{"id":40026,"date":"2025-04-10T09:02:41","date_gmt":"2025-04-10T01:02:41","guid":{"rendered":"https:\/\/17aitech.com\/?p=40026"},"modified":"2025-04-10T09:02:41","modified_gmt":"2025-04-10T01:02:41","slug":"ai-agent-%e5%8f%91%e5%b1%95%e5%8f%b2%ef%bc%9a%e4%bb%8e-rl-%e9%a9%b1%e5%8a%a8%e5%88%b0%e5%a4%a7%e6%a8%a1%e5%9e%8b%e9%a9%b1%e5%8a%a8-%ef%bd%9cair-2025","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=40026","title":{"rendered":"AI Agent \u53d1\u5c55\u53f2\uff1a\u4ece RL \u9a71\u52a8\u5230\u5927\u6a21\u578b\u9a71\u52a8 \uff5cAIR 2025"},"content":{"rendered":"<p><\/p>\n<p>Manus \u7684\u51fa\u73b0\u5c06\u667a\u80fd\u4f53\u63a8\u5165\u5f53\u4e0b AI \u683c\u5c40\u7684\u524d\u5217\uff0c\u4f7f\u5f97\u8fd9\u4e2a\u8fc7\u53bb\u7565\u62bd\u8c61\u7684\u6982\u5ff5\u53d8\u5f97\u5177\u4f53\u53ef\u611f\u77e5\u3002\u7136\u800c\u884c\u4e1a\u4e2d\u4e5f\u4e0d\u4e4f\u5bf9 Manus \u7684\u4e89\u8bae\uff0c\u8ba4\u4e3a Manus \u6ca1\u6709\u5e95\u5c42\u6280\u672f\u521b\u65b0\u529b\uff0c\u66f4\u591a\u7684\u662f\u5c06\u73b0\u6709\u6280\u672f\u878d\u5408\u4ece\u800c\u5728\u5de5\u7a0b\u4e0a\u521b\u65b0\uff0c\u5373\u6240\u8c13\u7684\u201c\u5957\u58f3\u201d\u3002<\/p>\n<p>\u867d\u8bf4\u5de5\u7a0b\u521b\u65b0\u4e5f\u662f\u4e00\u79cd\u62a4\u57ce\u6cb3\uff0c\u4f46\u201c\u5957\u58f3\u201d\u7684\u8bf4\u6cd5\u4e5f\u5e76\u975e\u5b8c\u5168\u6ca1\u9053\u7406\u3002\u8fd1\u51e0\u5e74\u7684\u65f6\u95f4\u91cc\uff0c\u5b66\u754c\u548c\u4e1a\u754c\u5173\u4e8e Agent \u7684\u6280\u672f\u548c\u5b9e\u8df5\u6210\u679c\u9887\u4e30\u3002\u5728 AI \u667a\u80fd\u4f53\u63a8\u7406\u4e0e\u51b3\u7b56\u7814\u8ba8\u4f1a\uff08AIR 2025\uff09\u4e0a\uff0c\u6765\u81ea\u4f26\u6566\u5927\u5b66\u5b66\u9662\u3001\u65b0\u52a0\u5761\u5357\u6d0b\u7406\u5de5\u5927\u5b66\u3001Weco AI\u3001Google DeepMind\u3001Meta\u3001\u534e\u4e3a\u3001\u963f\u91cc\u7b49\u591a\u4f4d\u5b66\u672f\u754c\u548c\u5de5\u4e1a\u754c\u7684\u7814\u7a76\u4eba\u5458\u56f4\u7ed5\u5f3a\u5316\u5b66\u4e60\u3001\u63a8\u7406\u51b3\u7b56\u3001AI \u667a\u80fd\u4f53\u5c55\u5f00\u8ba8\u8bba\u3002<\/p>\n<p>\u65b0\u52a0\u5761\u5357\u6d0b\u7406\u5de5\u5927\u5b66\u7684\u5b89\u6ce2\u6559\u6388\u63ed\u793a\u4e86\u4ece\u57fa\u4e8e\u5f3a\u5316\u5b66\u4e60\u7684\u667a\u80fd\u4f53\u5230\u7531\u5927\u578b\u8bed\u8a00\u6a21\u578b\u9a71\u52a8\u7684\u667a\u80fd\u4f53\u7684\u6f14\u53d8\uff0c\u5206\u4eab\u4e86\u56e2\u961f\u591a\u9879\u5173\u4e8e Agent \u7684\u5de5\u4f5c\u8fdb\u5c55\uff0c\u5176\u4e2d Q* \u7b97\u6cd5\u4ee5\u591a\u6b65\u9aa4\u63a8\u7406\u4f5c\u4e3a\u5ba1\u614e\u89c4\u5212\uff0c\u5728\u5b66\u4e60 Q \u503c\u6a21\u578b\u7684\u8fc7\u7a0b\u4e2d\uff0c\u9700\u8981\u7ecf\u5386\u79bb\u7ebf\u5f3a\u5316\u5b66\u4e60\u4ee5\u4ea4\u66ff\u66f4\u65b0 Q \u503c\u6807\u7b7e\u5e76\u62df\u5408 QVM\u3001\u4f7f\u7528\u8868\u73b0\u6700\u4f73\u7684\u56de\u6eda\u8f68\u8ff9\u7684\u5956\u52b1\u3001\u4f7f\u7528\u4e0e\u66f4\u5f3a\u5927 LLM \u4e00\u8d77\u5b8c\u6210\u7684\u8f68\u8ff9\u7684\u5956\u52b1\u4e09\u4e2a\u5173\u952e\u6b65\u9aa4\u3002<\/p>\n<p>\u521d\u521b\u516c\u53f8 Weco Al \u7684 CTO Yuxiang \u9610\u8ff0\u4e86\u5728\u89e3\u7a7a\u95f4\u4e2d\u5bfb\u627e\u667a\u80fd\u7684\u4e00\u4e9b\u65f6\u95f4\uff0c\u4ecb\u7ecd\u4e86\u7531\u4eba\u5de5\u667a\u80fd\u9a71\u52a8\u7684 Agent\u2014\u2014 AIDE\uff0c\u80fd\u591f\u5904\u7406\u5b8c\u6574\u7684\u673a\u5668\u548c\u5de5\u7a0b\u4efb\u52a1\u3002\u5982\u679c\u5c06\u673a\u5668\u5b66\u4e60\u548c\u5de5\u7a0b\u89c6\u4e3a\u4e00\u4e2a\u4ee3\u7801\u4f18\u5316\u95ee\u9898\uff0c\u90a3\u4e48\u5b83\u5c31\u4f1a\u5c06\u6574\u4e2a\u641c\u7d22\u6216\u4ee3\u7801\u4f18\u5316\u7684\u8fc7\u7a0b\u5f62\u5f0f\u5316\u4e3a\u5728\u89e3\u7a7a\u95f4\u4e2d\u7684\u6811\u641c\u7d22\u3002\u5728\u8fd9\u4e2a\u88ab\u5f62\u5f0f\u5316\u7684\u89e3\u7a7a\u95f4\u4e2d\uff0cAIDE \u662f\u4e00\u4e2a\u4efb\u4f55\u5927\u8bed\u8a00\u6a21\u578b\u90fd\u53ef\u4ee5\u7f16\u5199\u7684\u4ee3\u7801\u7a7a\u95f4\u3002<\/p>\n<p>\u6765\u81ea\u4f26\u6566\u5927\u5b66\u5b66\u9662\u7684\u5b8b\u7814\u4ece\u00a0DeepSeek\u00a0\u5207\u5165\uff0c\u8ba8\u8bba\u4e86\u5f3a\u5316\u5b66\u4e60\u5728\u5927\u578b\u8bed\u8a00\u6a21\u578b\u63a8\u7406\u4e2d\u7684\u4f5c\u7528\uff0c\u5e76\u6307\u51fa DS \u53c8\u4e00\u4e2a\u201cAha\u65f6\u523b\u201d\uff0c\u5373\u5728\u5f3a\u5316\u5b66\u4e60\u9636\u6bb5\uff0c\u5927\u578b\u8bed\u8a00\u6a21\u578b\u5b66\u4f1a\u4e86\u81ea\u6211\u7ea0\u6b63\uff0c\u8fd9\u53ef\u80fd\u662f\u7531\u4e8e\u5176\u57fa\u7840\u6a21\u578b\u5df2\u7ecf\u5177\u5907\u81ea\u6211\u7ea0\u6b63\u7684\u80fd\u529b\u3002\u57fa\u4e8e\u6b64\u8fdb\u4e00\u6b65\u53d1\u73b0\uff0c\u5f53 Agent \u4f7f\u7528\u67d0\u4e9b\u5173\u952e\u8bcd\u65f6\uff0c\u5b83\u4eec\u4f1a\u8fdb\u884c\u5404\u79cd\u56de\u6eaf\u3001\u81ea\u6211\u62a5\u544a\u548c\u590d\u6742\u63a8\u7406\u3002<\/p>\n<p>\u8c37\u6b4c\u00a0Deepmind\u00a0\u7814\u7a76\u5458\u51af\u7199\u680b\u521d\u6b65\u9610\u8ff0\u4e86\u5c06\u5f3a\u5316\u5b66\u4e60\u7684\u7ec4\u6210\u90e8\u5206\u7528\u81ea\u7136\u8bed\u8a00\u63cf\u8ff0\u51fa\u6765\u7684\u7406\u5ff5\uff0c\u5c06\u4f1a\u628a\u6240\u6709\u5f3a\u5316\u5b66\u4e60\u7684\u6982\u5ff5\u91cd\u65b0\u5b9a\u4e49\u4e3a\u81ea\u7136\u8bed\u8a00\u8868\u793a\u7684\u5185\u5bb9\uff0c\u5c1d\u8bd5\u5c06\u7b56\u7565\u3001\u503c\u51fd\u6570\u3001\u8d1d\u5c14\u66fc\u65b9\u7a0b\u3001\u8499\u7279\u5361\u6d1b\u91c7\u6837\u3001\u65f6\u95f4\u5dee\u5206\u5b66\u4e60\u4ee5\u53ca\u7b56\u7565\u6539\u8fdb\u64cd\u4f5c\u7b26\u7b49\uff0c\u6620\u5c04\u5230\u5b83\u4eec\u7684\u81ea\u7136\u8bed\u8a00\u5bf9\u5e94\u4e2d\u3002<\/p>\n<p>AIR2025 \u7531\u4f26\u6566\u5927\u5b66\u5b66\u9662\u6c6a\u519b\u3001Meta GenAI \u7530\u6e0a\u680b\u7b49\u6559\u6388\u8054\u5408\u4e3b\u529e\uff0c\u81f4\u529b\u4e8e\u63a8\u52a8\u667a\u80fd\u7cfb\u7edf\u7684\u53d1\u5c55\uff0c\u4f7f\u5176\u80fd\u591f\u81ea\u4e3b\u3001\u9002\u5e94\u6027\u5f3a\u4e14\u8d1f\u8d23\u4efb\u5730\u8fd0\u884c\uff08\u4f1a\u8bae\u8be6\u60c5\u53ca\u6ce8\u518c\u53ef\u8bbf\u95ee\u5b98\u7f51\uff1ahttps:\/\/ai-agent-reasoning.com\uff09\u3002\u672c\u6b21\u4f1a\u8bae\u7279\u522b\u9e23\u8c22\u6765\u81ea\u52a0\u5dde\u5927\u5b66\u4f2f\u514b\u5229\u5206\u6821\u7684\u535a\u58eb\u540e\u7814\u7a76\u5458\u987e\u5c1a\u5b9a\u3002<\/p>\n<p>AI \u79d1\u6280\u8bc4\u8bba\u622a\u53d6\u4f1a\u8bae\u90e8\u5206\u7cbe\u5f69\u5185\u8fdb\u884c\u7f16\u8bd1\uff0c\u4ee5\u4e0b\u4e3a\u6838\u5fc3\u5185\u5bb9\u7684\u6f14\u8bb2\u5b9e\u5f55\uff1a<\/p>\n<p><strong>\u4e00\u3001Agent \u9a71\u52a8\u529b\u53d8\u9769\uff1a\u4ece RL \u5230 LLM<\/strong><\/p>\n<\/p>\n<p>\u65b0\u52a0\u5761\u5357\u6d0b\u7406\u5de5\u5927\u5b66\u7684\u5b89\u6ce2\u6559\u6388\u505a\u4e86\u4e3b\u9898\u4e3a\u300aFrom RL-based to LLM-powered Agents\u300b\u7684\u6f14\u8bb2\uff0c\u63ed\u793a\u4e86\u8fd1\u5e74\u6765\u4ece\u57fa\u4e8e\u5f3a\u5316\u5b66\u4e60\u7684\u667a\u80fd\u4f53\u5230\u7531\u5927\u578b\u8bed\u8a00\u6a21\u578b\u9a71\u52a8\u7684\u667a\u80fd\u4f53\u7684\u6f14\u53d8\uff0c\u5206\u4eab\u4e86\u591a\u9879\u5173\u4e8e Agent \u7684\u5de5\u4f5c\u8fdb\u5c55\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-eca3219bff8c15c37c232ffd7a987cc3.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-eca3219bff8c15c37c232ffd7a987cc3.png\"><\/a><\/p>\n<p>\u53bb\u5e74\uff0c\u6211\u4eec\u505a\u4e86\u4e00\u4e9b\u5de5\u4f5c\uff0c\u7ed3\u5408\u4e86\u4e00\u4e2a\u4e34\u65f6\u6a21\u578b\uff0c\u4ee5\u63d0\u9ad8\u5176\u5728\u67d0\u4e9b\u57fa\u51c6\u95ee\u9898\u4e2d\u7684\u6027\u80fd\u3002\u6211\u4eec\u7684\u65b9\u6cd5\u662f\u5c1d\u8bd5\u4ece\u4e0e\u73af\u5883\u7684\u4ea4\u4e92\u4e2d\u5b66\u4e60\u7b56\u7565\uff0c\u56e0\u6b64\u5b83\u5177\u6709\u5f88\u5f3a\u7684\u843d\u5730\u80fd\u529b\uff0c\u6240\u4ee5\u6211\u60f3\u6211\u4eec\u5728\u8fd9\u91cc\u5c1d\u8bd5\u7ed3\u5408\u5148\u9a8c\u77e5\u8bc6\u7684\u4f18\u52bf\uff0c\u4ece\u6a21\u578b\u548c\u843d\u5730\u80fd\u529b\u4e2d\u6c72\u53d6\u4f18\u52bf\uff0c\u4ee5\u63d0\u9ad8\u6027\u80fd\u3002<\/p>\n<p>\u56e0\u4e3a\u5bf9\u4e8e\u8fd9\u9879\u5de5\u4f5c\uff0c\u6211\u4eec\u53d1\u73b0\u5229\u7528\u77e5\u8bc6\u6a21\u578b\u53ef\u4ee5\u63d0\u9ad8\u5176\u5728\u67d0\u4e9b\u5b9e\u9645\u5de5\u4f5c\u573a\u666f\u4e2d\u7684\u6027\u80fd\u3002<\/p>\n<p>\u63a8\u7406\u548c\u63a8\u65ad\u975e\u5e38\u91cd\u8981\uff0c\u5c24\u5176\u662f\u5728 OpenAI-o1 \u548c DeepSeek R1 \u53d1\u5e03\u4e4b\u540e\uff0c\u6211\u4eec\u6709\u4e00\u4e2a\u7eaf\u7cb9\u57fa\u4e8e\u81ea\u5df1\u7814\u7a76\u7684\u7248\u672c\uff0c\u8fd9\u786e\u5b9e\u975e\u5e38\u56f0\u96be\u3002<\/p>\n<p>\u4f46\u4e8b\u5b9e\u4e0a\uff0c\u6211\u4eec\u5728 