{"id":23006,"date":"2024-11-11T11:37:07","date_gmt":"2024-11-11T03:37:07","guid":{"rendered":"https:\/\/aif.amtbbs.org\/?p=23006"},"modified":"2024-11-11T11:37:07","modified_gmt":"2024-11-11T03:37:07","slug":"%e5%85%a8%e7%90%83%e9%a6%96%e4%b8%aa%e6%b3%95%e5%be%8bo1%e5%a4%a7%e6%a8%a1%e5%9e%8b%e5%8f%91%e5%b8%83%ef%bc%8csystem2%e8%8c%83%e5%bc%8f%e4%b8%8b%e6%85%a2%e6%80%9d%e8%80%83%e6%b3%95%e5%be%8b%e4%b8%93","status":"publish","type":"post","link":"https:\/\/aif.amtbbs.org\/index.php\/2024\/11\/11\/23006\/","title":{"rendered":"\u5168\u7403\u9996\u4e2a\u6cd5\u5f8bo1\u5927\u6a21\u578b\u53d1\u5e03\uff0cSystem2\u8303\u5f0f\u4e0b\u6162\u601d\u8003\u6cd5\u5f8b\u4e13\u5bb6\uff5c\u6e2f\u79d1\u5927&\u5317\u5927"},"content":{"rendered":"
\"\"<\/div>\n
<\/div>\n
HK-O1aw\uff0c\u662f\u7531\u9999\u6e2f\u751f\u6210\u5f0f\u4eba\u5de5\u667a\u80fd\u7814\u53d1\u4e2d\u5fc3(HKGAI)\u65d7\u4e0bAI for Reasoning\u56e2\u961f(HKAIR) \u8054\u5408\u5317\u4eac\u5927\u5b66\u5bf9\u9f50\u56e2\u961f\uff08PKU-Alignment Team\uff09\u63a8\u51fa\u7684\u5168\u7403\u9996\u4e2a\u6162\u601d\u8003\u8303\u5f0f\u6cd5\u5f8b\u63a8\u7406\u5927\u6a21\u578b\u3002<\/div>\n
\n
\n

OpenAI o1\u98ce\u683c\u7684\u63a8\u7406\u5927\u6a21\u578b\uff0c\u6709\u884c\u4e1a\u5782\u76f4\u7248\u4e86\u3002<\/p>\n

HK-O1aw\uff0c\u662f\u7531\u9999\u6e2f\u751f\u6210\u5f0f\u4eba\u5de5\u667a\u80fd\u7814\u53d1\u4e2d\u5fc3(HKGAI)\u65d7\u4e0bAI for Reasoning\u56e2\u961f(HKAIR) \u8054\u5408\u5317\u4eac\u5927\u5b66\u5bf9\u9f50\u56e2\u961f\uff08PKU-Alignment Team\uff09\u63a8\u51fa\u7684\u5168\u7403\u9996\u4e2a\u6162\u601d\u8003\u8303\u5f0f\u6cd5\u5f8b\u63a8\u7406\u5927\u6a21\u578b\u3002<\/p>\n

HK-O1aw\u7684\u63a8\u7406\u80fd\u529b\uff0c\u5728\u591a\u4e2a\u6cd5\u5f8b\u573a\u666f\u7684\u6d4b\u8bd5\u4e2d\u5f97\u5230\u8bc1\u660e\u3002\u8be5\u6a21\u578b\u5728\u5408\u540c\u6cd5\u3001\u6d88\u8d39\u8005\u6743\u76ca\u4fdd\u62a4\u6cd5\u7b49\u591a\u4e2a\u6cd5\u5f8b\u9886\u57df\u4e2d\u5747\u5c55\u73b0\u51fa\u8272\u7684\u8868\u73b0\uff0c\u5e76\u80fd\u7075\u6d3b\u9002\u5e94\u5404\u7c7b\u6cd5\u5f8b\u54a8\u8be2\u573a\u666f\u3002<\/p>\n

