4e4b\u540e\uff0c\u518d\u7528\u7c7b\u4f3c PPO \/ GRPO \u8fd9\u79cd\u7b56\u7565\u68af\u5ea6\u65b9\u6cd5\u53bb\u66f4\u65b0 Curriculum\uff1a\n<ul class=\"wp-block-list\">\n<li>\u9488\u5bf9\u201c\u5982\u4f55\u51fa\u9898\u201d\u8fd9\u4ef6\u4e8b\u672c\u8eab\u505a RL\uff1b<\/li>\n\n\n\n<li>\u9010\u6e10\u5b66\u4f1a\uff1a\u5728\u5f53\u524d Executor \u6c34\u5e73\u4e0b\uff0c\u4ec0\u4e48\u6837\u7684\u95ee\u9898\u6700\u80fd\u63d0\u5347\u5bf9\u65b9\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u56db\u3001\u6267\u884c Agent\uff1a\u5982\u4f55\u5728\u566a\u58f0\u5f88\u5927\u7684\u81ea\u76d1\u7763\u73af\u5883\u4e2d\u7a33\u6b65\u53d8\u5f3a<\/p>\n\n\n\n<p>\u6267\u884c Agent \u7684\u76ee\u6807\uff1a<strong>\u5728 Curriculum \u51fa\u7684\u9898\u4e0a\u5c3d\u91cf\u7b54\u5bf9<\/strong>\uff0c\u540c\u65f6\u5b66\u4f1a\u5408\u7406\u5730\u8c03\u7528\u5de5\u5177\u3002<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u6311\u51fa\u201c\u6700\u6709\u8bad\u7ec3\u4ef7\u503c\u7684\u9898\u201d<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u6bcf\u8f6e\u8fed\u4ee3\u5b8c Curriculum \u540e\uff0c\u628a\u5b83\u51bb\u7ed3\uff0c\u7528\u6765\u5927\u89c4\u6a21\u51fa\u9898\uff0c\u5f62\u6210\u9898\u5e93\u3002<\/li>\n\n\n\n<li>\u5bf9\u9898\u5e93\u91cc\u7684\u6bcf\u9053\u9898\uff0c\u8ba9\u5f53\u524d Executor \u591a\u6b21\u4f5c\u7b54\uff0c\u7edf\u8ba1\u201c\u81ea\u4e00\u81f4\u6027\u201d\uff08\u591a\u6b21\u7b54\u6848\u662f\u5426\u4e00\u81f4\uff09\u3002<\/li>\n\n\n\n<li>\u53ea\u4fdd\u7559\u90a3\u4e9b\u201c\u4e00\u81f4\u5ea6\u5c45\u4e2d\u201d\u7684\u9898\u4f5c\u4e3a\u8bad\u7ec3\u96c6\uff1a\n<ul class=\"wp-block-list\">\n<li>\u592a\u7b80\u5355\uff08\u4e00\u81f4\u5ea6\u63a5\u8fd1 100%\uff09\uff1a\u518d\u8bad\u7ec3\u610f\u4e49\u4e0d\u5927\uff1b<\/li>\n\n\n\n<li>\u592a\u96be\u6216\u592a\u6df7\u4e71\uff08\u4e00\u81f4\u5ea6\u5f88\u4f4e\uff09\uff1a\u53ef\u80fd\u8fde Executor \u4e5f\u5b8c\u5168\u6478\u4e0d\u7740\u5934\u7eea\uff0c\u5bb9\u6613\u5b66\u574f\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u6700\u7ec8\u5f97\u5230\u7684\u662f\u4e00\u6279\u201c\u4f4d\u4e8e\u80fd\u529b\u8fb9\u754c\u9644\u8fd1\u201d\u7684\u9898\u2014\u2014\u5bf9 Executor \u6765\u8bf4\u65e2\u4e0d\u7b80\u5355\uff0c\u53c8\u4e0d\u662f\u5b8c\u5168\u4e0d\u53ef\u505a\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"2\" class=\"wp-block-list\">\n<li>\u591a\u6b65 + \u5de5\u5177\u7684\u89e3\u9898\u8f68\u8ff9<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u89e3\u6bcf\u9053\u9898\u65f6\uff0c\u4e0d\u662f\u4e00\u6b21\u6027\u5410\u51fa\u7b54\u6848\uff0c\u800c\u662f\uff1a<br>1\uff09\u5148\u8f93\u51fa\u4e00\u6bb5\u63a8\u7406\u6587\u5b57\uff1b<br>2\uff09\u5f53\u9700\u8981\u8ba1\u7b97\u6216\u9a8c\u8bc1\u65f6\uff0c\u751f\u6210\u4e00\u4e2a\u201c\u5de5\u5177\u8c03\u7528\u5757\u201d\uff08\u6bd4\u5982 <code>python ...