OpenAI \u53d1\u5e03\u76f8\u5173\u6a21\u578b\u4e4b\u524d\u5c31\u53d1\u5e03\u4e86\u5173\u4e8e Q* \u7684\u7b2c\u4e00\u7bc7\u8bba\u6587\u3002\u6211\u4eec\u9700\u8981\u4e00\u4e2a G \u51fd\u6570\uff0c\u7528\u4e8e\u4f30\u7b97\u4ece\u521d\u59cb\u72b6\u6001\u5230\u5f53\u524d\u8282\u70b9\u7684\u6210\u672c\u3002\u5728\u6211\u4eec\u7684\u5de5\u4f5c\u4e2d\uff0c\u6211\u4eec\u4f7f\u7528\u7684 G \u51fd\u6570\u662f\u901a\u8fc7\u5229\u7528\u6587\u732e\u4e2d\u7684\u6570\u636e\u6765\u8bad\u7ec3\u6a21\u578b\u7684\u3002\u5bf9\u4e8e\u542f\u53d1\u5f0f\u51fd\u6570\uff08h \u503c\uff09\uff0c\u6211\u4eec\u5b9e\u9645\u4e0a\u662f\u81ea\u5df1\u8fdb\u884c\u4e86\u4fee\u6b63\u3002<\/p>\n<p>\u6240\u4ee5\uff0c\u57fa\u4e8e\u6211\u4eec\u7684\u6570\u636e\uff0c\u8bad\u7ec3\u8fd9\u6837\u4e00\u4e2a\u5f3a\u5927\u7684\u6a21\u578b\u6709\u5f88\u591a\u65b9\u6cd5\u3002\u6700\u7ec8\uff0c\u6211\u4eec\u5c06\u8fd9\u4e24\u8005\u7ed3\u5408\u8d77\u6765\uff0c\u5e76\u5e94\u7528 A* \u641c\u7d22\u7b97\u6cd5\uff0c\u4ee5\u63d0\u5347\u5927\u578b\u8bed\u8a00\u6a21\u578b\u7684\u63a8\u7406\u80fd\u529b\u3002<\/p>\n<p>\u6240\u4ee5\uff0c\u6211\u4eec\u65e9\u671f\u505a\u4e86\u4e9b\u5b9e\u9a8c\u3002\u4f60\u53ef\u4ee5\u964d\u4f4e\u90a3\u4e9b\u6570\u503c\uff0c\u56e0\u4e3a\u90a3\u65f6\u5019\u57fa\u7840\u6a21\u578b\u8fd8\u4e0d\u591f\u5f3a\u5927\u3002\u6211\u60f3\u5173\u952e\u70b9\u662f\uff0c\u5982\u679c\u4f60\u5e94\u7528\u8fd9\u79cd\u63a8\u7406\u65b9\u6cd5\uff0c\u5b83\u53ef\u4ee5\u63d0\u5347\u57fa\u7840\u6a21\u578b\u7684\u6027\u80fd\u3002<\/p>\n<p>\u7136\u540e\u6211\u4eec\u4ee5\u67d0\u79cd\u65b9\u5f0f\u8bad\u7ec3\u5b83\u4eec\u7684 Q \u503c\u51fd\u6570\u3002\u6240\u4ee5\uff0c\u6211\u4eec\u8fd8\u5728\u8003\u8651\u662f\u5426\u80fd\u591f\u514b\u670d\u56f0\u96be\uff0c\u4f8b\u5982\uff0c\u5c06\u8fd9\u79cd\u65b9\u6cd5\u5e94\u7528\u4e8e\u6539\u8fdb\u6700\u8fd1\u7684 DeepSeek \u6a21\u578b\u4ee5\u53ca\u5176\u4ed6\u6a21\u578b\u3002<\/p>\n<p>\u6240\u4ee5\uff0c\u6211\u4eec\u5728\u6bd4\u8f83\u63a7\u5236\u65b9\u9762\u4e5f\u505a\u4e86\u4e00\u4e9b\u5173\u4e8e Synapse \u7684\u5de5\u4f5c\uff0c\u662f\u6211\u4eec\u53bb\u5e74\u5e74\u521d\u53d1\u8868\u7684\u6210\u679c\u4e4b\u4e00\u3002\u8fd9\u4e9b\u5de5\u4f5c\u6d89\u53ca\u4e00\u4e9b\u60f3\u6cd5\uff0c\u6bd4\u5982\u72b6\u6001\u62bd\u8c61\u8bad\u7ec3\u3001\u4ece\u6f14\u793a\u4e2d\u5b66\u4e60\uff0c\u4ee5\u53ca\u4f7f\u7528\u8bb0\u5fc6\u7b49\u65b9\u6cd5\uff0c\u6765\u6539\u8fdb\u8ba1\u7b97\u673a\u63a7\u5236\u4efb\u52a1\u3002<\/p>\n<p>\u6211\u4eec\u8fd8\u63d0\u4f9b\u4e86\u4e00\u4e2a\u7528\u4e8e\u6784\u5efa\u901a\u7528\u865a\u62dfAgent\u7684\u5f00\u53d1\u8005\u5de5\u5177\u5305\u3002\u6211\u4eec\u63d0\u4f9b\u4e86\u66f4\u597d\u7684\u754c\u9762\uff0c\u8fd8\u63d0\u4f9b\u4e86\u66f4\u5f3a\u5927\u7684\u524d\u7aef\u652f\u6301\uff0c\u5e76\u4e14\u63d0\u4f9b\u4e86\u5927\u91cf\u7684\u57fa\u51c6\u6d4b\u8bd5\uff0c\u7528\u4e8e\u8bbe\u8ba1\u9002\u7528\u4e8ePC\u63a7\u5236\u3001\u8ba1\u7b97\u673a\u63a7\u5236\u7b49\u573a\u666f\u7684\u901a\u7528\u865a\u62dfAgent\uff0c\u4e5f\u9002\u7528\u4e8e\u79fb\u52a8\u8bbe\u5907\u7684\u63a7\u5236\u3002<\/p>\n<p>\u63a5\u4e0b\u6765\u7684\u8fd9\u9879\u5de5\u4f5c\u6d89\u53ca\u5229\u7528\u8bed\u8a00\u6a21\u578b\u9a71\u52a8\u7684\u667a\u80fd\u4f53\u6765\u73a9\u5177\u6709\u6311\u6218\u6027\u7684\u7535\u5b50\u6e38\u620f\u3002<\/p>\n<p>\u56e0\u6b64\uff0c\u6211\u4eec\u6784\u5efa\u4e86\u4e00\u4e2a\u667a\u80fd\u4f53\u67b6\u6784\uff0c\u5305\u62ec\u4e0d\u540c\u7684\u7ec4\u4ef6\uff0c\u4f8b\u5982\u6211\u4eec\u9700\u8981\u7406\u89e3\u73af\u5883\u3002\u5b83\u5305\u542b\u4e00\u4e2a\u53cd\u601d\u6a21\u578b\u3001\u8bb0\u5fc6\u6a21\u578b\u3001\u68c0\u7d22\u6a21\u578b\u7b49\uff0c\u7528\u4e8e\u5e94\u5bf9\u8bb8\u591a\u5177\u6709\u6311\u6218\u6027\u7684\u7535\u5b50\u6e38\u620f\u548c\u4e0d\u540c\u7684\u8f6f\u4ef6\u3002\u8fd9\u4e2a\u9879\u76ee\u662f\u5f00\u6e90\u7684\uff0c\u5bf9\u4e8e\u611f\u5174\u8da3\u7684\u4eba\u975e\u5e38\u6709\u5438\u5f15\u529b\u3002<\/p>\n<p>\u6211\u4eec\u6700\u8fd1\u505a\u4e86\u4e00\u4e9b\u5c1a\u672a\u53d1\u8868\u7684\u5de5\u4f5c\uff0c\u662f\u5173\u4e8e\u4f7f\u7528\u5f3a\u5316\u5b66\u4e60\uff08RL\uff09\u5bf9\u8bed\u8a00\u6a21\u578b\u8fdb\u884c\u5fae\u8c03\u7684\u3002<\/p>\n<p>\u6211\u60f3\u8fd9\u5728\u67d0\u79cd\u7a0b\u5ea6\u4e0a\u4e0e\u4e00\u4e9b\u65e9\u671f\u5de5\u4f5c\u6709\u6240\u4e0d\u540c\uff0c\u5728\u6211\u4eec\u8fc7\u53bb\u770b\u5230\u7684\u5927\u591a\u6570\u5de5\u4f5c\u4e2d\uff0c\u5f3a\u5316\u5b66\u4e60\u5e76\u6ca1\u6709\u6d89\u53ca\u667a\u80fd\u4f53\u3002\u4f60\u77e5\u9053\u7684\uff0c\u4eba\u4eec\u53ea\u662f\u6784\u5efa\u4e0d\u540c\u7684\u7ec4\u4ef6\uff0c\u4f7f\u7528\u8bed\u8a00\u6a21\u578b\u4f5c\u4e3a\u5927\u8111\uff0c\u5e76\u7ed3\u5408\u5176\u4ed6\u7ec4\u4ef6\u6765\u5904\u7406\u590d\u6742\u4efb\u52a1\u3002<\/p>\n<p>\u4f46\u5728\u8fd9\u91cc\uff0c\u6211\u8ba4\u4e3a\u5728\u672a\u6765\uff0c\u5bf9\u4e8e\u8bb8\u591a\u73b0\u5b9e\u4e16\u754c\u7684\u95ee\u9898\uff0c\u6211\u4eec\u9700\u8981\u5f3a\u5316\u5b66\u4e60\u7684\u80fd\u529b\u3002\u7136\u800c\uff0c\u5982\u679c\u6211\u4eec\u60f3\u5c06\u5f3a\u5316\u5b66\u4e60\u5e94\u7528\u4e8e\u8fd9\u4e9b\u573a\u666f\uff0c\u4f1a\u9762\u4e34\u8bb8\u591a\u6311\u6218\uff0c\u5176\u4e2d\u6700\u663e\u8457\u7684\u662f\u63a2\u7d22\u7a7a\u95f4\u7684\u6307\u6570\u7ea7\u589e\u957f\u3002\u56e0\u4e3a\u5f00\u653e\u548c\u5b9e\u9645\u52a8\u4f5c\u6280\u80fd\u7684\u91c7\u6837\u7a7a\u95f4\u4f1a\u968f\u7740\u77e9\u5f62\u7684\u5927\u5c0f\u548c\u539a\u5ea6\u5448\u6307\u6570\u7ea7\u589e\u957f\uff0c\u56e0\u4e3a\u63a2\u7d22\u7a7a\u95f4\u662f\u5728token\u7ea7\u522b\u4e0a\u7684\uff0c\u6240\u4ee5token\u7a7a\u95f4\u975e\u5e38\u5e9e\u5927\u3002\u56e0\u6b64\uff0c\u6211\u4eec\u9700\u8981\u89e3\u51b3\u63a2\u7d22\u95ee\u9898\u3002\u540c\u65f6\uff0c\u6211\u4eec\u6ce8\u610f\u5230\u5e76\u975e\u6240\u6709token\u5728\u6700\u7ec8\u51b3\u7b56\u52a8\u4f5c\u4e2d\u90fd\u53d1\u6325\u6709\u610f\u4e49\u7684\u4f5c\u7528\u3002<\/p>\n<p>\u6240\u4ee5\uff0c\u6211\u8ba4\u4e3a\u6211\u4eec\u5728\u8fd9\u91cc\u5f97\u5230\u7684\u542f\u793a\u662f\uff0c\u6211\u4eec\u5fc5\u987b\u8bbe\u8ba1\u4e00\u4e9b\u673a\u5236\u6765\u51b3\u5b9a\u5982\u4f55\u8fdb\u884c\u66f4\u6709\u6548\u7684\u63a2\u7d22\uff0c\u4ee5\u4fbf\u63d0\u9ad8\u5f3a\u5316\u5b66\u4e60\u5fae\u8c03\u7684\u6548\u7387\uff0c\u4ece\u800c\u63d0\u5347\u8bed\u8a00\u6a21\u578b\u7684\u6027\u80fd\u3002\u56e0\u6b64\uff0c\u6211\u4eec\u8bbe\u8ba1\u4e86\u4e00\u4e2a\u540d\u4e3a\u201cCoSo\u201d\u7684\u65b9\u6cd5\uff0c\u5b83\u5305\u542b\u51e0\u4e2a\u5173\u952e\u601d\u60f3\u3002\u9996\u5148\uff0c\u6211\u4eec\u4f7f\u7528\u4e8b\u5b9e\u63a8\u7406\u6765\u8bc6\u522b\u5bf9\u52a8\u4f5c\u81f3\u5173\u91cd\u8981\u7684token\u3002<\/p>\n<p>\u4e0d\u662f\u6bcf\u4e2atoken\u90fd\u5bf9\u667a\u80fd\u4f53\u6700\u7ec8\u91c7\u53d6\u7684\u52a8\u4f5c\u4ea7\u751f\u5f71\u54cd\uff0c\u6216\u8005\u4ea7\u751f\u76f8\u540c\u7684\u5f71\u54cd\u3002\u56e0\u6b64\uff0c\u6211\u4eec\u4f7f\u7528\u56e0\u679c\u63a8\u7406\u6765\u627e\u51fa\u8fd9\u4e9btoken\uff0c\u7136\u540e\u5229\u7528\u8fd9\u4e9b\u4fe1\u606f\u6765\u51b3\u5b9a\u5982\u4f55\u8fdb\u884c\u63a2\u7d22\u3002\u5176\u6b21\uff0c\u6211\u4eec\u53ef\u4ee5\u4e2d\u65ad\u4f18\u5316\u8fc7\u7a0b\uff0c\u5c06\u6211\u4eec\u7684\u63a2\u7d22\u96c6\u4e2d\u5728\u90a3\u4e9b\u6709\u5f71\u54cd\u7684token\u4e0a\u3002<\/p>\n<p>\u8fd9\u662f\u5229\u7528\u6211\u4eec\u5728\u7b2c\u4e00\u6b65\u4e2d\u5b66\u5230\u7684\u7ed3\u679c\u3002\u7136\u540e\u6211\u4eec\u5c1d\u8bd5\u8fdb\u884c\u4e86\u8bb8\u591a\u5b9e\u9a8c\uff0c\u53ef\u4ee5\u770b\u5230\u8fd9\u79cd\u65b9\u6cd5\u663e\u8457\u63d0\u9ad8\u4e86\u89c6\u89c9\u8bed\u8a00\u6a21\u578b\uff08VLM\uff09\u5728\u4e00\u4e9b\u975e\u5e38\u5177\u6709\u6311\u6218\u6027\u7684\u4efb\u52a1\u4e2d\u7684\u6027\u80fd\u3002\u6211\u8ba4\u4e3a\u8fd9\u8fd8\u662f\u4e00\u9879\u6b63\u5728\u8fdb\u884c\u7684\u5de5\u4f5c\uff0c\u4f8b\u5982\u6211\u521a\u624d\u63d0\u5230\u7684\u521b\u9020\u6027\u5de5\u4f5c\u3002<\/p>\n<\/p>\n<p><strong>\u4e8c\u3001\u5728\u89e3\u7a7a\u95f4\u4e2d\u5bfb\u627e\u667a\u80fd<\/strong><\/p>\n<\/p>\n<p>\u521d\u521b\u516c\u53f8 Weco Al \u7684 CTO Yuxiang \u505a\u4e86\u9898\u4e3a\u300aAlDE: Searching Intelligence in the Space of Solutions\u300b\u7684\u5206\u4eab\uff0c\u9610\u8ff0\u5728\u89e3\u7a7a\u95f4\u4e2d\u5bfb\u627e\u667a\u80fd\u7684\u65b0\u601d\u8003\uff0c\u4ecb\u7ecd\u4e86\u4e00\u79cd\u7531\u4eba\u5de5\u667a\u80fd\u9a71\u52a8\u7684\u5f3a\u5927\u7684 Agent\u2014\u2014 