\u4ee5\u4e0b\u662fHK-O1aw\u6a21\u578b\u7528\u6237\u4f53\u9a8c\u6f14\u793a\uff1a<\/p>\n

\u63d0\u95ee1:\u62ff\u8863\u670d\u53bb\u5e72\u6d17\u540e\uff0c\u5e72\u6d17\u5e97\u5f04\u574f\u4e86\u8863\u670d\uff0c\u6839\u636e\u9999\u6e2f\u6cd5\u4f8b\uff0c\u53ef\u5426\u5411\u5e97\u94fa\u7d22\u507f\uff1f\u5e94\u8be5\u4ee5\u4ec0\u4e48\u6e20\u9053\u7d22\u4ef7\uff1f<\/p>\n

\u5728\u4e2a\u6848\u54a8\u8be2\u573a\u666f\u4e2d\uff0c\u9488\u5bf9\u201c\u5e72\u6d17\u5e97\u6d17\u574f\u8863\u670d\u201d\u7684\u60c5\u5f62\uff0c\u7528\u6237\u54a8\u8be2HK-O1aw\u5982\u4f55\u5229\u7528\u9999\u6e2f\u6cd5\u4f8b\u8fdb\u884c\u6d88\u8d39\u8005\u7ef4\u6743\u3002HK-O1aw\u901a\u8fc7\u201c\u5206\u6790\u9999\u6e2f\u6cd5\u5f8b\u6761\u4f8b\u2014\u2014\u5206\u6790\u7d22\u507f\u9014\u5f84\u2014\u2014\u5c0f\u989d\u94b1\u503a\u5ba1\u88c1\u5904\u7684\u9002\u7528\u6027\u2014\u2014\u6c11\u4e8b\u6cd5\u5ead\u7684\u8bc9\u8bbc\u7a0b\u5e8f\u201d\u5b8c\u6574\u7684\u903b\u8f91\u94fe\u6761\u8fdb\u884c\u601d\u8003\u3002\u6700\u7ec8\u63d0\u4f9b\u7528\u6237\u7b54\u6848\uff1a\u53ef\u4ee5\u8fdb\u884c\u7d22\u8d54\uff0c\u9996\u5148\u5efa\u8bae\u76f4\u63a5\u8054\u7cfb\u5e97\u94fa\u7d22\u8d54\uff0c\u82e5\u65e0\u679c\u5219\u5411\u6d88\u8d39\u8005\u59d4\u5458\u4f1a\u6295\u8bc9\uff0c\u6216\u901a\u8fc7\u6cd5\u5f8b\u9014\u5f84\u89e3\u51b3\u3002<\/p>\n

\"\u56fe\u7247\"<\/p>\n

\u63d0\u95ee2:\u5546\u4e1a\u4e3a\u4ec0\u4e48\u9700\u8981\u8c03\u89e3\u548c\u4ef2\u88c1\uff1f<\/strong><\/p>\n

\u5728\u6cd5\u5f8b\u77e5\u8bc6\u89e3\u7b54\u573a\u666f\u4e2d\uff0c\u9488\u5bf9\u7528\u6237\u63d0\u51fa\u201c\u5546\u4e1a\u4e3a\u4ec0\u4e48\u9700\u8981\u8c03\u89e3\u548c\u4ef2\u88c1\u201d\uff0cHK-O1aw\u7684\u63a8\u7406\u80fd\u529b\u5e2e\u52a9\u5b83\u628a\u95ee\u9898\u8fdb\u884c\u62c6\u5206\uff0c\u6309\u5546\u4e1a\u8c03\u89e3\u548c\u5546\u4e1a\u4ef2\u88c1\u7684\u5b9a\u4e49\u548c\u529f\u80fd\u3001\u8c03\u89e3\u548c\u4ef2\u88c1\u5404\u81ea\u7684\u4f18\u70b9\u3001\u9002\u7528\u6027\u9a8c\u8bc1\u7684\u601d\u8def\u8fdb\u884c\u6df1\u5ea6\u601d\u8003\uff0c\u5e76\u7ed9\u51fa\u6700\u7ec8\u7b54\u6848\u3002<\/p>\n