<\/code>\uff09\uff1b<br>3\uff09\u6846\u67b6\u5728\u6c99\u76d2\u4e2d\u6267\u884c\u4ee3\u7801\uff0c\u5c06\u7ed3\u679c\u5c01\u88c5\u6210 <code>output ...<\/code> \u53cd\u9988\u7ed9\u6a21\u578b\uff1b<br>4\uff09\u6a21\u578b\u7ee7\u7eed\u57fa\u4e8e\u8fd9\u4e2a\u53cd\u9988\u63a8\u7406\uff0c\u53ef\u80fd\u518d\u6b21\u8c03\u7528\u5de5\u5177\uff0c\u76f4\u81f3\u7ed9\u51fa\u6700\u7ec8\u7b54\u6848\uff08\u7528\u7279\u5b9a\u6807\u8bb0\u5305\u4f4f\uff09\u3002<\/li>\n\n\n\n<li>\u8fd9\u6574\u4e2a\u8fc7\u7a0b\u5f62\u6210\u4e00\u6761<strong>\u6df7\u5408\u4e86\u81ea\u7136\u8bed\u8a00\u548c\u4ee3\u7801\u6267\u884c\u7ed3\u679c\u7684\u957f\u8f68\u8ff9<\/strong>\uff0c\u5f3a\u5316\u5b66\u4e60\u5c31\u662f\u5728\u8fd9\u4e9b\u8f68\u8ff9\u4e0a\u505a\u7684\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"3\" class=\"wp-block-list\">\n<li>\u7528\u591a\u6570\u6295\u7968\u6784\u9020\u4f2a\u6807\u7b7e<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5bf9\u540c\u4e00\u9898\u91c7\u6837\u591a\u6761\u5b8c\u6574\u89e3\u9898\u8f68\u8ff9\uff0c\u62ff\u5230\u591a\u4e2a\u6700\u7ec8\u7b54\u6848\u3002<\/li>\n\n\n\n<li>\u7528<strong>\u591a\u6570\u6295\u7968<\/strong>\u9009\u51fa\u4e00\u4e2a\u201c\u4f2a\u6b63\u786e\u7b54\u6848\u201d\uff0c\u628a\u6bcf\u6761\u8f68\u8ff9\u7684\u6700\u540e\u7b54\u6848\u548c\u5b83\u6bd4\uff1a\n<ul class=\"wp-block-list\">\n<li>\u76f8\u540c \u2192 \u8fd9\u4e00\u8f68\u8ff9\u5f97\u5230\u6b63\u5956\u52b1\uff1b<\/li>\n\n\n\n<li>\u4e0d\u540c \u2192 \u5f97\u5230\u96f6\u6216\u8d1f\u5956\u52b1\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u8fd9\u6837\u5c31\u4e0d\u9700\u8981\u5916\u90e8\u6807\u6ce8\uff1a\u662f\u6a21\u578b\u81ea\u5df1\u7ed9\u81ea\u5df1\u6253\u5206\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"4\" class=\"wp-block-list\">\n<li>ADPO\uff1a\u5728\u201c\u6709\u6b67\u4e49\u7684\u4f2a\u6807\u7b7e\u201d\u4e0b\u505a\u66f4\u806a\u660e\u7684\u5f3a\u5316\u5b66\u4e60<\/li>\n<\/ol>\n\n\n\n<p>\u56e0\u4e3a\u4f2a\u6807\u7b7e\u6765\u81ea\u591a\u6570\u6295\u7968\uff0c\u672c\u8eab\u53ef\u80fd\u662f\u9519\u7684\uff0c\u800c\u4e14\u9898\u76ee\u6709\u65f6\u5f88\u6a21\u68f1\u4e24\u53ef\u3002\u76f4\u63a5\u5f53\u6210\u771f\u6807\u7b7e\u53bb\u5f3a\u5316\u5b66\u4e60\uff0c\u5bb9\u6613\u5b66\u574f\u3002\u56e0\u6b64\u4f5c\u8005\u63d0\u51fa\u4e86\u4e00\u4e2a\u53eb <strong>ADPO\uff08\u6a21\u7cca\u611f\u77e5\u7684\u52a8\u6001\u7b56\u7565\u4f18\u5316\uff09<\/strong> \u7684\u6539\u8fdb\u7248\u672c\uff0c\u5927\u81f4\u505a\u4e86\u4e24\u4ef6\u4e8b\uff1a<\/p>\n\n\n\n<p>\uff081\uff09\u6839\u636e\u201c\u9898\u76ee\u6b67\u4e49\u5ea6\u201d\u8c03\u6574\u5b66\u4e60\u5f3a\u5ea6<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u9898\u76ee\u7684\u4e00\u81f4\u5ea6\u8d8a\u4f4e\uff0c\u8bf4\u660e\u6a21\u578b\u4e4b\u95f4\u5206\u6b67\u8d8a\u5927\uff0c\u8fd9\u79cd\u9898\u7684\u4f2a\u6807\u7b7e\u8d8a\u53ef\u80fd\u662f\u9519\u7684\u3002<\/li>\n\n\n\n<li>ADPO \u4f1a\u6839\u636e\u4e00\u81f4\u5ea6\u5bf9\u6bcf\u4e2a\u6837\u672c\u7684\u201c\u4f18\u52bf\u503c\u201d\uff08\u76f8\u5f53\u4e8e\u68af\u5ea6\u6743\u91cd\uff09\u505a\u7f29\u653e\uff1a\n<ul class=\"wp-block-list\">\n<li>\u4e00\u81f4\u5ea6\u9ad8\uff08\u6a21\u578b\u6bd4\u8f83\u786e\u5b9a\uff09 \u2192 \u4fdd\u7559\u8f83\u5f3a\u7684\u66f4\u65b0\u4fe1\u53f7\uff1b<\/li>\n\n\n\n<li>\u4e00\u81f4\u5ea6\u4f4e\uff08\u9ad8\u6b67\u4e49\uff09 \u2192 \u628a\u68af\u5ea6\u6743\u91cd\u6253\u6298\uff0c\u9632\u6b62\u6a21\u578b\u8fc7\u5ea6\u76f8\u4fe1\u53ef\u80fd\u662f\u9519\u7684\u4f2a\u6807\u7b7e\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\uff082\uff09\u9488\u5bf9\u9ad8\u6b67\u4e49\u9898\u653e\u5bbd\u201c\u4fe1\u4efb\u533a\u95f4\u201d\uff0c\u9f13\u52b1\u63a2\u7d22<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4f20\u7edf PPO \u4e2d\u6709\u4e00\u4e2a\u201c\u622a\u65ad\u6bd4\u7387\u201d\u7684\u673a\u5236\uff0c\u9632\u6b62\u7b56\u7565\u4e00\u6b65\u8de8\u592a\u5927\u5bfc\u81f4\u8bad\u7ec3\u4e0d\u7a33\u5b9a\u3002<\/li>\n\n\n\n<li>\u8bba\u6587\u7684\u89c2\u5bdf\u662f\uff1a\n<ul class=\"wp-block-list\">\n<li>\u5bf9\u5f88\u591a\u4f4e\u6982\u7387 token\uff0c\u8fd9\u4e2a\u201c\u4e0a\u622a\u65ad\u201d\u7ecf\u5e38\u88ab\u89e6\u53d1\uff0c\u53cd\u800c\u4f1a<strong>\u6291\u5236\u63a2\u7d22\u65b0\u7684\u63a8\u7406\u8def\u5f84<\/strong>\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>ADPO \u6839\u636e\u9898\u76ee\u4e00\u81f4\u5ea6\uff0c<strong>\u52a8\u6001\u8c03\u6574\u8fd9\u4e2a\u4e0a\u622a\u65ad\u7684\u5bbd\u5ea6<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li>\u5bf9\u9ad8\u5ea6\u4e0d\u786e\u5b9a\u7684\u9898\uff0c\u9002\u5f53\u653e\u5bbd\u7ea6\u675f\uff0c\u8ba9\u7b56\u7565\u53ef\u4ee5\u5927\u6b65\u8c03\u6574\uff0c\u628a\u4e00\u4e9b\u201c\u539f\u6765\u6982\u7387\u5f88\u4f4e\u4f46\u53ef\u80fd\u662f\u6b63\u786e\u65b9\u5411\u201d\u7684 token \u6d8c\u73b0\u51fa\u6765\uff1b<\/li>\n\n\n\n<li>\u5bf9\u5f88\u786e\u5b9a\u7684\u9898\uff0c\u4fdd\u6301\u4e25\u683c\u7ea6\u675f\uff0c\u4fdd\u8bc1\u7a33\u5b9a\u548c\u4e0d\u9057\u5fd8\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u6574\u4f53\u6548\u679c\uff1a\u5728\u591a\u6570\u6295\u7968\u4f2a\u6807\u7b7e\u7684\u566a\u58f0\u73af\u5883\u4e0b\uff0cExecutor \u4ecd\u80fd\u6bd4\u8f83\u7a33\u5730\u5411\u201c\u66f4\u597d\u3001\u66f4\u4f1a\u7528\u5de5\u5177\u201d\u7684\u65b9\u5411\u6536\u655b\u3002<\/p>\n\n\n\n<p>\u4e94\u3001\u5b9e\u9a8c\u7ed3\u679c\uff1a\u771f\u7684\u53d8\u5f3a\u4e86\u5417\uff1f<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u5b9e\u9a8c\u8bbe\u7f6e<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5b9e\u73b0\u57fa\u4e8e VeRL \u6846\u67b6\uff0c\u57fa\u7840\u6a21\u578b\u9009\u7528\uff1a\n<ul class=\"wp-block-list\">\n<li>Qwen3-4B-Base<\/li>\n\n\n\n<li>Qwen3-8B-Base<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u6bcf\u9898\u91c7\u6837 10 \u6761\u56de\u7b54\u7528\u4e8e\u4e0d\u786e\u5b9a\u6027\u4f30\u8ba1\u548c\u591a\u6570\u6295\u7968\u3002<\/li>\n\n\n\n<li>Self-consistency \u8fc7\u6ee4\u9608\u503c\u8bbe\u7f6e\u6210\u4e00\u4e2a\u533a\u95f4\uff0c\u53ea\u4fdd\u7559\u4e2d\u7b49\u4e00\u81f4\u5ea6\u7684\u9898\u3002<\/li>\n\n\n\n<li>\u5de5\u5177\u662f\u4e00\u4e2a\u6c99\u76d2\u4ee3\u7801\u89e3\u91ca\u5668\uff0c\u6267\u884c\u5e26\u7279\u5b9a\u6807\u8bb0\u7684 python \u4ee3\u7801\u5757\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"2\" class=\"wp-block-list\">\n<li>\u5bf9\u6bd4\u5bf9\u8c61<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4e0d\u505a\u4efb\u4f55\u8bad\u7ec3\u7684\u57fa\u7840\u6a21\u578b\uff08\u6709\u65e0\u5de5\u5177\u7248\u672c\uff09\uff1b<\/li>\n\n\n\n<li>\u591a\u79cd\u201c\u96f6\u6570\u636e\u81ea\u8fdb\u5316\u201d\u65b9\u6cd5\uff1a\n<ul class=\"wp-block-list\">\n<li>R-Zero<\/li>\n\n\n\n<li>Absolute Zero\uff08\u4e5f\u6709\u4ee3\u7801\u6267\u884c\u5668\uff0c\u4f46\u4e3b\u8981\u7528\u4e8e\u9a8c\u8bc1\uff09<\/li>\n\n\n\n<li>SPIRAL<\/li>\n\n\n\n<li>Socratic-Zero \u7b49\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ol start=\"3\" class=\"wp-block-list\">\n<li>\u6570\u5b66\u63a8\u7406\u4e0a\u7684\u63d0\u5347<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ee5 Qwen3-8B \u4e3a\u4f8b\uff1a\n<ul class=\"wp-block-list\">\n<li>\u57fa\u7840\u6a21\u578b\u5728\u591a\u6570\u5b66\u57fa\u51c6\u4e0a\u7684\u5e73\u5747\u51c6\u786e\u7387\u5927\u7ea6\u5728 49 \u5de6\u53f3\uff1b<\/li>\n\n\n\n<li>\u52a0\u5de5\u5177\u4f46\u4e0d\u8bad\u7ec3\u7565\u6709\u63d0\u5347\uff1b<\/li>\n\n\n\n<li>R-Zero\u3001Socratic-Zero \u7b49\u65b9\u6cd5\u80fd\u63d0\u5347\u5230 54\uff5e56 \u5de6\u53f3\uff1b<\/li>\n\n\n\n<li><strong>Agent0 \u53ef\u4ee5\u5230\u7ea6 58 \u5de6\u53f3<\/strong>\uff0c\u63d0\u5347\u660e\u663e\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u5bf9 