AIDE\u3002<\/p>\n<p>\u6211\u4eec\u4e4b\u6240\u4ee5\u79f0\u4e4b\u4e3a AIDE \u662f\u56e0\u4e3a\uff0c\u5b83\u5c31\u50cf\u4e00\u79cd\u7531\u4eba\u5de5\u667a\u80fd\u9a71\u52a8\u7684\u5f3a\u5927\u7684 Agent\uff0c\u80fd\u591f\u5904\u7406\u5b8c\u6574\u7684\u673a\u5668\u548c\u5de5\u7a0b\u4efb\u52a1\u3002\u6240\u4ee5\uff0c\u5982\u679c\u5c06\u673a\u5668\u5b66\u4e60\u548c\u5de5\u7a0b\u89c6\u4e3a\u4e00\u4e2a\u4ee3\u7801\u4f18\u5316\u95ee\u9898\uff0c\u90a3\u4e48\u5b83\u5c31\u4f1a\u5c06\u6574\u4e2a\u641c\u7d22\u6216\u4ee3\u7801\u4f18\u5316\u7684\u8fc7\u7a0b\u5f62\u5f0f\u5316\u4e3a\u5728\u89e3\u7a7a\u95f4\u4e2d\u7684\u6811\u641c\u7d22\u3002\u5728\u8fd9\u4e2a\u88ab\u5f62\u5f0f\u5316\u7684\u89e3\u7a7a\u95f4\u4e2d\uff0c\u5b83\u53ea\u662f\u4e00\u4e2a\u4efb\u4f55\u5927\u8bed\u8a00\u6a21\u578b\u90fd\u53ef\u4ee5\u7f16\u5199\u7684\u4ee3\u7801\u7a7a\u95f4\u3002<\/p>\n<p>\u4f60\u53ef\u80fd\u89c1\u8fc7\u5176\u4ed6\u66f4\u5177\u4f53\u7684Agent\uff0c\u6bd4\u5982\u90a3\u4e9b\u63d0\u793a APIAgent \u6216\u53cd\u5e94\u5f0f Agent\uff0c\u5b83\u4eec\u5c06\u6240\u6709\u5386\u53f2\u89e3\u51b3\u65b9\u6848\u7ec4\u7ec7\u6210\u6811\u72b6\u7ed3\u6784\u3002\u7136\u540e\uff0c\u5c06\u6240\u6709\u8fd9\u4e9b\u5386\u53f2\u89e3\u51b3\u65b9\u6848\u7eb3\u5165\u4e0a\u4e0b\u6587\u4e2d\uff0c\u4f46\u8fd9\u4e2a\u8fc7\u7a0b\u5b9e\u9645\u4e0a\u662f\u9012\u589e\u7684\u3002\u56e0\u6b64\uff0c\u5b83\u4f1a\u8fc5\u901f\u79ef\u7d2f\u4e0a\u4e0b\u6587\u4fe1\u606f\uff0c\u6240\u4ee5\u5728\u957f\u671f\u7684\u4ee3\u7801\u4f18\u5316\u8fc7\u7a0b\u4e2d\uff0c\u5b83\u53ef\u80fd\u4e0d\u4f1a\u8868\u73b0\u5f97\u5f88\u597d\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-f5dbf6b2663f0faae50ebafc19c90e00.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-f5dbf6b2663f0faae50ebafc19c90e00.png\"><\/a><\/p>\n<p>\u6240\u4ee5\uff0c\u8fd9\u4e2a\u95ee\u9898\u88ab\u91cd\u65b0\u5b9a\u4e49\u4e3a\u4e00\u4e2a\u4f18\u5316\u95ee\u9898\u3002\u673a\u5668\u5b66\u4e60\u53ef\u4ee5\u5728\u673a\u5668\u4e2d\u5b8c\u6210\uff0c\u7136\u540e\u6211\u4eec\u53ef\u4ee5\u5b9a\u4e49\u6240\u6709\u76f8\u5173\u7684\u8bc4\u4f30\u6307\u6807\u3002\u8fd9\u4e0e\u6211\u4eec\u63d0\u51fa\u7684\u673a\u5668\u5b66\u4e60\u5de5\u7a0b Agent \u975e\u5e38\u5951\u5408\uff0c\u5176\u5b9a\u4e49\u7684\u5956\u52b1\u6216\u4f18\u5316\u76ee\u6807\u4e5f\u975e\u5e38\u7b80\u5355\u3002\u6211\u4eec\u53ea\u662f\u5728\u8fd9\u4e2a\u4ee3\u7801\u7a7a\u95f4\u4e2d\u8fdb\u884c\u641c\u7d22\uff0c\u76ee\u6807\u662f\u4f18\u5316\u673a\u5668\u5b66\u4e60\u4ee3\u7801\u548c\u673a\u5668\u5b66\u4e60\u5de5\u7a0b\u4efb\u52a1\u4e2d\u7684\u76ee\u6807\u51fd\u6570\u3002\u8fd9\u4e2a\u76ee\u6807\u51fd\u6570\u53ef\u4ee5\u662f\u9a8c\u8bc1\u7cbe\u5ea6\u3001\u635f\u5931\uff0c\u6216\u8005\u662f\u4efb\u4f55\u4e0e\u4f60\u7684\u673a\u5668\u5b66\u4e60\u6210\u672c\u76f8\u5173\u7684\u6307\u6807\u3002<\/p>\n<p>\u800c\u4ee3\u7801\u7a7a\u95f4\u5728\u672c\u4f8b\u4e2d\u88ab\u975e\u5e38\u5177\u4f53\u5730\u5b9a\u4e49\u4e3a\u7528\u4e8e\u89e3\u51b3\u8be5\u95ee\u9898\u7684Python\u811a\u672c\u7a7a\u95f4\u3002\u597d\u5904\u662f\u6211\u4eec\u73b0\u5728\u53ef\u4ee5\u5728\u4e00\u4e2a\u516c\u5e73\u7684\u6307\u6807\u4e0a\u6bd4\u8f83\u89e3\u51b3\u65b9\u6848\uff0c\u5e76\u4e14\u4f7f\u8fd9\u4e9b\u4f9d\u8d56\u4e8e\u5355\u4e00\u6807\u51c6\u5df2\u77e5\u8bc4\u4f30\u7684\u7814\u7a76\u65b9\u6cd5\u66f4\u52a0\u7edf\u4e00\uff0c\u6574\u4e2a\u641c\u7d22\u8fc7\u7a0b\u4e5f\u4f1a\u66f4\u52a0\u7a33\u5065\u3002<\/p>\n<p>\u56e0\u6b64\uff0c\u6211\u4eec\u5f00\u53d1\u4e86\u8fd9\u79cd\u7b97\u6cd5\uff0c\u5b83\u672c\u8d28\u4e0a\u662f\u4e00\u4e2a\u6811\u641c\u7d22\u95ee\u9898\u3002\u4f60\u4ece\u4e00\u68f5\u7a7a\u6811\u5f00\u59cb\uff0c\u9996\u5148\u4f1a\u751f\u6210\u4e00\u4e2a\u521d\u59cb\u8282\u70b9\uff0c\u5b9e\u9645\u4e0a\u662f\u4e00\u7ec4\u57fa\u7840\u89e3\u51b3\u65b9\u6848\u3002\u7136\u540e\uff0c\u5b83\u901a\u8fc7\u67e5\u770b\u73b0\u6709\u7684\u4ee3\u7801\u548c\u73b0\u6709\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u8fed\u4ee3\u5730\u63d0\u51fa\u65b0\u7684\u89e3\u51b3\u65b9\u6848\u3002\u8fd9\u4e9b\u89e3\u51b3\u65b9\u6848\u5df2\u7ecf\u751f\u6210\u4e86\uff0c\u7136\u540e\u5b83\u4f1a\u63d0\u51fa\u4f60\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u5e76\u4e14\u57fa\u4e8e\u8fd9\u4e2a\u60f3\u6cd5\uff0c\u5b83\u4f1a\u751f\u6210\u90a3\u6bb5\u4ee3\u7801\uff0c\u7136\u540e\u8fd0\u884c\u4ee3\u7801\u4ee5\u8bc4\u4f30\u89e3\u51b3\u65b9\u6848\uff0c\u5e76\u8bb0\u5f55\u65b0\u7684\u8282\u70b9\u3002<\/p>\n<p>\u8fd9\u91cc\u7684\u8bc4\u4f30\u6307\u6807\u662f\u6eda\u52a8\uff08scroll\uff09\uff0c\u901a\u5e38\u5728\u673a\u5668\u5b66\u4e60\u4efb\u52a1\u4e2d\uff0c\u8fd9\u4e2a\u6307\u6807\u53ef\u4ee5\u662f\u7cbe\u5ea6\uff08accuracy\uff09\u3001\u635f\u5931\uff08loss\uff09\u6216\u8005\u968f\u4fbf\u4f60\u600e\u4e48\u79f0\u547c\u5b83\u3002\u7136\u540e\u5b83\u4f1a\u6839\u636e\u8fd9\u4e2a\u6307\u6807\u9009\u62e9\u4e0b\u4e00\u4e2a\u8282\u70b9\uff0c\u4ee5\u4fbf\u8fdb\u4e00\u6b65\u4f18\u5316\u3002\u6240\u4ee5\uff0c\u5b83\u6d89\u53ca\u4e86\u6240\u6709\u8fd9\u4e9b\u641c\u7d22\u7b56\u7565\u3001\u603b\u7ed3\u64cd\u4f5c\u7b26\u4ee5\u53ca\u7f16\u7801\u64cd\u4f5c\u7b26\u3002\u8fd9\u4e9b\u64cd\u4f5c\u7b26\u4e0d\u518d\u5b8c\u5168\u7531\u7b97\u6cd5\u5b9a\u4e49\uff0c\u800c\u662f\u90e8\u5206\u7531\u5927\u578b\u8bed\u8a00\u6a21\u578b\u5b9a\u4e49\u3002<\/p>\n<p>\u6240\u4ee5\uff0c\u4e3a\u4e86\u66f4\u76f4\u89c2\u5730\u5c55\u793a\uff0c\u6211\u4eec\u4ece S0 \u5f00\u59cb\uff0c\u8fd9\u662f\u4e00\u4e2a\u521d\u59cb\u7684\u7a7a\u89e3\u51b3\u65b9\u6848\uff0c\u4e5f\u5c31\u662f\u6211\u4eec\u7684\u6570\u636e\u72b6\u6001\u3002\u6211\u4eec\u8fd8\u6ca1\u6709\u4efb\u4f55\u73b0\u6709\u7684\u673a\u5668\u5b66\u4e60\u4efb\u52a1\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u7136\u540e\u5b83\u5f00\u59cb\u8d77\u8349\u4e09\u4e2a\u3002\u4f8b\u5982\uff0c\u8d77\u8349\u4e09\u4e2a\u65b9\u5411\u4e0d\u540c\u7684\u89e3\u51b3\u65b9\u6848\u3002\u6240\u4ee5\u5728\u63d0\u793a\u4e2d\uff0c\u6709\u4e00\u4e2a\u6280\u5de7\u662f\u6211\u4eec\u4f1a\u660e\u786e\u8981\u6c42\u5b83\u63a2\u7d22\u4e0d\u540c\u7684\u65b9\u5411\uff0c\u4ee5\u786e\u4fdd S01\u3001 S2 \u548c S3 \u4e4b\u95f4\u6709\u8db3\u591f\u7684\u591a\u6837\u6027\u3002\u7136\u540e\u5728\u4e0b\u4e00\u6b65\uff0c\u5b83\u4f1a\u9009\u62e9\u4e00\u4e2a\u8282\u70b9\u5f00\u59cb\u4f18\u5316\u3002<\/p>\n<p>\u4f8b\u5982\uff0c\u5c1d\u8bd5\u4e0d\u540c\u7684\u6b65\u9aa4\u6765\u4fee\u590d\u95ee\u9898\uff0c\u5982\u679c\u6210\u529f\u4fee\u590d\u4e86\uff0c\u5b83\u5c31\u6210\u4e3a\u4e00\u4e2a\u6709\u6548\u7684\u89e3\u51b3\u65b9\u6848\u3002\u7136\u540e\u8fd9\u4e2a\u89e3\u51b3\u65b9\u6848\u5c31\u88ab\u5b58\u50a8\u4e3a\u4e00\u4e2a\u6709\u6548\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u6b64\u65f6\u4f60\u6709\u4e86\u4e00\u4e2a\u5f53\u524d\u6700\u4f73\u8282\u70b9\uff0c\u6bd4\u5982 S5\uff0c\u7136\u540e\u5b83\u5f00\u59cb\u63a2\u7d22\u4e0b\u4e00\u4e2a\u8981\u4f18\u5316\u7684\u8282\u70b9\u3002\u5b83\u4f1a\u4fdd\u8bc1\u6bcf\u4e2a\u8349\u62df\u7684\u89e3\u51b3\u65b9\u6848\u81f3\u5c11\u88ab\u63a2\u7d22\u4e00\u6b21\uff0c\u5e76\u4e14\u4f1a\u4ece S2 \u7b49\u8282\u70b9\u5206\u522b\u751f\u6210\u53e6\u4e00\u4e2a\u6539\u8fdb\u65b9\u6848\uff0c\u7136\u540e\u8bc4\u4f30\u4e3a\u89e3\u51b3\u65b9\u68486\u62167\uff0c\u8fd9\u4e2a\u8fc7\u7a0b\u4f1a\u4e0d\u65ad\u6301\u7eed\uff0c\u76f4\u5230\u7528\u5c3d\u6240\u6709\u7684\u4f18\u5316\u6b65\u9aa4\u3002<\/p>\n<p>\u6240\u4ee5\u6700\u7ec8\uff0c\u9009\u62e9\u6700\u4f18\u89e3\u5176\u5b9e\u76f8\u5f53\u7b80\u5355\uff0c\u56e0\u4e3a\u6240\u6709\u8fd9\u4e9b\u89e3\u51b3\u65b9\u6848\u90fd\u662f\u7528\u76f8\u540c\u7684\u8bc4\u4f30\u6307\u6807\u6765\u8bc4\u4f30\u7684\u3002\u6240\u4ee5\uff0c\u57fa\u4e8e\u8bc4\u4f30\u6307\u6807\uff0c\u4f60\u5c31\u80fd\u5f97\u5230\u90a3\u4e2a\u6700\u4f18\u89e3\u3002<\/p>\n<p>\u662f\u4ec0\u4e48\u5b9a\u4e49\u4e86\u6574\u4e2a\u8fc7\u7a0b\u5462\uff1f\u6709\u51e0\u4e2a\u5173\u952e\u7ec4\u4ef6\u3002\u9996\u5148\u662f\u641c\u7d22\u7b56\u7565\u3002\u5728\u8fd9\u4e2a\u6848\u4f8b\u4e2d\uff0c\u6211\u4eec\u5b9e\u9645\u4e0a\u91c7\u7528\u4e86\u4e00\u4e2a\u975e\u5e38\u7b80\u5355\u7684\u70ed\u7f16\u7801\u7b56\u7565\u3002<\/p>\n<p>\u5728\u8d77\u8349\u9636\u6bb5\uff0c\u5f53\u5b83\u8d77\u8349\u591a\u4e2a\u89e3\u51b3\u65b9\u6848\u65f6\uff0c\u7531\u4e8e\u5b83\u8fd8\u6ca1\u6709\u4e00\u68f5\u6811\uff0c\u4e5f\u5c31\u662f\u8bf4\u6211\u4eec\u8fd8\u6ca1\u6709\u5206\u914d\u521d\u59cb\u89e3\u51b3\u65b9\u6848\uff0c\u5b83\u4f1a\u521b\u5efa\u591a\u4e2a\u89e3\u51b3\u65b9\u6848\u6765