\"\u56fe\u7247\"<\/p>\n

\u63a8\u7406\u80fd\u529b\u7684\u5b9e\u73b0\uff0c\u8fdb\u4e00\u6b65\u4fdd\u969c\u4e86\u5927\u6a21\u578b\u5728\u6cd5\u5f8b\u573a\u666f\u4e0b\u6cd5\u5f8b\u6761\u4f8b\u5f15\u7528\u3001\u4e8b\u5b9e\u5224\u65ad\u3001\u51b3\u7b56\u610f\u89c1\u7684\u51c6\u786e\u6027\uff0c\u5206\u6790\u89d2\u5ea6\u7684\u5b8c\u5907\u6027\uff0c\u8ba9\u5927\u6a21\u578b\u5728\u6cd5\u5f8b\u670d\u52a1\u4e2d\u66f4\u52a0\u4e13\u4e1a\u53ef\u9760\u3002<\/p>\n

\u50cf\u8fd9\u6837\uff0c\u4e00\u76f4\u4ee5\u6765\u90fd\u662f\u8017\u65f6\u8017\u529b\u4e14\u6210\u672c\u9ad8\u6602\u6cd5\u5f8b\u7ea0\u7eb7\u5904\u7406\uff0c\u5982\u4eca\u53ef\u4ee5\u6162\u6162\u4ea4\u7ed9\u6cd5\u5f8b\u63a8\u7406\u5927\u6a21\u578b\u6765\u5e72\u4e86\u3002<\/p>\n

\u6df1\u5ea6\u63a8\u7406\uff1aHK-O1aw\u6cd5\u5f8b\u5927\u6a21\u578b\u7684\u6838\u5fc3\u7ade\u4e89\u529b<\/h4>\n

\u65e9\u671f\u5e02\u573a\u4e0a\u7684\u6cd5\u5f8b\u5927\u6a21\u578b\u5728\u6df1\u5ea6\u63a8\u7406\u80fd\u529b\u4e0a\u6709\u6240\u6b20\u7f3a\u3002<\/p>\n

\u65e9\u671f\u7684\u6cd5\u5f8b\u5927\u6a21\u578b\u5373\u4fbf\u80fd\u591f\u63d0\u4f9b\u4e30\u5bcc\u7684\u6cd5\u5f8b\u77e5\u8bc6\uff0c\u4f46\u9762\u5bf9\u73b0\u5b9e\u793e\u4f1a\u4e2d\u7684\u590d\u6742\u6cd5\u5f8b\u7ea0\u7eb7\uff0c\u5b83\u4eec\u5f80\u5f80\u5728\u5173\u952e\u903b\u8f91\u4e0a\u51fa\u73b0\u758f\u6f0f\uff0c\u6700\u7ec8\u96be\u4ee5\u89e3\u51b3\u6cd5\u5f8b\u573a\u666f\u4e0b\u7684\u5b9e\u9645\u95ee\u9898\u3002<\/p>\n

HK-O1aw\u901a\u8fc7\u91c7\u7528O1\u98ce\u683c\u7684\u6570\u636e\u96c6\u548c\u8bad\u7ec3\u8bbe\u7f6e\uff0c\u5b9e\u73b0\u6cd5\u5f8b\u6a21\u578b\u6df1\u5ea6\u63a8\u7406\u80fd\u529b\u7684\u8dc3\u5347\uff0c\u8fd9\u4f7f\u5f97\u6cd5\u5f8b\u5927\u6a21\u578b\u80fd\u591f\u6df1\u5165\u5206\u6790\u590d\u6742\u7684\u6cd5\u5f8b\u6587\u672c\u548c\u6848\u4f8b\uff0c\u63d0\u4f9b\u903b\u8f91\u7f1c\u5bc6\u7684\u6cd5\u5f8b\u610f\u89c1\uff0c\u6ee1\u8db3\u884c\u4e1a\u5bf9\u6df1\u5ea6\u5206\u6790\u548c\u590d\u6742\u95ee\u9898\u89e3\u51b3\u7684\u8feb\u5207\u9700\u6c42\u3002<\/p>\n