4B \u6a21\u578b\u4e5f\u6709\u7c7b\u4f3c\u8d8b\u52bf\uff0cAgent0 \u57fa\u672c\u90fd\u662f\u540c\u7c7b\u65b9\u6cd5\u4e2d\u7684\u6700\u4f18\u6216\u63a5\u8fd1\u6700\u4f18\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"4\" class=\"wp-block-list\">\n<li>\u6cdb\u5316\u5230\u901a\u7528\u63a8\u7406\u4efb\u52a1<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8bba\u6587\u8fd8\u5728 SuperGPQA\u3001MMLU-Pro\u3001BBEH \u8fd9\u4e9b\u66f4\u504f\u901a\u7528\u77e5\u8bc6\u548c\u590d\u6742\u63a8\u7406\u7684\u57fa\u51c6\u4e0a\u6d4b\u8bd5\u3002<\/li>\n\n\n\n<li>\u5373\u4f7f\u8bad\u7ec3\u4e3b\u6218\u573a\u662f\u6570\u5b66\u63a8\u7406\uff0c<strong>Agent0 \u4f9d\u7136\u80fd\u5728\u8fd9\u4e9b\u901a\u7528\u4efb\u52a1\u4e0a\u5e26\u6765 20% \u5de6\u53f3\u7684\u76f8\u5bf9\u63d0\u5347<\/strong>\uff0c\u5e76\u8d85\u8fc7\u5176\u4ed6\u81ea\u8fdb\u5316\u65b9\u6cd5\u3002<\/li>\n\n\n\n<li>\u8bf4\u660e\uff1a\n<ul class=\"wp-block-list\">\n<li>\u901a\u8fc7\u5de5\u5177\u9a71\u52a8\u3001\u591a\u8f6e\u63a8\u7406\u5f62\u6210\u7684\u201c\u62bd\u8c61\u63a8\u7406\u80fd\u529b\u201d\uff0c\u53ef\u4ee5\u5728\u6570\u5b66\u4ee5\u5916\u7684\u9886\u57df\u8fc1\u79fb\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ol start=\"5\" class=\"wp-block-list\">\n<li>\u6d88\u878d\u5b9e\u9a8c\u7684\u7ed3\u8bba<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u53bb\u6389\u8bfe\u7a0b Agent \u7684\u8bad\u7ec3\uff0c\u76f4\u63a5\u7528\u201c\u672a\u8bad\u7ec3\u7684\u51fa\u9898\u5668\u201d \u2192 \u6027\u80fd\u5927\u5e45\u4e0b\u964d\uff0c\u8bf4\u660e\u201c\u5b66\u4f1a\u5982\u4f55\u51fa\u9898\u201d\u672c\u8eab\u5f88\u5173\u952e\uff1b<\/li>\n\n\n\n<li>\u53bb\u6389\u5de5\u5177\u5956\u52b1 \u2192 \u5de5\u5177\u76f8\u5173\u9898\u76ee\u51cf\u5c11\uff0c\u6574\u4f53\u6027\u80fd\u660e\u663e\u53d8\u5dee\uff0c\u8bf4\u660e\u660e\u786e\u5956\u52b1\u201c\u9700\u8981\u5de5\u5177\u7684\u9898\u201d\u975e\u5e38\u91cd\u8981\uff1b<\/li>\n\n\n\n<li>\u53bb\u6389\u591a\u6837\u6027\u60e9\u7f5a \u2192 \u9898\u578b\u66f4\u5355\u4e00\uff0c\u5c24\u5176\u662f\u901a\u7528\u4efb\u52a1\u4e0a\u7684\u8868\u73b0\u4e0b\u964d\uff1b<\/li>\n\n\n\n<li>Executor \u4e0d\u7528 ADPO \u6539\u8fdb\u3001\u53ea\u7528\u666e\u901a GRPO \u2192 \u6027\u80fd\u7565\u5fae\u4e0b\u964d\uff0c\u8bc1\u660e\u201c\u6309\u6b67\u4e49\u52a8\u6001\u8c03\u6574\u5b66\u4e60\u529b\u5ea6\u548c\u622a\u65ad\u201d\u786e\u5b9e\u6709\u5e2e\u52a9\uff1b<\/li>\n\n\n\n<li>\u53bb\u6389\u591a\u8f6e\u63a8\u7406\uff0c\u53ea\u7528\u5355\u8f6e \u2192 \u6570\u5b66\u7b49\u590d\u6742\u4efb\u52a1\u7684\u8868\u73b0\u660e\u663e\u53d8\u5dee\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"6\" class=\"wp-block-list\">\n<li>\u5171\u8fdb\u5316\u8fc7\u7a0b\u7684\u6f14\u5316\u73b0\u8c61<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ed6\u4eec\u56fa\u5b9a\u7b2c\u4e00\u8f6e\u8bad\u7ec3\u5f97\u5230\u7684 Executor\uff0c\u7528\u5b83\u53bb\u505a\u7b2c 1\u30012\u30013 \u8f6e Curriculum \u5206\u522b\u751f\u6210\u7684\u9898\u96c6\uff1a\n<ul class=\"wp-block-list\">\n<li>\u901a\u8fc7\u7387\u4ece\u7b2c 1 \u8f6e\u5230\u7b2c 3 \u8f6e\u6301\u7eed\u4e0b\u964d\uff0c\u8bf4\u660e\u9898\u786e\u5b9e\u5728\u53d8\u96be\uff1b<\/li>\n\n\n\n<li>\u540c\u65f6\u5e73\u5747\u5de5\u5177\u8c03\u7528\u6b21\u6570\u4ece\u7ea6 1.6 \u6b21\u5347\u5230 2.6 \u6b21\uff0c\u8bf4\u660e\u9898\u76ee\u8d8a\u6765\u8d8a\u4f9d\u8d56\u5de5\u5177\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u8fd9\u5370\u8bc1\u4e86\u4f5c\u8005\u7684\u4e3b\u5f20\uff1a\n<ul class=\"wp-block-list\">\n<li>\u5de5\u5177\u52a0\u6301\u7684 Executor \u53d8\u5f3a \u2192 \u8bfe\u7a0b Agent \u51fa\u9898\u66f4\u96be\u3001\u66f4\u5de5\u5177\u5316\uff1b<\/li>\n\n\n\n<li>\u6574\u4e2a\u7cfb\u7edf\u5f62\u6210\u4e86\u4e00\u4e2a\u6b63\u53cd\u9988\u7684\u201c\u87ba\u65cb\u5f0f\u4e0a\u5347\u201d\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u516d\u3001\u4e0e\u76f8\u5173\u5de5\u4f5c\u7684\u5173\u7cfb<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u76f8\u6bd4\u4e00\u822c\u81ea\u8fdb\u5316\u6846\u67b6<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ee5\u524d\u7684\u81ea\u8fdb\u5316\u65b9\u6cd5\u5927\u591a\uff1a\n<ul class=\"wp-block-list\">\n<li>\u4e0d\u5e26\u5916\u90e8\u5de5\u5177\uff0c\u96be\u5ea6\u88ab\u6a21\u578b\u672c\u8eab\u80fd\u529b\u5c01\u9876\uff1b<\/li>\n\n\n\n<li>\u6216\u8005\u53ea\u7528\u5916\u90e8\u5de5\u5177\u505a\u9a8c\u8bc1\uff0c\u6ca1\u628a\u201c\u5de5\u5177\u4f7f\u7528\u672c\u8eab\u201d\u7eb3\u5165\u8bfe\u7a0b\u8bbe\u8ba1\u4e0e\u7b56\u7565\u5b66\u4e60\u7684\u6838\u5fc3\u73af\u8282\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>Agent0 \u7684\u7279\u70b9\uff1a\n<ul class=\"wp-block-list\">\n<li><strong>\u5de5\u5177\u96c6\u6210\u662f\u53cc\u5411\u7684<\/strong>\uff1a\u51fa\u9898\u4e5f\u8003\u8651\u5de5\u5177\u3001\u89e3\u9898\u66f4\u4f9d\u8d56\u5de5\u5177\uff1b<\/li>\n\n\n\n<li>\u901a\u8fc7\u5956\u52b1\u8bbe\u8ba1\uff0c\u8ba9\u201c\u9700\u8981\u591a\u6b65\u63a8\u7406 + \u5de5\u5177\u8c03\u7528\u201d\u7684\u9898\u81ea\u7136\u5360\u636e\u8bfe\u7a0b\u4e3b\u6d41\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ol