\u63a2\u7d22\u4e0d\u540c\u7684\u65b9\u6cd5\u3002\u800c\u5728\u8c03\u8bd5\u9636\u6bb5\uff0c\u5f53\u5b83\u8fdb\u5165\u8c03\u8bd5\u9636\u6bb5\u540e\uff0c\u5b83\u4f1a\u6709\u4e00\u4e2a\u6700\u5927\u8c03\u8bd5\u6b65\u6570\u9650\u5236\uff0c\u5b83\u4f1a\u5728\u90a3\u4e2a\u8282\u70b9\u505c\u7559\uff0c\u76f4\u5230\u8fbe\u5230\u5141\u8bb8\u7684\u6700\u5927\u8c03\u8bd5\u6b65\u6570\u3002<\/p>\n<p>\u901a\u5e38\u6211\u4eec\u4f1a\u5c06\u8fd9\u4e2a\u6700\u5927\u8c03\u8bd5\u6b65\u6570\u8bbe\u7f6e\u4e3a10\u523020\u6b65\uff0c\u4ee5\u907f\u514d\u8fd9\u4e2aAgent\u82b1\u8d39\u8fc7\u591a\u65f6\u95f4\u5728\u8c03\u8bd5\u4e0a\uff0c\u4ece\u800c\u9677\u5165\u51e0\u4e4e\u65e0\u9650\u5faa\u73af\uff0c\u6d6a\u8d39\u5927\u91cf\u65f6\u95f4\u548c\u8ba1\u7b97\u8d44\u6e90\u3002\u5f53\u7136\uff0c\u6700\u91cd\u8981\u4e5f\u6700\u6709\u8da3\u7684\u90e8\u5206\u5e76\u4e0d\u662f\u4ec0\u4e48\u65f6\u5019\u9009\u62e9\u4e00\u4e2a\u8282\u70b9\u6765\u8fdb\u884c\u6539\u8fdb\u3002<\/p>\n<p>\u6240\u4ee5\u5f53\u5b83\u5b8c\u6210\u8c03\u8bd5\u6216\u8d77\u8349\u540e\uff0c\u5c31\u4f1a\u8fdb\u5165\u4e00\u4e2a\u9636\u6bb5\uff0c\u6765\u6539\u8fdb\u4e00\u4e2a\u6876\u8282\u70b9\u3002\u8fd9\u53ea\u662f\u4e00\u4e2a\u8d2a\u5a6a\u7b97\u6cd5\uff0c\u5b83\u4f1a\u9009\u62e9\u6811\u4e2d\u5f53\u524d\u8868\u73b0\u6700\u597d\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u7136\u540e\u51b3\u5b9a\u8fdb\u4e00\u6b65\u4f18\u5316\u6811\u4e2d\u8868\u73b0\u6700\u9ad8\u7684\u90a3\u4e2a\u8282\u70b9\u3002<\/p>\n<p>\u6240\u4ee5\u5728\u7f16\u7801\u64cd\u4f5c\u7b26\u4e2d\uff0c\u6211\u4eec\u4e5f\u4f1a\u6839\u636e\u4e0d\u540c\u7684\u9636\u6bb5\u91c7\u7528\u4e0d\u540c\u7684\u63d0\u793a\u7b56\u7565\u3002\u6bd4\u5982\u5728\u8d77\u8349\u9636\u6bb5\uff0c\u6211\u4eec\u4f1a\u9f13\u52b1\u5b83\u4e3a\u6a21\u578b\u67b6\u6784\u548c\u7279\u5f81\u5de5\u7a0b\u5236\u5b9a\u4e00\u4e2a\u8ba1\u5212\uff0c\u5e76\u8981\u6c42\u5b83\u751f\u6210\u4e00\u4e2a\u5355\u6587\u4ef6Python\u7a0b\u5e8f\u6765\u5b9e\u73b0\u8fd9\u4e2a\u8ba1\u5212\u3002\u5728\u5e95\u5c42\u9636\u6bb5\uff0cAgent\u4f1a\u6536\u5230\u9519\u8bef\u65e5\u5fd7\u548c\u5806\u6808\u8ddf\u8e2a\uff0c\u4ee5\u8bc6\u522b\u95ee\u9898\u6240\u5728\u3002<\/p>\n<p>\u7136\u540e\uff0c\u5b83\u4f1a\u901a\u8fc7\u4fdd\u7559\u6574\u4f53\u5148\u524d\u7684\u65b9\u6cd5\u6765\u7ea0\u6b63\u95ee\u9898\u3002\u56e0\u6b64\uff0c\u6211\u4eec\u786e\u4fdd\u8c03\u8bd5\u5b9e\u9645\u4e0a\u4e0d\u4f1a\u6539\u53d8\u89e3\u51b3\u65b9\u6848\u672c\u8eab\u3002\u5728\u6539\u8fdb\u6a21\u5f0f\u6216\u6539\u8fdb\u9636\u6bb5\uff0c\u6211\u4eec\u4f1a\u63d0\u793aAgent\u63d0\u51fa\u4e00\u4e2a\u539f\u5b50\u7ea7\u522b\u7684\u6539\u53d8\u3002\u8fd9\u662f\u53e6\u4e00\u4e2a\u6211\u4eec\u5e0c\u671b\u7eb3\u5165\u8fd9\u4e2a\u6846\u67b6\u7684\u89c2\u5bdf\u7ed3\u679c\uff0c\u5373\u6bcf\u4e00\u6b65\u5b9e\u9645\u4e0a\u90fd\u662f\u53ef\u89e3\u91ca\u7684\u3002\u884c\u52a8\u672c\u8eab\u662f\u53ef\u89e3\u91ca\u7684\uff0c\u5e76\u4e14\u662f\u539f\u5b50\u6027\u7684\u3002<\/p>\n<p>\u56e0\u6b64\uff0c\u6211\u4eec\u4e0d\u5141\u8bb8Agent\u6216\u5927\u578b\u8bed\u8a00\u6a21\u578b\u4e00\u6b21\u63d0\u51fa\u591a\u4e2a\u6539\u8fdb\u3002\u76f8\u53cd\uff0c\u6211\u4eec\u4f1a\u63d0\u793a\u5b83\u9010\u6b65\u3001\u589e\u91cf\u5730\u8fdb\u884c\u6539\u8fdb\u3002\u5728\u8fd9\u4e2a\u8fc7\u7a0b\u4e2d\uff0c\u6211\u4eec\u4e0d\u4f1a\u8df3\u8fc7\u4efb\u4f55\u4e2d\u95f4\u6b65\u9aa4\u7684\u4f18\u5316\u60f3\u6cd5\uff0c\u8fd9\u4f7f\u5f97\u5b83\u80fd\u591f\u8fdb\u884c\u66f4\u7ec6\u81f4\u7684\u63a2\u7d22\uff0c\u5e76\u4e14\u5728\u6574\u4f53\u4e0a\u66f4\u5177\u53ef\u89e3\u91ca\u6027\u3002<\/p>\n<p>\u4e5f\u5c31\u662f\u8bf4\uff0c\u5b83\u80fd\u591f\u66f4\u597d\u5730\u5c55\u793a\u51fa\u8fbe\u5230\u6700\u4f18\u89e3\u7684\u6700\u4f73\u8def\u5f84\u662f\u4ec0\u4e48\u3002\u4f8b\u5982\uff0c\u5207\u6362\u4f18\u5316\u5668\u3001\u6dfb\u52a0\u4e00\u5c42\u3001\u4f7f\u7f51\u7edc\u53d8\u5f97\u66f4\u6df1\uff0c\u6216\u8005\u4ece\u4e00\u79cd\u67b6\u6784\u8f6c\u6362\u5230\u53e6\u4e00\u79cd\u67b6\u6784\u3001\u6dfb\u52a0\u6b63\u5219\u5316\u7b49\u3002\u5982\u679c\u4f60\u68c0\u67e5\u5b83\u6700\u7ec8\u751f\u6210\u7684\u6811\u8f68\u8ff9\u6216\u6811\u7ed3\u6784\uff0c\u4f60\u4f1a\u53d1\u73b0\u5f88\u591a\u8fd9\u6837\u7684\u539f\u5b50\u4f18\u5316\u6b65\u9aa4\uff0c\u800c\u4e14\u5f88\u591a\u65f6\u5019\u8fd9\u4e9b\u6b65\u9aa4\u662f\u975e\u5e38\u6709\u6d1e\u5bdf\u529b\u7684\u3002<\/p>\n<p>\u6700\u540e\uff0c\u56e0\u4e3a\u4e00\u4e2a\u6700\u5927\u7684\u95ee\u9898\u662f\u4f60\u9700\u8981\u7ba1\u7406\u4e0a\u4e0b\u6587\uff0c\u6bd4\u5982\u53ef\u80fd\u9700\u8981\u8fd0\u884c8\u4e2a\u6b65\u9aa4\u3002\u4f8b\u5982\uff0cOpenAI\u8fd0\u884c\u4e86500\u4e2a\u6b65\u9aa4\uff0c\u5373\u4f7f\u662fGemini\uff0c\u4e5f\u6ca1\u6709\u529e\u6cd5\u771f\u6b63\u5904\u7406\u90a3\u4e48\u957f\u7684\u4e0a\u4e0b\u6587\u3002\u6240\u4ee5\uff0c\u5fc5\u987b\u6709\u4e00\u79cd\u65b9\u6cd5\u6765\u7ba1\u7406\u4e0a\u4e0b\u6587\u3002\u8fd9\u5c31\u662f\u6211\u4eec\u6240\u8bf4\u7684\u603b\u7ed3\u64cd\u4f5c\u7b26\uff0c\u5b83\u4f1a\u63d0\u53d6\u76f8\u5173\u4fe1\u606f\uff0c\u4ee5\u907f\u514d\u4e0a\u4e0b\u6587\u8fc7\u8f7d\u3002<\/p>\n<p>\u603b\u7ed3\u64cd\u4f5c\u7b26\u4f1a\u5305\u542b\u6027\u80fd\u6307\u6807\uff0c\u6bd4\u5982\u5f53\u524d\u7684\u51c6\u786e\u7387\u3001\u9ad8\u53c2\u6570\u8bbe\u7f6e\u548c\u8c03\u8bd5\u9636\u6bb5\u7684\u4fe1\u606f\u3002\u8fd9\u662f\u975e\u5e38\u91cd\u8981\u7684\uff0c\u5c24\u5176\u662f\u5728\u8c03\u8bd5\u9636\u6bb5\u3002\u597d\u5904\u662f\u6211\u4eec\u53ef\u4ee5\u622a\u65ad\u5b83\u4e4b\u524d\u53ef\u4ee5\u5904\u7406\u7684\u8282\u70b9\u6570\u91cf\u3002<\/p>\n<p>\u6211\u4eec\u53ef\u4ee5\u5c06\u603b\u7ed3\u540e\u7684\u4fe1\u606f\u653e\u5165\u5927\u578b\u8bed\u8a00\u6a21\u578b\u7684\u4e0a\u4e0b\u6587\u4e2d\uff0c\u4ee5\u751f\u6210\u8c03\u8bd5\u8282\u70b9\u6216\u6539\u8fdb\u8282\u70b9\u3002\u8fd9\u5c06\u4fdd\u6301\u4e00\u4e2a\u51e0\u4e4e\u6052\u5b9a\u7684\u7a97\u53e3\u5927\u5c0f\uff0c\u4f9bAgent\u4f7f\u7528\uff0c\u8fd9\u4f7f\u6211\u4eec\u80fd\u591f\u771f\u6b63\u6269\u5c55\u5230\u5f88\u957f\u7684\u65f6\u95f4\u8303\u56f4\uff0c\u6bd4\u5982\u5bf9\u6bd4\u6b65\u9aa4\u3002<\/p>\n<p>\u800c\u4e14\uff0c\u56e0\u4e3a\u6211\u4eec\u5c06\u5176\u5b9a\u4e49\u4e3a\u9010\u6b65\u6539\u8fdb\uff0c\u8fd9\u4e5f\u4f7f\u5f97\u6574\u4e2a\u4f18\u5316\u64cd\u4f5c\u7b26\u53d8\u5f97\u65e0\u72b6\u6001\u3002\u5b83\u4e0d\u518d\u4f9d\u8d56\u4e8e\u6574\u4e2a\u8f68\u8ff9\uff0c\u800c\u662f\u65e0\u72b6\u6001\u7684\uff0c\u4e0d\u4f1a\u50cf\u63d0\u793a\u6216\u4e0a\u4e0b\u6587\u5927\u5c0f\u90a3\u6837\u5448\u7206\u70b8\u5f0f\u589e\u957f\u3002<\/p>\n<\/p>\n<p><strong>\u4e09\u3001\u805a\u7126\u901a\u7528\u578b GUI Agent<\/strong><\/p>\n<\/p>\n<p>\u6765\u81ea Huawei London \u7684\u90b5\u5764\u505a\u4e86\u4e3b\u9898\u4e3a\u300aTowards generalist GUl Agents: model and optimization\u300b\u7684\u6f14\u8bb2\uff0c\u4ecb\u7ecd\u4e86\u9762\u5411\u901a\u7528\u578b GUI Agent \u7684\u6a21\u578b\u548c\u4f18\u5316\u3002<\/p>\n<p>\u4ee5\u4e0b\u662fGUI Agent\u7684\u6f14\u793a\uff0c\u4ed6\u4eec\u6709\u4e0d\u540c\u7684\u4efb\u52a1\uff0c\u6bd4\u5982\u6211\u4eec\u6709\u4e09\u4e2a\u4e0d\u540c\u7684\u5e73\u53f0\u3002\u7b2c\u4e00\u4e2a\u662f\u83b7\u53d6\u5173\u4e8e\u7f8e\u5143\u7684\u7814\u7a76\u7ed3\u679c\uff0c\u6211\u4eec\u53ef\u4ee5\u4ece\u4e3b\u7528\u6237\u754c\u9762\u9875\u9762\u5f00\u59cb\u3002\u7136\u540e\uff0c\u6211\u4eec\u53ef\u4ee5\u6267\u884c\u4e00\u4e9b\u6b65\u9aa4\u524d\u5f80\u9ea6\u5f53\u52b3\uff0c\u8fdb\u5165\u9ea6\u5f53\u52b3\u9910\u5385\u5e76\u641c\u7d22\u90a3\u91cc\u7684\u85af\u6761\uff0c\u6211\u4eec\u8fd8\u53ef\u4ee5\u8bbe\u7f6e\u591a\u4e2a\u6b65\u9aa4\u5e76\u63d0\u9ad8\u76ee\u6807\u3002\u8fd9\u5c31\u662fGUI Agent\u53ef\u4ee5\u5e2e\u52a9\u6211\u4eec\u7684\u5730\u65b9\u3002<\/p>\n<p>\u5728\u53e6\u4e00\u4e2a\u7f51\u7ad9\u4e0a\uff0cGUI Agent \u4e5f\u8bb8\u53ef\u4ee5\u627e\u5230\u4e00\u4e9b\u66f4\u597d\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u5e2e\u52a9\u4eba\u7c7b\u5b8c\u6210\u8fd9\u7c7b\u4efb\u52a1\u3002\u8fd9\u5c31\u662f GUI Agent \u7684\u610f\u4e49\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-78cbd009ebfe9c9a5f56309c4263ce0f.