O1\u98ce\u683c\u662f\u4ec0\u4e48\uff1f<\/h4>\n

\u5b83\u6307\u7684\u662f\u4e00\u79cd\u7c7b\u4f3c\u4e8eOpenAI o1\u6a21\u578b\u7684\u957f\u8003\u80fd\u529b\uff0c\u5b83\u4fa7\u91cd\u4e8e\u6162\u601d\u8003\uff08slow thinking\uff09\u548c\u94fe\u5f0f\u63a8\u7406\uff08Chain of Thought, CoT\uff09\u3002\u5728\u6cd5\u5f8b\u9886\u57df\uff0cO1\u98ce\u683c\u4e13\u6ce8\u4e8e\u6df1\u5165\u5206\u6790\u548c\u63a8\u7406\uff0c\u4ee5\u5e94\u5bf9\u6cd5\u5f8b\u6587\u672c\u548c\u6848\u4f8b\u7684\u590d\u6742\u6027\u3002\u8fd9\u79cd\u98ce\u683c\u7684\u6a21\u578b\u6280\u672f\u96be\u70b9\u4e3b\u8981\u4f53\u73b0\u5728\u4ee5\u4e0b\u51e0\u4e2a\u65b9\u9762\uff1a<\/p>\n

    \n
  • \u590d\u6742\u6027\u5904\u7406\uff1a\u6cd5\u5f8b\u6587\u672c\u5e38\u5305\u542b\u4e13\u4e1a\u672f\u8bed\u548c\u590d\u6742\u7684\u903b\u8f91\u7ed3\u6784\uff0cO1\u98ce\u683c\u7684\u6a21\u578b\u5fc5\u987b\u80fd\u591f\u8fdb\u884c\u6df1\u5165\u7684\u601d\u8003\u548c\u63a8\u7406\u3002<\/li>\n
  • \u957f\u94fe\u63a8\u7406\uff1a\u4e3a\u4e86\u5728\u590d\u6742\u6848\u4ef6\u4e2d\u8fdb\u884c\u5168\u9762\u5206\u6790\u548c\u5224\u65ad\uff0cO1\u98ce\u683c\u7684\u6a21\u578b\u5fc5\u987b\u80fd\u591f\u652f\u6301\u66f4\u957f\u7684\u63a8\u7406\u94fe\u6761\u3002<\/li>\n
  • \u51c6\u786e\u6027\u548c\u53ef\u9760\u6027\uff1aO1\u98ce\u683c\u7684\u6a21\u578b\u5fc5\u987b\u786e\u4fdd\u6a21\u578b\u5728\u63a8\u7406\u8fc7\u7a0b\u4e2d\u7684\u4e00\u81f4\u6027\u548c\u53ef\u9760\u6027\uff0c\u4ee5\u907f\u514d\u9519\u8bef\u5224\u65ad\u3002<\/li>\n
  • \u53ef\u89e3\u91ca\u6027\uff1aO1\u98ce\u683c\u7684\u6a21\u578b\u9700\u8981\u63d0\u4f9b\u53ef\u89e3\u91ca\u7684\u6a21\u578b\u8f93\u51fa\uff0c\u4ee5\u4fbf\u6cd5\u5f8b\u4e13\u4e1a\u4eba\u58eb\u7406\u89e3\u548c\u4fe1\u4efb\u6a21\u578b\u7684\u7ed3\u8bba\u3002<\/li>\n<\/ul>\n