start=\"2\" class=\"wp-block-list\">\n<li>\u76f8\u6bd4\u4e00\u822c\u5de5\u5177\u5f3a\u5316\u5b66\u4e60\uff08TIR\uff09<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8bb8\u591a\u5de5\u5177\u5f3a\u5316\u5b66\u4e60\u5de5\u4f5c\u4f9d\u8d56\u4eba\u5de5\u6784\u9020\u7684\u6570\u636e\u96c6\u6216\u4efb\u52a1\u73af\u5883\uff1b<\/li>\n\n\n\n<li>Agent0 \u5219\u5f3a\u8c03\u201c<strong>\u4ece\u96f6\u5916\u90e8\u6570\u636e<\/strong>\u201d\u51fa\u53d1\uff0c\u901a\u8fc7\u81ea\u8fdb\u5316\u751f\u6210\u6240\u6709\u8bad\u7ec3\u4fe1\u53f7\uff1b<\/li>\n\n\n\n<li>\u8fd9\u4f7f\u5b83\u66f4\u50cf\u4e00\u79cd\u201c\u9762\u5411\u672a\u77e5\u73af\u5883\u7684\u5927\u89c4\u6a21\u81ea\u5b66\u673a\u5236\u201d\uff0c\u800c\u4e0d\u4ec5\u662f\u201c\u5728\u56fa\u5b9a\u4efb\u52a1\u4e0a\u8c03\u4f18\u5de5\u5177\u4f7f\u7528\u7b56\u7565\u201d\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u4e03\u3001\u603b\u7ed3\u4e0e\u53ef\u80fd\u7684\u5c40\u9650<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u521b\u65b0\u70b9\u6982\u62ec<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u628a\u201c\u51fa\u9898\u201d\u548c\u201c\u89e3\u9898\u201d\u5206\u6210\u4e24\u4e2a\u89d2\u8272\uff0c\u4e14\u90fd\u53ef\u7528\u5de5\u5177\u3001\u591a\u8f6e\u63a8\u7406\uff1b<\/li>\n\n\n\n<li>\u7528 Executor \u7684\u201c\u4e0d\u786e\u5b9a\u6027 + \u5de5\u5177\u4f7f\u7528\u9891\u7387 + \u591a\u6837\u6027\u201d\u8bbe\u8ba1\u51fa\u9898\u5956\u52b1\uff0c\u771f\u6b63\u56f4\u7ed5\u201c\u80fd\u529b\u8fb9\u754c\u201d\u6784\u5efa\u8bfe\u7a0b\uff1b<\/li>\n\n\n\n<li>\u5728\u4f2a\u6807\u7b7e\u3001\u566a\u58f0\u6781\u5927\u7684\u81ea\u76d1\u7763\u573a\u666f\u4e0b\uff0c\u63d0\u51fa ADPO \u8fd9\u79cd\u8003\u8651\u6b67\u4e49\u5ea6\u7684\u5f3a\u5316\u5b66\u4e60\u4f18\u5316\u7b56\u7565\uff1b<\/li>\n\n\n\n<li>\u5728\u591a\u6570\u5b66\u548c\u901a\u7528\u63a8\u7406\u57fa\u51c6\u4e0a\u53d6\u5f97\u4e86\u6bd4\u73b0\u6709\u591a\u79cd\u201c\u96f6\u6570\u636e\u81ea\u8fdb\u5316\u201d\u65b9\u6cd5\u66f4\u597d\u7684\u6210\u7ee9\u3002<\/li>\n<\/ul>\n\n\n\n<ol start=\"2\" class=\"wp-block-list\">\n<li>\u53ef\u80fd\u7684\u5c40\u9650\u4e0e\u601d\u8003<\/li>\n<\/ol>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u76ee\u524d\u4e3b\u8981\u7684\u5de5\u5177\u662f\u4ee3\u7801\u89e3\u91ca\u5668\uff0c\u9002\u5408\u6570\u5b66\u3001\u903b\u8f91\u548c\u7a0b\u5e8f\u5f0f\u4efb\u52a1\uff0c\u6269\u5c55\u5230\u66f4\u5f00\u653e\u7684\u771f\u5b9e\u73af\u5883\uff08\u7f51\u9875\u3001API\u3001\u6587\u4ef6\u7cfb\u7edf\u7b49\uff09\u65f6\u4f9d\u7136\u9700\u8981\u5de5\u7a0b\u8bbe\u8ba1\uff1b<\/li>\n\n\n\n<li>\u591a\u8f6e\u91c7\u6837\u3001\u81ea\u4e00\u81f4\u6027\u4f30\u8ba1\u3001\u4ee3\u7801\u6267\u884c\u7b49\u90fd\u5e26\u6765\u5f88\u5927\u7684\u7b97\u529b\u5f00\u9500