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-78cbd009ebfe9c9a5f56309c4263ce0f.png\"><\/a><\/p>\n<p>\u4ece2023\u5e74\u52302025\u5e74\uff0c\u4f60\u53ef\u4ee5\u770b\u5230 GUI Agent \u5df2\u7ecf\u5e7f\u6cdb\u6d41\u884c\u8d77\u6765\u3002\u5b83\u91cd\u65b0\u5f15\u53d1\u4e86\u5bf9Agent\u7814\u7a76\u7684\u5173\u6ce8\uff0c\u65e0\u8bba\u662f\u5b66\u672f\u754c\u8fd8\u662f\u5927\u578b\u79d1\u6280\u516c\u53f8\u90fd\u5728\u5173\u6ce8GUI Agent\u3002\u8fd9\u79cd\u5173\u6ce8\u4e0d\u4ec5\u5c40\u9650\u4e8e\u79fb\u52a8\u8bbe\u5907\uff0c\u8fd8\u6db5\u76d6\u4e86\u7f51\u7ad9\u548c\u8ba1\u7b97\u9886\u57df\u3002<\/p>\n<\/p>\n<p>\u6211\u4eec\u8fd8\u5c06\u8fd9\u4e9b\u4e3b\u9898\u5f15\u5165\u5230 GUI Agent \u4e2d\uff0c\u4f8b\u5982\u7b2c\u4e00\u4e2a\u662f\u5173\u4e8e\u751f\u6210\u7684\u3002\u5b9e\u9645\u4e0a\uff0c\u4f7f\u7528\u8fd9\u79cd\u6a21\u578b\u6765\u6307\u5bfc\u884c\u52a8\u5e76\u4e0d\u597d\u3002\u56e0\u6b64\uff0c\u5f53\u6211\u4eec\u4e3aGUI Agent\u63d0\u4f9b\u5f53\u524d\u6a21\u578b\u80fd\u529b\u65f6\uff0c\u6211\u4eec\u6709\u4e0d\u540c\u7c7b\u578b\u7684\u6539\u8fdb\u65b9\u6cd5\u6765\u5b9e\u73b0\u66f4\u597d\u7684\u6a21\u578b\u3002\u4f46\u4f60\u5fc5\u987b\u7406\u89e3\uff0c\u5982\u4f55\u5b9e\u73b0\uff0c\u5982\u4f55\u4e3aUI\u8bbe\u8ba1Pythonian\uff0c\u4ee5\u53ca\u5982\u4f55\u8bbe\u8ba1\u5206\u5c42\u67b6\u6784\u3002<\/p>\n<p>\u7b2c\u4e8c\u4e2a\u662f\u5173\u4e8eAgent\u7cfb\u7edf\u3002\u5728\u6211\u4eec\u6709\u4e86\u9879\u76ee\u6a21\u578b\u4e4b\u540e\uff0c\u6211\u4eec\u8fd8\u9700\u8981\u4e00\u4e9b\u5176\u4ed6\u81ea\u52a8\u5316\u7684\u6a21\u578b\u6765\u63d0\u9ad8GUI Agent\u7684\u6027\u80fd\u548c\u6548\u7387\u3002\u4e3a\u4e86\u5b9e\u73b0\u8fd9\u4e00\u70b9\uff0c\u6211\u4eec\u6709\u4e0d\u540c\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u6211\u4eec\u53ef\u4ee5\u6709\u6570\u636e\u5e93\u89c4\u5212\u3001\u53cd\u5c04\u673a\u5236\u3001\u4e5f\u53ef\u4ee5\u4f7f\u7528\u8bb0\u5fc6\u548c\u68c0\u7d22\u5668\u3002<\/p>\n<p>\u6211\u8fd8\u60f3\u5f3a\u8c03\u7684\u53e6\u4e00\u4e2a\u91cd\u70b9\u662f\uff0c\u6211\u4eec\u8fd8\u60f3\u505a\u5f88\u591a\u5173\u4e8e\u5fae\u8c03\u7684\u5de5\u4f5c\uff0c\u56e0\u4e3a\u53ef\u4ee5\u8bf4\u5f3a\u5316\u5b66\u4e60\u5bf9\u4e8eAgent\u7684\u5fae\u8c03\u662f\u975e\u5e38\u91cd\u8981\u7684\uff0c\u6211\u4eec\u9700\u8981\u67d0\u79cd\u76ee\u7684\u3002\u4f8b\u5982\uff0c\u5982\u4f55\u5229\u7528\u5f53\u524d\u6a21\u578b\u4f5c\u4e3a\u8bc4\u5224\u3002\u5982\u4f55\u53d1\u6325\u751f\u6210\u89d2\u8272\u6a21\u578b\u7684\u4f5c\u7528\uff0c\u4ee5\u53ca\u5982\u4f55\u4e3a\u89d2\u8272\u6a21\u578b\u8fdb\u884c\u66f4\u597d\u7684\u5fae\u8c03\u3002\u540c\u65f6\uff0c\u6211\u4eec\u9700\u8981\u627e\u5230\u66f4\u597d\u7684\u65b9\u6cd5\u6765\u5b9e\u73b0\u6211\u4eec\u9ad8\u6548\u3001\u53ef\u9760\u4e14\u7a33\u5065\u7684\u5f3a\u5316\u5b66\u4e60\u8bad\u7ec3\uff0c\u5e76\u4e14\u6211\u4eec\u9700\u8981\u627e\u5230\u6700\u9002\u5408GUI Agent\u7684\u7b97\u6cd5\u3002<\/p>\n<p>\u6700\u540e\u4e00\u4e2a\u95ee\u9898\u662f\u5173\u4e8e\u8bc4\u4f30\u7684\u3002\u5f53\u8bbe\u8ba1\u4e0d\u540c\u7684\u57fa\u51c6\u4efb\u52a1\u65f6\uff0c\u5bf9\u4e8e\u4e24\u4e2aAgent\u6765\u8bf4\u8bc4\u4f30\u975e\u5e38\u91cd\u8981\uff0c\u6211\u4eec\u9700\u8981\u8bbe\u8ba1\u8bc4\u4f30\u6307\u6807\u3002<\/p>\n<p>\u540c\u6837\uff0c\u6211\u4eec\u4e5f\u63d0\u51fa\u4e00\u4e9b\u7814\u7a76\u95ee\u9898\u3002<\/p>\n<p>\u7b2c\u4e00\u4e2a\u95ee\u9898\u662f\uff0c\u6211\u4eec\u9700\u8981\u627e\u5230\u5e76\u63d0\u51fa\u4e00\u4e2a\u57fa\u51c6\u6d4b\u8bd5\u3002\u56e0\u4e3a\u76ee\u524d\u6211\u4eec\u53ef\u4ee5\u770b\u5230\u5f88\u591a\u5173\u4e8eGUI Agent\u7684\u8bba\u6587\u3002\u6240\u4ee5\u6211\u4eec\u9700\u8981\u8bbe\u8ba1\u4e00\u4e2a\u5168\u9762\u7684\u57fa\u51c6\u6d4b\u8bd5\uff0c\u5b83\u53ef\u4ee5\u7528\u4e8e\u4e0d\u540c\u7684\u5e94\u7528\u7a0b\u5e8f\uff0c\u4ee5\u53ca\u4e0d\u540c\u7684Agent\u3002\u56e0\u6b64\uff0c\u5f53\u4f60\u627e\u5230\u4e00\u4e2a\u7528\u4e8e\u8bc4\u4f30Agent\u7684\u6d41\u7a0b\u65f6\uff0c\u5b83\u4e0d\u4ec5\u4ec5\u4f9d\u8d56\u4eba\u5de5\u5e72\u9884\u3002<\/p>\n<p>\u7b2c\u4e8c\u90e8\u5206\u662f\uff0c\u6211\u4eec\u9700\u8981\u8bbe\u8ba1\u4e00\u4e2a\u884c\u52a8\u6a21\u578b\u3002\u4f17\u6240\u5468\u77e5\uff0c\u5982\u679c\u6211\u4eec\u4ec5\u4ec5\u4f7f\u7528\u5f53\u524d\u7684\u57fa\u7840\u6a21\u578b\u6765\u6267\u884cAgent\u4efb\u52a1\uff0c\u90a3\u4e48\u6211\u4eec\u9700\u8981\u627e\u5230\u4e00\u4e9b\u65b9\u6cd5\u6765\u8bad\u7ec3\u51fa\u6027\u80fd\u826f\u597d\u4e14\u9ad8\u6548\u7684\u6a21\u578b\u3002<\/p>\n<p>\u6700\u540e\u4e00\u4e2a\u95ee\u9898\u662f\u5173\u4e8e\u5982\u4f55\u4e3aGUI Agent\u8fdb\u884c\u9ad8\u6548\u7684\u5f3a\u5316\u5b66\u4e60\u5fae\u8c03\u3002\u6211\u4eec\u5fc5\u987b\u8ba9Agent\u80fd\u591f\u5145\u5206\u5229\u7528\u6709\u9650\u7684\u6570\u636e\uff0c\u5e76\u9010\u6b65\u63d0\u9ad8\u6027\u80fd\u3002\u5bf9\u4e8eGUI Agent\u6765\u8bf4\uff0c\u8fdb\u884c\u5fae\u8c03\u5e76\u4e0d\u662f\u4e00\u4ef6\u5bb9\u6613\u7684\u4e8b\uff0c\u56e0\u6b64\u6211\u4eec\u8fd8\u9700\u8981\u627e\u5230\u4e00\u4e9b\u65b9\u6cd5\u6765\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\u3002<\/p>\n<p><strong>\u56db\u3001DeepSeek \u5f3a\u5316\u5b66\u4e60\u7684\u201cAha\u65f6\u523b\u201d<\/strong><\/p>\n<p>\u6765\u81ea UCL \u7684\u5b8b\u7814 \uff0c\u505a\u4e86\u4e3b\u9898\u4e3a\u300aThe Power of Reinforcement Learning in LLM Reasoning\u300b\u7684\u6f14\u8bb2\uff0c\u8ba8\u8bba\u4e86\u5f3a\u5316\u5b66\u4e60\u5728\u5927\u578b\u8bed\u8a00\u6a21\u578b\u63a8\u7406\u4e2d\u7684\u4f5c\u7528\u3002<\/p>\n<p>\u8fd9\u662fR1-zero\u7684\u7ed3\u679c\uff0c\u5b83\u7684\u57fa\u51c6\u6d4b\u8bd5\u7ed3\u679c\u975e\u5e38\u597d\uff0c\u751a\u81f3\u6bd4OpenAI\u7684o1\u8fd8\u8981\u51fa\u8272\u3002\u66f4\u4ee4\u4eba\u5370\u8c61\u6df1\u523b\u7684\u662f\uff0c\u5b83\u7684\u8bad\u7ec3\u8fc7\u7a0b\u975e\u5e38\u7a33\u5b9a\u3002\u800c\u4e14\u5b83\u5728\u6269\u5c55\u6027\u65b9\u9762\u8868\u73b0\u51fa\u8272\u3002\u5bf9\u4e8eR1-zero\u6765\u8bf4\uff0c\u5b83\u5728\u8fc7\u53bb\u7684\u51c6\u786e\u7387\u65b9\u9762\u5e76\u4e0d\u6bd4\u6700\u65b0\u7684\u539f\u59cb\u7248\u672c\u66f4\u597d\u3002\u4f46\u5f53\u4f60\u5c1d\u8bd5\u751f\u6210\u5185\u5bb9\u65f6\uff0c\u5b83\u663e\u7136\u80fd\u591f\u4ea7\u751f\u66f4\u597d\u7684\u7ed3\u679c\u3002<\/p>\n<p>\u8fd8\u6709\u4ed6\u4eec\u5728\u8bba\u6587\u4e2d\u63d0\u5230\u7684\u4e00\u4e2a\u201cAha\u65f6\u523b\u201d\u3002\u5728\u5f3a\u5316\u5b66\u4e60\u9636\u6bb5\uff0c\u8bed\u8a00\u6a21\u578b\u5b66\u4f1a\u4e86\u589e\u52a0\u601d\u8003\u9884\u7b97\u3002\u8fd9\u53ef\u80fd\u662f\u56e0\u4e3a\u4f60\u9700\u8981\u66f4\u591a\u7684token\u6765\u8fdb\u884c\u601d\u8003\uff0c\u4ece\u800c\u89e3\u51b3\u4e00\u4e9b\u96be\u9898\u3002\u8fd9\u4e00\u89c2\u70b9\u5f97\u5230\u4e86\u4e00\u4e9b\u652f\u6301\u3002\u4ed6\u4eec\u53d1\u73b0\uff0c\u5f53Agent\u4f7f\u7528\u67d0\u4e9b\u5173\u952e\u8bcd\u65f6\uff0c\u5b83\u4eec\u4f1a\u8fdb\u884c\u5404\u79cd\u56de\u6eaf\u3001\u81ea\u6211\u62a5\u544a\u548c\u590d\u6742\u63a8\u7406\u3002\u4f46\u5bf9\u6b64\u4e5f\u6709\u53e6\u4e00\u79cd\u53ef\u80fd\u7684\u89e3\u91ca\u3002\u9996\u5148\uff0c\u57fa\u7840\u6a21\u578b\u672c\u8eab\u5df2\u7ecf\u5177\u5907\u81ea\u6211\u7ea0\u6b63\u7684\u80fd\u529b\u3002\u6240\u4ee5\uff0c\u4ece\u6280\u672f\u89d2\u5ea6\u6765\u770b\uff0c\u8fd9\u5e76\u4e0d\u662f\u4e00\u4e2a\u975e\u5e38\u201cAha\u201d\u7684\u65f6\u523b\uff0c\u800c\u662f\u8868\u660e\u5f3a\u5316\u5b66\u4e60\u786e\u5b9e\u53ef\u4ee5\u5728\u7b80\u5355\u7684\u8bbe\u7f6e\u4e0b\u53d1\u6325\u4f5c\u7528\u3002\u6211\u8ba4\u4e3a\u8fd9\u624d\u662f\u6700\u91cd\u8981\u7684\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-33043f0d6c217facd3be9ce3694c740a.