    \u60f3\u8981\u8bad\u7ec3\u51fa\u903b\u8f91\u8d85\u5f3a\u7684\u6a21\u578b\uff0c\u4e30\u5bcc\u5e76\u4e14\u9ad8\u8d28\u91cf\u7684\u6570\u636e\u96c6\u5fc5\u4e0d\u53ef\u5c11\u3002<\/p>\n

    \u4f9d\u6258\u4e8e\u7ecf\u8fc7\u4e25\u683c\u6e05\u6d17\u548c\u9884\u5904\u7406\u7684\u6cd5\u5f8b\u6750\u6599\uff0cHKAIR\u56e2\u961f\u4e3aHK-O1aw\u6784\u5efa\u4e8615959\u4e2a\u5305\u542b\u201c\u95ee\u9898-\u601d\u8003-\u7b54\u6848\u201d\u7684\u4e09\u5143\u7ec4\uff0c\u5168\u90e8\u4ee5\u7b80\u4f53\u4e2d\u6587\u5448\u73b0\uff0c\u5e76\u4ee5\u7ed3\u6784\u5316\u7684JSON\u683c\u5f0f\u5b58\u50a8\uff0c\u5e76\u5728Hugging Face\u5e73\u53f0\u4e0a\u5f00\u6e90\uff08\u8be6\u89c1\u6587\u672b\u94fe\u63a5\uff09\u3002\u8fd9\u79cd\u8bbe\u8ba1\u65e8\u5728\u6a21\u62df\u6cd5\u5f8b\u4e13\u4e1a\u4eba\u58eb\u7684\u601d\u8003\u8fc7\u7a0b\uff0c\u589e\u5f3a\u6cd5\u5f8b\u5927\u6a21\u578b\u7684\u6df1\u5165\u5206\u6790\u80fd\u529b\u3002<\/p>\n

    \u6570\u636e\u96c6\u4e2d\u7684\u95ee\u9898\u7c7b\u578b\u5e7f\u6cdb\uff0c\u6db5\u76d6\u6848\u4f8b\u5206\u6790\u3001\u6cd5\u5f8b\u5e94\u7528\u548c\u6cd5\u5f8b\u6982\u5ff5\u89e3\u91ca\u7b49\u591a\u4e2a\u65b9\u9762\uff0c\u96be\u5ea6\u7ea7\u522b\u4e3a\u4e2d\u7ea7\u5230\u9ad8\u7ea7\uff0c\u5bf9\u4e8e\u6cd5\u5f8b\u4e13\u4e1a\u4eba\u58eb\u548c\u6cd5\u5b66\u9662\u5b66\u751f\u6765\u8bf4\u5177\u6709\u6311\u6218\u6027\u3002<\/p>\n

    \"\u56fe\u7247\"<\/p>\n

    HKGAI\u56e2\u961f\u4ecb\u7ecd<\/h4>\n

    HKAIR Lab\u96b6\u5c5e\u4e8e\u9999\u6e2f\u751f\u6210\u5f0f\u4eba\u5de5\u667a\u80fd\u7814\u53d1\u4e2d\u5fc3\uff08HKGAI\uff09\u3002HKAIR Lab \u7531\u9999\u6e2f\u79d1\u6280\u5927\u5b66\u9996\u5e2d\u526f\u6821\u957f\u90ed\u6bc5\u53ef\u6559\u6388\u548c\u8de8\u5b66\u79d1\u5b66\u9662\u97e9\u65af\u777f\u52a9\u7406\u6559\u6388\u6307\u5bfc\uff0c\u6210\u5458\u4e3b\u8981\u7531\u9999\u6e2f\u79d1\u6280\u5927\u5b66\u8de8\u5b66\u79d1\u5b66\u9662\u7684\u535a\u58eb\u7814\u7a76\u751f\u548c\u7814\u7a76\u5458\u56e2\u961f\u7ec4\u6210\u3002HKAIR Lab \u4e0e\u5317\u5927\u5bf9\u9f50\u56e2\u961f\u5728 HK-O1aw \u548c\u201c\u6162\u601d\u8003\u201d\u8303\u5f0f\u4e0b\uff0c\u5c31\u751f\u6210\u5f0f\u4eba\u5de5\u667a\u80fd\u5982\u4f55\u8d4b\u80fd\u6cd5\u5f8b\u5e94\u7528\u573a\u666f\u5c55\u5f00\u4e86\u5408\u4f5c\u3002<\/p>\n