\uff0c\u81ea\u8fdb\u5316\u8fc7\u7a0b\u672c\u8eab\u201c\u5f88\u8d35\u201d\uff1b<\/li>\n\n\n\n<li>\u4f2a\u6807\u7b7e\u673a\u5236\u518d\u600e\u4e48\u505a\u6b67\u4e49\u8c03\u6574\uff0c\u4ecd\u7136\u6709\u53ef\u80fd\u653e\u5927\u6a21\u578b\u81ea\u8eab\u7684\u504f\u5dee\uff0c\u9700\u8981\u548c\u5b89\u5168\u6027\u3001\u5bf9\u9f50\u673a\u5236\u4e00\u8d77\u8003\u8651\uff1b<\/li>\n\n\n\n<li>\u8bfe\u7a0b\u8d28\u91cf\u9ad8\u5ea6\u4f9d\u8d56\u4e8e\u5956\u52b1\u8bbe\u8ba1\uff1a\u5982\u679c\u4e0d\u786e\u5b9a\u5ea6\u3001\u5de5\u5177\u5956\u52b1\u3001\u591a\u6837\u6027\u7684\u5e73\u8861\u8bbe\u7f6e\u4e0d\u5f53\uff0c\u53ef\u80fd\u4f1a\u5bfc\u81f4\u201c\u8d70\u504f\u201d\u7684\u8bfe\u7a0b\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u603b\u4f53\u6765\u8bf4\uff0c\u8fd9\u7bc7\u8bba\u6587\u5c55\u793a\u4e86\u4e00\u6761\u5f88\u6709\u5438\u5f15\u529b\u7684\u8def\u7ebf\uff1a\u4e0d\u4f9d\u8d56\u5916\u90e8\u6807\u6ce8\u6570\u636e\uff0c\u53ea\u9760\u53cc\u667a\u80fd\u4f53 + \u5de5\u5177 + \u5f3a\u5316\u5b66\u4e60\uff0c\u5c31\u80fd\u8ba9\u4e00\u4e2a\u57fa\u7840\u6a21\u578b\u5728\u63a8\u7406\u80fd\u529b\u4e0a\u6301\u7eed\u201c\u81ea\u6211\u5347\u7ea7\u201d\uff0c\u800c\u4e14\u8fd9\u4e9b\u80fd\u529b\u8fd8\u80fd\u6cdb\u5316\u5230\u66f4\u4e00\u822c\u7684\u63a8\u7406\u4efb\u52a1\u4e0a\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>Agent0 on GitHub: <a href=\"https:\/\/github.com\/aiming-lab\/Agent0\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/github.com\/aiming-lab\/Agent0<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u8bba\u6587Agent0: Unleashing Self-Evolving Agents from Zero Dat [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[8],"tags":[39,94,86],"class_list":["post-6554","post","type-post","status-publish","format-standard","hentry","category-tech","tag-ai","tag-agentic-ai","tag-llm"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/nullthought.net\/?cat=8\" rel=\"category\">Tech<\/a>","rttpg_excerpt":"\u8bba\u6587Agent0: Unleashing Self-Evolving Agents from Zero Dat&hellip;","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6554","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6554"}],"version-history":[{"count":1,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6554\/revisions"}],"predecessor-version":[{"id":6555,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6554\/revisions\/6555"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6554"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6554"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6554"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}