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-33043f0d6c217facd3be9ce3694c740a.png\"><\/a><\/p>\n<p>\u5bf9\u4e8eR1\u7248\u672c\uff0c\u4ed6\u4eec\u4f7f\u7528\u521d\u59cb\u6570\u636e\u5f00\u59cb\u8bad\u7ec3\uff0c\u7136\u540e\u5728\u5f3a\u5316\u5b66\u4e60\uff08RL\uff09\u7684\u573a\u666f\u4e2d\u8fdb\u884c\u8bad\u7ec3\u3002\u7ed3\u8bba\u662f\uff0c\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u5177\u6709\u5f3a\u5927\u7684\u80fd\u529b\uff0c\u80fd\u591f\u8fdb\u884c\u7a33\u5b9a\u7684\u5f3a\u5316\u5b66\u4e60\uff0c\u800c\u5c0f\u578b\u8bed\u8a00\u6a21\u578b\u5219\u91c7\u7528\u77e5\u8bc6\u84b8\u998f\u7684\u65b9\u5f0f\u8fdb\u884c\u8bad\u7ec3\u3002<\/p>\n<p>\u5728R1\u53d1\u5e03\u4e4b\u540e\uff0c\u6211\u4eec\u5728\u540e\u7eed\u7684\u8bad\u7ec3\u548c\u589e\u91cf\u66f4\u65b0\u4e2d\u63a8\u51fa\u4e86TinyZero\uff0c\u8fd9\u662f\u4e00\u4e2a30\u4ebf\u53c2\u6570\u7684\u5927\u578b\u6a21\u578b\u3002\u6b64\u5916\uff0cSimpleRL\u662f\u572870\u4ebf\u53c2\u6570\u7684LLM\u4e0a\u5b9e\u73b0\u7684Zero-RL\u3002\u6240\u4ee5\uff0c\u6240\u6709\u8fd9\u4e9b\u5de5\u4f5c\u90fd\u4f7f\u7528\u4e86Zero-RL\u3002\u81f3\u5c11\u57fa\u672c\u7684\u60f3\u6cd5\u662f\uff0c\u4f60\u9700\u8981\u6709\u4e00\u4e2a\u5f3a\u5927\u7684\u57fa\u7840\u6a21\u578b\uff0c\u4ee5\u4fbf\u80fd\u591f\u5b66\u4f1a\u63a8\u7406\u3002\u4e0b\u4e00\u6b65\u662f\u5b83\u4eec\u53ef\u4ee5\u5b66\u4f1a\u63a2\u7d22\uff0c\u7136\u540e\u5b83\u4eec\u53ef\u4ee5\u5b66\u4f1a\u81ea\u6211\u7ea0\u6b63\u3002<\/p>\n<p>\u6700\u8fd1\u4e5f\u6709\u4e00\u4e9b\u591a\u6a21\u6001\u7684Zero-RL\u5de5\u4f5c\u3002\u8fd9\u4e9b\u5de5\u4f5c\u57fa\u4e8eOpen-R1\u4ee3\u7801\u5e93\u3001OpenRLHF\u4ee3\u7801\u5e93\u6216Verl\u4ee3\u7801\u5e93\u3002<\/p>\n<p>\u6211\u4eec\u4e5f\u5728\u5c0f\u89c4\u6a21\u4e0a\u8fdb\u884c\u4e86\u4e00\u4e9b\u5b9e\u9a8c\u3002\u57fa\u672c\u8bbe\u7f6e\u662f\u6211\u4eec\u5c1d\u8bd5\u5728\u6570\u5b66\u95ee\u9898\u4e0a\u8fdb\u884c\u8bad\u7ec3\uff0c\u9009\u62e9\u7684\u96be\u5ea6\u7b49\u7ea7\u4e3a\u4e09\u5230\u4e94\uff0c\u8fd9\u4e0eSimpleRL\u4e4b\u524d\u7684\u4ee3\u7801\u5e93\u8bbe\u7f6e\u76f8\u540c\uff0c\u6211\u4eec\u53d1\u73b0\u8fd9\u76f8\u5f53\u91cd\u8981\u3002\u6211\u4eec\u9700\u8981\u8fdb\u884c\u7b5b\u9009\uff0c\u800c\u6211\u4eec\u662f\u5728Qwen2.5-Math-7B\u4e0a\u8fdb\u884c\u7684\u3002\u5176\u6027\u80fd\u8868\u73b0\u826f\u597d\uff0c\u5982\u56fe\u6240\u793a\u3002\u84dd\u8272\u7ebf\u6761\u8868\u793a\u4ece\u57fa\u7840\u6a21\u578b\u5f00\u59cb\u7684\u5f3a\u5316\u5b66\u4e60\uff0c\u5b83\u80fd\u591f\u63a8\u5e7f\u5230AIME2024\uff0c\u8fd9\u975e\u5e38\u96be\u4ee5\u89e3\u51b3\u3002<\/p>\n<p>\u4f46\u5b83\u4ec5\u5728\u6570\u5b66\u95ee\u9898\u4e0a\u8fdb\u884c\u8bad\u7ec3\uff0c\u901a\u8fc7\u4f7f\u7528\u76d1\u7763\u5fae\u8c03\u6570\u636e\uff0c\u5b83\u53ef\u4ee5\u5728GSM8k\u4e0a\u83b7\u5f97\u66f4\u597d\u7684\u6027\u80fd\u3002\u4f46\u5728AIME2024\u4e0a\uff0c\u5b83\u7ed9\u51fa\u7684\u7ed3\u679c\u975e\u5e38\u7cdf\u7cd5\u3002\u8fd9\u610f\u5473\u7740\uff0c\u76d1\u7763\u5fae\u8c03\u6570\u636e\u53ef\u80fd\u4f1a\u635f\u5bb3\u5f3a\u5316\u5b66\u4e60\u5e26\u6765\u7684\u6cdb\u5316\u80fd\u529b\u3002\u6211\u4eec\u8fd8\u5728LLaMA\u4e0a\u8fdb\u884c\u4e86\u5b9e\u9a8c\uff0c\u4f46\u7ed3\u679c\u5e76\u4e0d\u7406\u60f3\u3002<\/p>\n<p>\u63a5\u4e0b\u6765\uff0cYan Song \u4e5f\u5206\u4eab\u4e00\u4e9b\u6700\u8fd1\u5de5\u4f5c\u4e2d\u7684\u524d\u77bb\u89c1\u89e3\u3002<\/p>\n<p><strong>\u4e94\u3001\u81ea\u7136\u8bed\u8a00\u5f3a\u5316\u5b66\u4e60\u65b0\u8303\u5f0f<\/strong><\/p>\n<\/p>\n<\/p>\n<p>\u6765\u81ea UCL\u3001Google Deepmind \u7684\u51af\u7199\u680b\u63a2\u8ba8\u4e86\u300aNatural Language Reinforcement Learning\u300b\uff0c\u4ecb\u7ecd\u4e86\u628a\u5f3a\u5316\u5b66\u4e60\u7684\u7ec4\u6210\u90e8\u5206\u7528\u81ea\u7136\u8bed\u8a00\u63cf\u8ff0\u51fa\u6765\u7684\u7406\u5ff5\u3002<\/p>\n<p>\u5982\u679c\u6211\u4eec\u6709\u4e86\u65b0\u7684\u5f3a\u5316\u5b66\u4e60\u8303\u5f0f\uff0c\u90a3\u4f1a\u600e\u6837\uff1f\u5982\u679c\u6211\u4eec\u4e0d\u5b66\u4e60\u5df2\u7ecf\u786e\u5b9a\u7684\u503c\u51fd\u6570\uff0c\u800c\u662f\u5b66\u4e60\u8bed\u8a00\u503c\u51fd\u6570\u5462\uff1f\u4e5f\u5c31\u662f\u8bf4\uff0c\u6211\u4eec\u5c1d\u8bd5\u7528\u81ea\u7136\u8bed\u8a00\u6765\u63cf\u8ff0\u72b6\u6001\u7684\u503c\u3001\u63cf\u8ff0\u72b6\u6001-\u52a8\u4f5c\u5bf9\u3002\u4e3a\u4ec0\u4e48\u4e0d\u5e72\u8106\u628a\u6240\u6709\u5f3a\u5316\u5b66\u4e60\u7684\u7ec4\u6210\u90e8\u5206\u90fd\u7528\u8bed\u8a00\u8868\u8fbe\u51fa\u6765\u5462\uff1f\u8fd9\u6b63\u662f\u6211\u4eec\u6700\u8fd1\u52aa\u529b\u7684\u65b9\u5411\u3002<\/p>\n<p>\u6211\u4eec\u662f\u4ece\u4f20\u7edf\u7684\u5f3a\u5316\u5b66\u4e60\u6982\u5ff5\u4e2d\u6c72\u53d6\u7075\u611f\uff0c\u4f46\u6211\u4eec\u6b63\u5728\u5c06\u6240\u6709\u8fd9\u4e9b\u5f3a\u5316\u5b66\u4e60\u7684\u6982\u5ff5\u91cd\u65b0\u5b9a\u4e49\u4e3a\u81ea\u7136\u8bed\u8a00\u8868\u793a\u7a7a\u95f4\u4e2d\u7684\u5185\u5bb9\u3002\u6211\u4eec\u5c1d\u8bd5\u5c06\u7b56\u7565\u3001\u503c\u51fd\u6570\u3001\u8d1d\u5c14\u66fc\u65b9\u7a0b\u3001\u8499\u7279\u5361\u6d1b\u91c7\u6837\u3001\u65f6\u95f4\u5dee\u5206\u5b66\u4e60\u4ee5\u53ca\u7b56\u7565\u6539\u8fdb\u64cd\u4f5c\u7b26\u7b49\uff0c\u6620\u5c04\u5230\u5b83\u4eec\u7684\u81ea\u7136\u8bed\u8a00\u5bf9\u5e94\u4e2d\u3002<\/p>\n<p>\u8fd9\u91cc\u6211\u5c55\u793a\u4e86\u4e00\u4e2a\u8fd9\u6837\u7684\u4f8b\u5b50\u3002\u5728\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u4f60\u6709\u4e00\u4e2a\u7b56\u7565\uff0c\u5b83\u662f\u5206\u5e03\u5f0f\u7684\u3002\u4f46\u5728\u81ea\u7136\u8bed\u8a00\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u4f60\u53ef\u80fd\u4f1a\u6709\u4e00\u4e2a\u8bed\u8a00\u7b56\u7565\u3002\u4e5f\u5c31\u662f\u8bf4\uff0c\u4f60\u4e0d\u4e00\u5b9a\u9700\u8981\u76f4\u63a5\u5c06\u4f60\u7684\u72b6\u6001\u548c\u52a8\u4f5c\u6620\u5c04\u8fc7\u53bb\u3002\u76f8\u53cd\uff0c\u4f60\u53ef\u4ee5\u5c1d\u8bd5\u8bf4\uff1a\u201c\u6211\u5148\u5bf9\u72b6\u6001\u8fdb\u884c\u4e00\u4e9b\u5206\u6790\uff0c\u7136\u540e\u518d\u91c7\u53d6\u884c\u52a8\u3002\u201d\u6240\u4ee5\uff0c\u8ba9\u6211\u4eec\u7528\u8bed\u8a00\u7b56\u7565\u6765\u54cd\u5e94\u3002<\/p>\n<p>\u5728\u4f20\u7edf\u7684\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u4f60\u6709\u6807\u91cf\u5956\u52b1\uff0c\u5373\u5373\u65f6\u5956\u52b1\u3002\u4f46\u5728\u81ea\u7136\u8bed\u8a00\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u4f60\u53ef\u80fd\u4f1a\u6709\u8bed\u8a00\u53cd\u9988\uff0c\u6bd4\u5982\u201c\u4f60\u8fbe\u5230\u4e86\u76ee\u6807\u201d\uff0c\u8fd9\u4e0d\u4ec5\u4ec5\u50cf\u4f20\u7edf\u5f3a\u5316\u5b66\u4e60\u4e2d\u7684+1\u5956\u52b1\uff0c\u5b83\u53ef\u4ee5\u5305\u542b\u66f4\u4e30\u5bcc\u7684\u4fe1\u606f\u3002<\/p>\n<p>\u5728\u72b6\u6001\u65b9\u9762\uff0c\u4f60\u4e0d\u9700\u8981\u662f\u4e00\u4e2a\u9ad8\u7ef4\u72b6\u6001\uff0c\u4f60\u4e5f\u53ef\u4ee5\u7528\u8bed\u8a00\u6765\u63cf\u8ff0\u72b6\u6001\u3002\u4f8b\u5982\uff0c\u4f60\u53ef\u4ee5\u8fd9\u6837\u8bf4\u201c\u4f60\u6b63\u5728\u6eda\u52a8\u2026\u2026\u201d\u4e4b\u7c7b\u7684\u5185\u5bb9\u3002\u5bf9\u4e8e\u503c\u51fd\u6570\uff0c\u5728\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u6211\u4eec\u8fc7\u53bb\u4e60\u60ef\u4e8e\u4f7f\u7528\u9884\u671f\u7684\u7d2f\u79ef\u5956\u52b1\u3002\u4f46\u73b0\u5728\uff0c\u6211\u4eec\u53ef\u4ee5\u6709\u4e00\u4e2a\u66f4\u81ea\u7136\u7684\u8bed\u8a00\u8868\u793a\u3002\u6211\u4eec\u53ef\u4ee5\u5c1d\u8bd5\u7528\u8bed\u8a00\u63cf\u8ff0\u7b26\u6765\u603b\u7ed3\u672a\u6765\u7684\u8f68\u8ff9\uff0c\u6700\u7ec8\uff0c\u8fd8\u6709\u8d1d\u5c14\u66fc\u65b9\u7a0b\u3002<\/p>\n<p>\u4f20\u7edf\u7684\u8d1d\u5c14\u66fc\u65b9\u7a0b\u8bd5\u56fe\u8861\u91cf\u5f53\u524d\u72b6\u6001\u4e0e\u5176\u540e\u7eed\u72b6\u6001\u4e4b\u95f4\u7684\u5173\u7cfb\u3002\u6211\u4eec\u4e5f\u53ef\u4ee5\u5728\u81ea\u7136\u8bed\u8a00\u8bc4\u4f30\u4e2d\u505a\u7c7b\u4f3c\u7684\u4e8b\u60c5\u3002\u5982\u679c\u4f60\u8bd5\u56fe\u8bc4\u4f30\u5f53\u524d\u7684\u72b6\u6001\uff0c\u90a3\u4e48\u4f60\u7684\u5f53\u524d\u72b6\u6001\u8bc4\u4f30\u4e0d\u53ef\u80fd\u4e0e\u540e\u7eed\u72b6\u6001\u8bc4\u4f30\u76f8\u5dee\u751a\u8fdc\u3002\u56e0\u6b64\uff0c\u4f60\u7684\u81ea\u7136\u8bed\u8a00\u8bc4\u4f30\u5fc5\u987b\u5728\u8fde\u7eed\u72b6\u6001\u4e4b\u95f4\u5177\u6709\u4e00\u81f4\u6027\u3001\u81ea\u6211\u4e00\u81f4\u6027\u3002\u6240\u4ee5\uff0c\u8fd9\u4e5f\u662f\u8d1d\u5c14\u66fc\u65b9\u7a0b\uff0c\u4f46\u5b83\u53d1\u751f\u5728\u81ea\u7136\u8bed\u8a00\u7a7a\u95f4\u4e2d\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-6b94a5231a57b2bd7479e24a9ad700db.