    \u4e3a\u4e86\u628a\u63e1\u672a\u6765\u673a\u9047\u5e76\u5e94\u5bf9\u6311\u6218\uff0cHKAIR\u7814\u7a76\u56e2\u961f\u5c06\u6301\u7eed\u5bf9HK-O1aw\u8fdb\u884c\u4f18\u5316\u3002\u8ba1\u5212\u4e2d\u5305\u62ec\u5f15\u5165\u6cd5\u5f8b\u9886\u57df\u4e13\u5bb6\u53c2\u4e0e\u6a21\u578b\u7684\u8bad\u7ec3\u4e0e\u8bc4\u4f30\uff0c\u8fd9\u5c06\u4fc3\u8fdb\u6a21\u578b\u5bf9\u6cd5\u5f8b\u6982\u5ff5\u7684\u6df1\u5165\u7406\u89e3\uff0c\u63d0\u5347\u5176\u63a8\u7406\u80fd\u529b\uff0c\u5e76\u62d3\u5bbd\u6a21\u578b\u5728\u591a\u6837\u5316\u6cd5\u5f8b\u573a\u666f\u4e2d\u7684\u5e94\u7528\u8303\u56f4\u3002<\/p>\n

    HK-O1aw\u6a21\u578b\u5f00\u6e90\u8d44\u6e90\u94fe\u63a5\uff1a
    \n\u6a21\u578b\u5730\u5740\uff1ahttps:\/\/huggingface.co\/HKAIR-Lab\/HK-O1aw\u3002
    \n\u6570\u636e\u96c6\u5730\u5740\uff1ahttps:\/\/huggingface.co\/datasets\/HKAIR-Lab\/HK-O1aw-SFT-16K\u3002
    \n\u4ee3\u7801\u5730\u5740\uff1ahttps:\/\/github.com\/HKAIR-Lab\/HK-O1aw\/\u3002<\/p>\n<\/div>\n

    \u6587\u7ae0\u6765\u81ea\uff1a51CTO<\/p>\n<\/div>\n

    <\/div>\n

    <\/g><\/svg><\/i> \"Loading\"<\/p>\n

    <\/div>\n","protected":false},"excerpt":{"rendered":"

    HK-O1aw\uff0c\u662f\u7531\u9999\u6e2f\u751f\u6210\u5f0f\u4eba\u5de5\u667a\u80fd\u7814\u53d1\u4e2d\u5fc3(HKGAI)\u65d7\u4e0bAI for Reasoning\u56e2\u961f(HKAI […]<\/p>\n

    <\/div>\n

    <\/g><\/svg><\/i> \"Loading\"<\/p>\n

    <\/div>\n","protected":false},"author":56,"featured_media":23008,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[23,20,80],"tags":[1421,667,1422],"class_list":["post-23006","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-23","category-20","category-80","tag-hk-o1aw-","tag-667","tag-1422"],"_links":{"self":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/23006","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/users\/56"}],"replies":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/comments?post=23006"}],"version-history":[{"count":1,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/23006\/revisions"}],"predecessor-version":[{"id":23011,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/23006\/revisions\/23011"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/media\/23008"}],"wp:attachment":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/media?parent=23006"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/categories?post=23006"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/tags?post=23006"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}