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-6b94a5231a57b2bd7479e24a9ad700db.png\"><\/a><\/p>\n<p>\u540c\u6837\u5730\uff0c\u6211\u4eec\u53ef\u4ee5\u5c1d\u8bd5\u5c06\u4f20\u7edf\u7684\u8499\u7279\u5361\u6d1b\u91c7\u6837\u548c\u65f6\u95f4\u5dee\u5206\u5b66\u4e60\u65b9\u6cd5\u6620\u5c04\u5230\u81ea\u7136\u8bed\u8a00\u4e2d\u3002\u5047\u8bbe\u6211\u4eec\u6709\u4e24\u79cd\u8bed\u8a00\u805a\u5408\u5668 G1 \u548c G2 \u3002\u5728\u4f20\u7edf\u7684\u8499\u7279\u5361\u6d1b\u65b9\u6cd5\u4e2d\uff0c\u6211\u4eec\u901a\u8fc7\u91c7\u6837\u5927\u91cf\u7684\u8f68\u8ff9\u6765\u4f30\u8ba1\u5176\u6298\u6263\u5956\u52b1\u7684\u7d2f\u79ef\u548c\uff0c\u5e76\u8bd5\u56fe\u901a\u8fc7\u8ba1\u7b97\u8fd9\u4e9b\u8f68\u8ff9\u7684\u5e73\u5747\u5956\u52b1\u6765\u4f30\u8ba1\u6211\u4eec\u7684\u8f68\u8ff9\u3002<\/p>\n<p>\u5728\u81ea\u7136\u8bed\u8a00\u4e2d\uff0c\u6211\u4eec\u4e5f\u53ef\u4ee5\u505a\u7c7b\u4f3c\u7684\u4e8b\u60c5\u3002\u6211\u4eec\u4ece\u65f6\u95f4\u6b65 (t+1) \u5f00\u59cb\u91c7\u6837\uff0c\u76f4\u5230\u65e0\u7a77\u5927\u3002\u6211\u4eec\u6709\u5f88\u591a\u57fa\u4e8e\u8bed\u8a00\u7684\u51e0\u4f55\u91c7\u6837\u7ed3\u679c\uff0c\u5e76\u4e14\u5047\u8bbe\u6211\u4eec\u6709\u4e00\u4e9b\u8bed\u8a00\u4fe1\u606f\u805a\u5408\u5668\uff0c\u56e0\u4e3a\u5f53\u7136\uff0c\u6211\u4eec\u4e0d\u80fd\u5bf9\u8bed\u8a00\u8fdb\u884c\u6c42\u5e73\u5747\u6216\u6c42\u548c\u64cd\u4f5c\uff0c\u56e0\u4e3a\u5b83\u4eec\u4e0d\u662f\u6570\u5b57\u3002\u4f46\u5982\u679c\u6709\u4e00\u4e2a\u8bed\u8a00\u805a\u5408\u5668\uff0c\u6211\u4eec\u53ef\u4ee5\u8981\u6c42\u5b83\u4ece\u8fd9\u4e9b\u4e0d\u540c\u7684\u91c7\u6837\u8bed\u8a00\u8f68\u8ff9\u4e2d\u603b\u7ed3\u548c\u805a\u5408\u4fe1\u606f\u3002<\/p>\n<p>G1 \u53ef\u4ee5\u8d1f\u8d23\u4ece\u591a\u4e2a\u8f68\u8ff9\u4e2d\u805a\u5408\u8bc4\u4f30\u7ed3\u679c\uff0c\u5e76\u5c06\u6240\u6709\u6b65\u9aa4\u805a\u5408\u5230\u4e00\u4e2a\u9879\u76ee\u4e2d\u3002\u8fd9\u5728\u7269\u7406\u4e0a\u662f\u5b8c\u5168\u76f8\u540c\u7684\u4e8b\u60c5\uff0c\u53ea\u662f\u53d1\u751f\u5728\u4e0d\u540c\u7684\u7a7a\u95f4\u4e2d\u3002\u56e0\u6b64\uff0c\u6211\u4eec\u9700\u8981\u5728\u4f20\u7edf\u5f3a\u5316\u5b66\u4e60\u4e2d\u5b9e\u73b0\u4e0d\u540c\u7684\u805a\u5408\u64cd\u4f5c\u7b26\u3002\u5728\u4f20\u7edf\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u5b83\u662f\u5e73\u5747\u503c\uff1b\u5728\u57fa\u4e8e\u8bed\u8a00\u7684\u6848\u4f8b\u4e2d\uff0c\u5b83\u53ea\u662f\u4e00\u4e2a\u8bed\u8a00\u805a\u5408\u5668\u3002<\/p>\n<p>\u5bf9\u4e8e\u65f6\u95f4\u5dee\u5206\u5b66\u4e60\u4e5f\u662f\u5982\u6b64\u3002\u65f6\u95f4\u5dee\u5206\u5b66\u4e60\u7684\u7406\u5ff5\u662f\uff0c\u5982\u679c\u4f60\u8bd5\u56fe\u8bc4\u4f30\u5f53\u524d\u72b6\u6001\u7684\u503c\uff0c\u4f60\u53ef\u4ee5\u5411\u524d\u8fc8\u51fa\u4e00\u6b65\uff0c\u5e76\u5c06\u5373\u65f6\u5956\u52b1\u4e0e\u672a\u6765\u72b6\u6001\u8bc4\u4f30\u7ed3\u679c\u7ed3\u5408\u8d77\u6765\uff0c\u4ee5\u5f62\u6210\u4f60\u5bf9\u5f53\u524d\u72b6\u6001\u7684\u8bc4\u4f30\u3002<\/p>\n<p>\u5728\u81ea\u7136\u8bed\u8a00\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u6211\u4eec\u4e5f\u53ef\u4ee5\u505a\u5b8c\u5168\u76f8\u540c\u7684\u4e8b\u60c5\u3002\u5047\u8bbe\u6211\u4eec\u6709 G1 \u548c G2 \u8fd9\u4e24\u4e2a\u8bed\u8a00\u805a\u5408\u5668\uff0c G2\u8d1f\u8d23\u5c06\u5373\u65f6\u5956\u52b1\u548c\u672a\u6765\u8bc4\u4f30\u7ed3\u679c\u5408\u5e76\u3002\u6211\u4eec\u53ef\u4ee5\u8ba9 G2 \u63a5\u6536\u5373\u65f6\u8f6c\u6362\u7684\u63cf\u8ff0\u548c\u672a\u6765\u72b6\u6001\u8bc4\u4f30\u7ed3\u679c\uff0c\u5e76\u8ba9 G1 \u8d1f\u8d23\u4ece\u591a\u4e2a\u8f68\u8ff9\u4e2d\u805a\u5408\u8bc4\u4f30\u7ed3\u679c\u3002\u867d\u7136\u5728\u4f20\u7edf\u5f3a\u5316\u5b66\u4e60\u548c\u81ea\u7136\u8bed\u8a00\u5f3a\u5316\u5b66\u4e60\u4e2d\uff0c\u5b83\u4eec\u7684\u5b9e\u73b0\u65b9\u5f0f\u4e0d\u540c\uff0c\u4f46\u7406\u5ff5\u662f\u76f8\u4f3c\u7684\u3002\u4f60\u53ef\u4ee5\u901a\u8fc7\u4e0d\u540c\u7684\u805a\u5408\u5668\u770b\u5230\u6211\u4eec\u7684\u65b0\u8bed\u8a00\u65f6\u95f4\u5dee\u5206\u5b66\u4e60\u662f\u5982\u4f55\u54cd\u5e94\u4f20\u7edf\u65f6\u95f4\u5dee\u5206\u5b66\u4e60\u7684\u3002<\/p>\n<p>\u9610\u8ff0\u5b8c\u6982\u5ff5\u4e4b\u540e\uff0cXidong Feng \u4e5f\u8c08\u4e86\u8fd9\u79cd\u65b9\u6cd5\u7684\u5177\u4f53\u5b9e\u73b0\uff0c\u7ed9\u51fa\u4e86\u5982\u4f55\u5229\u7528\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u5b9e\u73b0\u81ea\u7136\u8bed\u8a00\u5f3a\u5316\u5b66\u4e60\uff08NLRL\uff09\u7684\u51e0\u6761\u8def\u5f84\u3002<\/p>\n<p><strong>\u516d\u3001Qwen\u957f\u6587\u672c\u80fd\u529b\uff1a\u66f4\u5feb\u901f\u5ea6\u3001\u66f4\u4f4e\u6210\u672c<\/strong><\/p>\n<p>\u963f\u91cc\u901a\u4e49\u5343\u95ee\u7684\u6797\u4fca\u65f8\u505a\u4e86\u9898\u4e3a\u300aQwen: Towards Generalist Models\u300b\u7684\u6f14\u8bb2\uff0c\u7cfb\u7edf\u4ecb\u7ecd\u4e86\u901a\u4e49\u5343\u95ee\u5728\u6280\u672f\u3001\u5de5\u7a0b\u5316\u65b9\u9762\u7684\u8fdb\u5c55\uff0c\u4ee5\u4e0b\u662f\u5173\u4e8e\u901a\u4e49\u5343\u95ee\u5927\u6a21\u578b\u5728\u6570\u636e\u91cf\u65b9\u9762\u7684\u7ec6\u8282\u3002<\/p>\n<p>\u5728Qwen2.5\u7248\u672c\u4e2d\uff0c\u6570\u636e\u91cf\u6269\u5c55\u523018T\u3002\u4f46\u73b0\u5728\u6211\u4eec\u6b63\u5728\u8003\u8651\u7528\u66f4\u591a\u7684token\u8fdb\u884c\u8bad\u7ec3\uff0c\u6bd4\u598230\u4e07\u4ebf\u523040\u4e07\u4ebf\u4e2atoken\u3002\u8fd9\u53ea\u662f\u4e00\u4e2a\u5f88\u5927\u7684\u6570\u91cf\uff0c\u4f46\u5bf9\u4e8e\u8bad\u7ec3\u5927\u578b\u6a21\u578b\u6765\u8bf4\u5f88\u91cd\u8981\uff0c\u4e0d\u4ec5\u9700\u8981\u9ad8\u8d28\u91cf\u7684\u6570\u636e\uff0c\u8fd8\u9700\u8981\u5927\u91cf\u7684\u6570\u636e\u3002\u8fd9\u5c31\u662f\u4e3a\u4ec0\u4e48\u6211\u4eec\u5728\u8fdb\u884c\u591a\u9636\u6bb5\u9884\u8bad\u7ec3\uff0c\u56e0\u4e3a\u4f60\u9700\u8981\u8bad\u7ec3\u5927\u91cf\u7684\u6570\u636e\u3002\u4f46\u6570\u636e\u4e2d\u6709\u810f\u6570\u636e\uff0c\u5728\u7b2c\u4e00\u9636\u6bb5\uff0c\u7136\u540e\u662f\u66f4\u9ad8\u8d28\u91cf\u7684\uff0c\u518d\u5230\u4e0d\u540c\u9636\u6bb5\u7684\u66f4\u9ad8\u8d28\u91cf\u3002<\/p>\n<p>\u5728\u6a21\u578b\u89c4\u6a21\u65b9\u9762\uff0c\u6211\u4eec\u5f00\u6e90\u4e867\u79cd\u4e0d\u540c\u89c4\u6a21\u7684\u6a21\u578b\uff0c\u8303\u56f4\u4ece0.5B\u523072B\u3002\u56e0\u6b64\uff0c\u4f60\u53ef\u4ee5\u901a\u8fc7\u8fd9\u4e9b\u4e0d\u540c\u89c4\u6a21\u7684\u6a21\u578b\u6765\u7814\u7a76\u6269\u5c55\u89c4\u5f8b\u3002\u4f46\u76ee\u524d\u6211\u4eec\u8ba8\u8bba\u7684\u662f\u5bc6\u96c6\u578b\u6a21\u578b\uff0c\u6211\u4eec\u73b0\u5728\u6b63\u5728\u8003\u8651MoE\uff08Mixture of Experts\uff09\u6a21\u578b\uff0c\u5b83\u4e0e\u5bc6\u96c6\u578b\u6a21\u578b\u6709\u4e9b\u76f8\u4f3c\uff0c\u4f46\u5728\u8bad\u7ec3\u548c\u5185\u5b58\u6a21\u578b\u65b9\u9762\u6709\u66f4\u591a\u7684\u6280\u672f\u7ec6\u8282\uff0c\u4f46\u603b\u4f53\u4e0a\u5b83\u4ecd\u7136\u9075\u5faa\u6269\u5c55\u89c4\u5f8b\u3002<\/p>\n<p>\u5728\u6269\u5c55\u6a21\u578b\u89c4\u6a21\u65b9\u9762\uff0c\u4e0d\u4ec5\u4ec5\u662f\u6a21\u578b\u672c\u8eab\u7684\u89c4\u6a21\uff0c\u8fd8\u6709\u6fc0\u6d3b\u53c2\u6570\u7684\u89c4\u6a21\u3002\u8fd8\u6709\u4e00\u79cd\u626b\u63cf\u65b9\u5f0f\uff0c\u4f46\u6a21\u578b\u89c4\u6a21\u95e8\u63a7\u5b9e\u9645\u4e0a\u5df2\u88ab\u8bc1\u660e\u662f\u975e\u5e38\u6709\u6548\u7684\u3002\u6211\u4eec\u7684\u5f00\u6e90\u6a21\u578b\u4e2d\u6709\u8bb8\u591a\u7ec6\u8282\uff0c\u4f60\u53ef\u4ee5\u53bb\u67e5\u770b\u8fd9\u4e9b\u7ec6\u8282\u3002<\/p>\n<p>\u5728\u4e0a\u4e0b\u6587\u957f\u5ea6\u6269\u5c55\u65b9\u9762\uff0c\u8fd9\u4e5f\u662f\u6211\u4eec\u4e4b\u524d\u975e\u5e38\u5173\u6ce8\u7684\u4e00\u4e2a\u95ee\u9898\u3002\u4eba\u4eec\u4e4b\u524d\u8bad\u7ec3\u7684\u6a21\u578b\u4e0a\u4e0b\u6587\u957f\u5ea6\u4e3a 32K tokens \u7136\u540e\u6269\u5c55\u5230 128k tokens token\u3002\u771f\u7684\u6709\u8fd9\u4e48\u957f\u5417\uff1f\u4f46\u4f60\u751a\u81f3\u53ef\u4ee5\u5c06\u5176\u6269\u5c55\u5230\u751a\u81f31000\u4e07\uff0c\u8fd9\u4e5f\u662f\u6709\u53ef\u80fd\u7684\u3002\u73b0\u5728\u4eba\u4eec\u6b63\u5728\u8003\u8651\u8f6c\u5411\u65e0\u9650\u957f\u5ea6\uff0c\u6240\u4ee5\u4ed6\u4eec\u5e0c\u671b\u4f7f\u7528\u7c7b\u4f3c\u4f20\u7edf\u7684\u591a\u5934\u6ce8\u610f\u529b\u673a\u5236\u6765\u5b9e\u73b0\u3002\u4e5f\u8bb8\u7ebf\u6027\u6ce8\u610f\u529b\u662f\u672a\u6765\u7684\u53d1\u5c55\u65b9\u5411\uff0c\u4f46\u5728\u8fd9\u4e00\u9886\u57df\u6211\u4eec\u8fd8\u6ca1\u6709\u53d6\u5f97\u91cd\u5927\u7a81\u7834\u3002<\/p>\n<p>\u4f46\u6211\u4eec\u6709\u4e00\u4e9b\u6280\u672f\u53ef\u4ee5\u5c06\u5176\u8fdb\u4e00\u6b65\u6269\u5c55\uff0c\u800c\u65e0\u9700\u8fdb\u4e00\u6b65\u8bad\u7ec3\u3002\u8fd9\u662f\u4e00\u79cd\u65e0\u9700\u8bad\u7ec3\u7684\u6280\u672f\uff0c\u79f0\u4e3a Trunk Attention\uff08\u622a\u65ad\u6ce8\u610f\u529b\uff09\uff0c\u4f60\u53ef\u4ee5\u67e5\u770b Trunk Attention \u7684\u6280\u672f\u62a5\u544a\u3002<\/p>\n<p>\u6211\u4eec\u56e2\u961f\u5728ICML\u4e0a\u53d1\u8868\u4e86\u4e00\u7bc7\u76f8\u5173\u8bba\u6587\uff0c\u8fd8\u6709\u53e6\u4e00\u4e2a\u6280\u672f\u95ee\u9898\u662f\u5728\u90e8\u7f72\u65b9\u9762\u3002\u62e5\u6709100\u4e07\u4e0a\u4e0b\u6587token\u7684\u6a21\u578b\uff0c\u90e8\u7f72\u8d77\u6765\u975e\u5e38\u56f0\u96be\u3002\u56e0\u6b64\uff0c\u4f60\u9700\u8981\u4f7f\u7528\u4e00\u4e9b\u6280\u672f\u6765\u5f15\u5165\u7a00\u758f\u6027\uff0c\u4ece\u800c\u52a0\u5feb\u63a8\u7406\u901f\u5ea6\u3002\u4f60\u53ef\u4ee5\u770b\u5230\uff0c\u4e4b\u524d\u9996\u6b21\u751f\u6210\u4e00\u4e2atoken\u9700\u89815\u5206\u949f\uff0c\u800c\u73b0\u5728\u53ea\u9700\u89811\u5206\u949f\uff0c\u8fd9\u4e5f\u610f\u5473\u7740\u5b83\u4f1a\u66f4\u4fbf\u5b9c\uff0c\u56e0\u4e3a\u6ce8\u610f\u529b\u673a\u5236\u4e2d\u5b58\u5728\u7a00\u758f\u6027\uff0c\u4f46\u6211\u4eec\u53d1\u73b0\u6027\u80fd\u5e76\u6ca1\u6709\u4e0b\u964d\uff0c\u6216\u8005\u5728100\u4e07token\u7684\u4e0a\u4e0b\u6587\u4e2d\uff0c\u6211\u4eec\u5f97\u5230\u4e86\u4e00\u4e2a\u6027\u4ef7\u6bd4\u5f88\u9ad8\u7684\u9884\u671f\u6027\u80fd\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-89b3f7e4584edab9434282530227bf2b.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-89b3f7e4584edab9434282530227bf2b.png\"><\/a><\/p>\n<p>\u9664\u4e86\u7406\u89e3\u957f\u4e0a\u4e0b\u6587\u4e4b\u5916\uff0c\u8fd8\u6709\u53e6\u4e00\u79cd\u6269\u5c55\uff0c\u90a3\u5c31\u662f\u751f\u6210\u957f\u4e0a\u4e0b\u6587\u7684\u80fd\u529b\u3002\u4ee5\u524d\u6211\u4eec\u80fd\u591f\u751f\u6210 8k tokens\uff0c\u4f46\u4f60\u4f1a\u53d1\u73b0\u8fd9\u8fd8\u4e0d\u591f\u3002\u56e0\u4e3a\u73b0\u5728\u6211\u4eec\u6709\u957f\u94fe\u5f0f\u63a8\u7406\uff08Long Chain of Thought\uff09\uff0c\u6240\u4ee5\uff0c\u5b83\u4e4b\u524d\u8fd8\u4e0d\u8db3\u4ee5\u751f\u6210\u975e\u5e38\u3001\u975e\u5e38\u957f\u7684\u6587\u672c\uff0c\u4f46\u5982\u4eca\u60c5\u51b5\u5df2\u7ecf\u5927\u4e0d\u76f8\u540c\u4e86\u3002\u4e5f\u8bb8\u73b0\u5728\u5b83\u80fd\u591f\u751f\u6210\u50cf 12.8k \u4e2a token \u7684\u6587\u672c\uff0c\u8fd9\u5b9e\u9645\u4e0a\u662f\u53ef\u4ee5\u5b9e\u73b0\u7684\u3002\u5bf9\u4e8e\u76ee\u524d\u7684Qwen\u6765\u8bf4\uff0c\u5b83\u5b9e\u9645\u4e0a\u80fd\u591f\u751f\u6210\u5927\u7ea6 32k \u7684 token\u3002<\/p>\n<\/p>\n<\/p>\n<\/p>\n<p><span style=\"color: #FFFFFF\">\u96f7\u5cf0\u7f51\u96f7\u5cf0\u7f51\u96f7\u5cf0\u7f51<span style=\"color:#FD5D3C\">(\u516c\u4f17\u53f7\uff1a\u96f7\u5cf0\u7f51)<\/span><\/span><\/p>\n<p>\u96f7\u5cf0\u7f51\u539f\u521b\u6587\u7ae0\uff0c\u672a\u7ecf\u6388\u6743\u7981\u6b62\u8f6c\u8f7d\u3002\u8be6\u60c5\u89c1<a href=\"https:\/\/r.xiumi.us\/board\/v5\/3qhbI\/392977584\" rel=\"nofollow\" target=\"_blank\">\u8f6c\u8f7d\u987b\u77e5<\/a>\u3002<\/p>\n<p>\n                        <a href=\"https:\/\/vip.yanxishe.com\/?from=leiphonearticle\" target=\"_blank\"><br \/>\n                            <a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-ef1b38dfb738ad25d47c69ef4a8cea34.jpeg\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-ef1b38dfb738ad25d47c69ef4a8cea34.jpeg\"><\/a><br \/>\n                        <\/a>\n                    <\/p>\n<p>                                            \u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.leiphone.com\/category\/ai\/9wAXk0zd36CYiGlT.html\" target=\"_blank\">AI Agent \u53d1\u5c55\u53f2\uff1a\u4ece RL \u9a71\u52a8\u5230\u5927\u6a21\u578b\u9a71\u52a8 \uff5cAIR 2025<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Manus \u7684\u51fa\u73b0\u5c06\u667a\u80fd\u4f53\u63a8\u5165\u5f53\u4e0b AI [&hellip;]<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"","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":"","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":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","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":[27],"tags":[70,73,68,55],"class_list":["post-40026","post","type-post","status-publish","format-standard","hentry","category-news","tag-agent","tag-73","tag-68","tag-55"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.4 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>AI Agent \u53d1\u5c55\u53f2\uff1a\u4ece RL \u9a71\u52a8\u5230\u5927\u6a21\u578b\u9a71\u52a8 \uff5cAIR 2025 - \u4e00\u8d77AI\u6280\u672f<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/17aitech.com\/?p=40026\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/17aitech.com\/?p=40026\",\"url\":\"https:\/\/17aitech.com\/?p=40026\",\"name\":\"AI Agent \u53d1\u5c55\u53f2\uff1a\u4ece RL \u9a71\u52a8\u5230\u5927\u6a21\u578b\u9a71\u52a8 \uff5cAIR 2025 - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\/\/17aitech.com\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/17aitech.com\/?p=40026#primaryimage\"},\"image\":{\"@id\":\"https:\/\/17aitech.com\/?p=40026#primaryimage\"},\"thumbnailUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-eca3219bff8c15c37c232ffd7a987cc3.png\",\"datePublished\":\"2025-04-10T01:02:41+00:00\",\"author\":{\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\"},\"breadcrumb\":{\"@id\":\"https:\/\/17aitech.com\/?p=40026#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/17aitech.com\/?p=40026\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/?p=40026#primaryimage\",\"url\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-eca3219bff8c15c37c232ffd7a987cc3.png\",\"contentUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-eca3219bff8c15c37c232ffd7a987cc3.png\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/17aitech.com\/?p=40026#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/17aitech.com\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"AI Agent \u53d1\u5c55\u53f2\uff1a\u4ece RL \u9a71\u52a8\u5230\u5927\u6a21\u578b\u9a71\u52a8 \uff5cAIR 2025\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/17aitech.com\/#website\",\"url\":\"https:\/\/17aitech.com\/\",\"name\":\"\u4e00\u8d77AI\u6280\u672f\",\"description\":\"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca\",\"alternateName\":\"\u4e00\u8d77AI\u6280\u672f\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/17aitech.com\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\",\"name\":\"AI\u5c0f\u52a9\u624b\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/image\/\",\"url\":\"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png\",\"contentUrl\":\"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png\",\"caption\":\"AI\u5c0f\u52a9\u624b\"},\"description\":\"\u8fd9\u4e2a\u4eba\u5f88\u61d2\uff0c\u4ec0\u4e48\u90fd\u6ca1\u6709\u7559\u4e0b\uff5e\",\"url\":\"https:\/\/17aitech.com\/?page_id=33738&user=3\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"AI Agent \u53d1\u5c55\u53f2\uff1a\u4ece RL \u9a71\u52a8\u5230\u5927\u6a21\u578b\u9a71\u52a8 \uff5cAIR 2025 - \u4e00\u8d77AI\u6280\u672f","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/17aitech.com\/?p=40026","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=40026","url":"https:\/\/17aitech.com\/?p=40026","name":"AI Agent \u53d1\u5c55\u53f2\uff1a\u4ece RL \u9a71\u52a8\u5230\u5927\u6a21\u578b\u9a71\u52a8 \uff5cAIR 2025 - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=40026#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=40026#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-eca3219bff8c15c37c232ffd7a987cc3.png","datePublished":"2025-04-10T01:02:41+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=40026#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=40026"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=40026#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-eca3219bff8c15c37c232ffd7a987cc3.png","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/04\/frc-eca3219bff8c15c37c232ffd7a987cc3.png"},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=40026#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"AI Agent \u53d1\u5c55\u53f2\uff1a\u4ece RL \u9a71\u52a8\u5230\u5927\u6a21\u578b\u9a71\u52a8 \uff5cAIR 2025"}]},{"@type":"WebSite","@id":"https:\/\/17aitech.com\/#website","url":"https:\/\/17aitech.com\/","name":"\u4e00\u8d77AI\u6280\u672f","description":"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca","alternateName":"\u4e00\u8d77AI\u6280\u672f","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/17aitech.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d","name":"AI\u5c0f\u52a9\u624b","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/#\/schema\/person\/image\/","url":"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png","contentUrl":"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png","caption":"AI\u5c0f\u52a9\u624b"},"description":"\u8fd9\u4e2a\u4eba\u5f88\u61d2\uff0c\u4ec0\u4e48\u90fd\u6ca1\u6709\u7559\u4e0b\uff5e","url":"https:\/\/17aitech.com\/?page_id=33738&user=3"}]}},"_links":{"self":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/40026","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=40026"}],"version-history":[{"count":0,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/40026\/revisions"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=40026"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=40026"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=40026"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}