{"id":40306,"date":"2025-05-22T21:02:59","date_gmt":"2025-05-22T13:02:59","guid":{"rendered":"https:\/\/17aitech.com\/?p=40306"},"modified":"2025-05-22T21:02:59","modified_gmt":"2025-05-22T13:02:59","slug":"%e7%ba%af%e8%92%b8%e9%a6%8f%e6%a8%a1%e5%9e%8b-sota-%e5%87%ba%e7%8e%b0%ef%bc%81%e7%9b%b4%e6%8e%a5-sft-%e6%88%90%e6%9c%ac%e7%9b%b4%e9%99%8d-50-%e5%80%8d%ef%bc%8c%e6%95%b0%e6%8d%ae%e5%b7%b2%e5%85%a8","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=40306","title":{"rendered":"\u7eaf\u84b8\u998f\u6a21\u578b SOTA \u51fa\u73b0\uff01\u76f4\u63a5 SFT \u6210\u672c\u76f4\u964d 50 \u500d\uff0c\u6570\u636e\u5df2\u5168\u90e8\u5f00\u6e90"},"content":{"rendered":"<p><\/p>\n<p>a-m-team \u53c8\u53d1\u65b0\u8bba\u6587\u4e86\u3002<\/p>\n<p>\u8fd9\u4e2a\u56e2\u961f\u4e0a\u5468\u521a\u521a\u5728 Hugging Face \u4f4e\u8c03\u5f00\u6e90\u4e8632B\u7a20\u5bc6\u6a21\u578b\uff0c\u4f46\u5728\u591a\u9879\u5173\u952e\u63a8\u7406\u8bc4\u6d4b\u4e2d\u51fb\u8d25\u4e86 DeepSeek-R1\uff0c\u5e76\u4e0e\u8d85\u5927\u89c4\u6a21\u7684 MoE \u6a21\u578bQwen3-235B-A22B\u3001Seed1.5-Thinking \u4e0d\u76f8\u4e0a\u4e0b\uff0c\u56e0\u6b64\u8d62\u5f97\u4e86\u6d77\u5185\u5916\u7684\u4e0d\u5c11\u5173\u6ce8\u3002<\/p>\n<p>\u4eca\u5929\uff0ca-m-team \u53d1\u5e03\u4e86\u4e00\u7bc7\u540d\u4e3a\u201cNot All Correct Answers Are Equal: Why Your Distillation Source Matters\u201d\u7684\u8bba\u6587\uff0c\u4ecb\u7ecd\u4e86 R1 \u4e4b\u540e\u4e0b\u4e00\u4ee3\u63a8\u7406\u6a21\u578b\u7684\u5c0f\u8fdb\u5c55\u3002<\/p>\n<p>\u5728\u8fd9\u7bc7\u8bba\u6587\u4e2d\uff0c\u7814\u7a76\u56e2\u961f\u53d1\u73b0\u57fa\u4e8e AM-Thinking-v1 \u7684\u95ee\u7b54\u6570\u636e\u201c\u7eaf\u84b8\u998f\u201d\uff08\u5373\u53ea\u7528SFT\uff09\u8bad\u7ec3\u51fa\u7684\u201c\u5b66\u751f\u6a21\u578b\u201d\uff0c\u7adf\u7136\u5728\u591a\u4e2a\u9ad8\u96be\u63a8\u7406\u4efb\u52a1\u4e0a\u63a5\u8fd1\u751a\u81f3\u8fbe\u5230\u5f53\u524d\u6700\u4f18\u6c34\u5e73\uff08SOTA\uff09\uff0c\u4e0d\u4ec5\u8d85\u8fc7\u4e86Qwen3-32B\uff0c\u751a\u81f3\u63a5\u8fd1\u4e86 Qwen3-235B \u8fd9\u6837\u5927\u4e00\u4e2a\u6570\u91cf\u7ea7\u7684\u6a21\u578b\u8868\u73b0\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3387c20d40788ad9020506d9b326b2c1.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3387c20d40788ad9020506d9b326b2c1.png\"><\/a><\/p>\n<p style=\"text-align: right\">\u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/pdf\/2505.14464<\/p>\n<p>\u5728\u5927\u6a21\u578b\u80fd\u529b\u7ade\u8d5b\u6301\u7eed\u63a8\u8fdb\u7684\u4eca\u5929\uff0c\u5982\u4f55\u4ee5\u66f4\u4f4e\u7684\u8bad\u7ec3\u6210\u672c\u83b7\u5f97\u66f4\u5f3a\u7684\u63a8\u7406\u80fd\u529b\uff0c\u6210\u4e3a\u5f00\u6e90\u793e\u533a\u9762\u4e34\u7684\u6838\u5fc3\u6311\u6218\u4e4b\u4e00\u3002<\/p>\n<p>\u76f8\u6bd4\u4e8e SFT+RL \u7684\u8bad\u7ec3\u65b9\u5f0f\uff0c\u53ea\u7528 SFT \u610f\u5473\u7740\u7740\u6570\u5341\u500d\u7684\u6210\u672c\u7f29\u51cf\u2014\u2014\u4e5f\u5c31\u662f\u8bf4\uff0c\u6ca1\u6709\u5f3a\u5316\u5b66\u4e60\u80fd\u529b\u7684\u4f01\u4e1a\u7ea7\u73a9\u5bb6\u5c06 AM-Thinking-v1 \u84b8\u998f\u7248\u62ff\u53bb\u76f4\u63a5 SFT \u540e\u843d\u5730\uff0c\u80fd\u591f\u5927\u5e45\u63d0\u5347\u5e94\u7528\u6548\u7387\u3001\u964d\u4f4e\u843d\u5730\u6210\u672c\u3002<\/p>\n<p>\u8fd9\u610f\u5473\u7740\u5f00\u6e90\u793e\u533a\u672a\u6765\u80fd\u4ee5\u66f4\u4f4e\u7684\u8bad\u7ec3\u6210\u672c\uff0c\u83b7\u5f97\u66f4\u5f3a\u7684\u63a8\u7406\u80fd\u529b\u3002\u800c\u8fd9\u9879\u5de5\u4f5c\u6210\u7acb\u7684\u524d\u63d0\u6761\u4ef6\u662f\uff0c\u9700\u8981\u80fd\u6709\u826f\u597d\u7684\u6570\u636e\u6e90\u3002<\/p>\n<p>\u6362\u8a00\u4e4b\uff0c\u5047\u5982\u6570\u636e\u6e90\u80fd\u9a71\u52a8\u84b8\u998f\u6a21\u578b\u7684\u6210\u957f\uff0c\u90a3\u4e48\u84b8\u998f\u4e5f\u5c06\u4e0d\u53ea\u662f\u4e00\u4e2a\u7b80\u7b80\u5355\u5355\u7684\u667a\u80fd\u201c\u538b\u7f29\u201d\u52a8\u4f5c\uff0c\u800c\u662f\u4f1a\u6709\u6f5c\u529b\u6210\u4e3a\u5728\u5f00\u6e90\u793e\u533a\u751f\u6001\u4e2d\u87ba\u65cb\u751f\u957f\u7684\u7cfb\u7edf\u6027\u5de5\u7a0b\u3002<\/p>\n<p><\/p>\n<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_78 ez-toc-wrap-left-text counter-hierarchy ez-toc-counter ez-toc-light-blue ez-toc-container-direction\">\n<div class=\"ez-toc-title-container\">\n<p class=\"ez-toc-title\" style=\"cursor:inherit\">\u6587\u7ae0\u76ee\u5f55<\/p>\n<span class=\"ez-toc-title-toggle\"><a href=\"#\" class=\"ez-toc-pull-right ez-toc-btn ez-toc-btn-xs ez-toc-btn-default ez-toc-toggle\" aria-label=\"Toggle Table of Content\"><span class=\"ez-toc-js-icon-con\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #999;color:#999\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #999;color:#999\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/span><\/a><\/span><\/div>\n<nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-1'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/17aitech.com\/?p=40306\/#%E4%BB%80%E4%B9%88%E6%A0%B7%E7%9A%84%E6%95%B0%E6%8D%AE%E6%BA%90%E6%9B%B4%E6%9C%89%E6%95%88%EF%BC%9F\" >\u4ec0\u4e48\u6837\u7684\u6570\u636e\u6e90\u66f4\u6709\u6548\uff1f<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-1'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/17aitech.com\/?p=40306\/#%E4%BB%8E%E4%B8%8D%E6%9C%8D%E8%B7%91%E4%B8%AA%E5%88%86%EF%BC%8C%E5%88%B0%E4%B8%8D%E6%9C%8D%E8%92%B8%E4%B8%AA%E9%A6%8F%EF%BC%9F\" >\u4ece\u4e0d\u670d\u8dd1\u4e2a\u5206\uff0c\u5230\u4e0d\u670d\u84b8\u4e2a\u998f\uff1f<\/a><\/li><\/ul><\/nav><\/div>\n<h1 style=\"font-size: 32px;font-weight: bold;padding: 0px 4px 0px 0px;text-align: center;margin: 0px 0px 20px\"><span class=\"ez-toc-section\" id=\"%E4%BB%80%E4%B9%88%E6%A0%B7%E7%9A%84%E6%95%B0%E6%8D%AE%E6%BA%90%E6%9B%B4%E6%9C%89%E6%95%88%EF%BC%9F\"><\/span>\u4ec0\u4e48\u6837\u7684\u6570\u636e\u6e90\u66f4\u6709\u6548\uff1f<span class=\"ez-toc-section-end\"><\/span><\/h1>\n<p><\/p>\n<p>\u84b8\u998f\uff08Distillation\uff09\u4f5c\u4e3a\u4e00\u79cd\u4f4e\u6210\u672c\u3001\u9ad8\u6548\u7387\u7684\u8bad\u7ec3\u65b9\u5f0f\uff0c\u5df2\u88ab\u5e7f\u6cdb\u7528\u4e8e\u6a21\u578b\u538b\u7f29\u4e0e\u80fd\u529b\u8fc1\u79fb\u3002\u4e00\u822c\u60c5\u51b5\u4e0b\uff0c\u57fa\u4e8eGRPO \u6216\u8005 PPO \u7684 RL \u8bad\u7ec3\uff0c\u4f46\u4e00\u4e2a\u88ab\u957f\u671f\u5ffd\u89c6\u7684\u95ee\u9898\u662f\uff1a\u4f60\u7684\u84b8\u998f\u6e90\u9009\u5bf9\u4e86\u5417\uff1f<\/p>\n<p>a-m-team \u8fd1\u671f\u5f00\u6e90\u4e86\u4e00\u5957\u57fa\u4e8e AM-Thinking-v1 \u548c Qwen3-235B-A22B \u4e24\u4e2a\u9876\u5c16\u5927\u6a21\u578b\u751f\u6210\u7684\u63a8\u7406\u84b8\u998f\u6570\u636e\u96c6\u3002\u901a\u8fc7\u5bf9 189 \u4e07\u6761\u9ad8\u8d28\u91cf\u63a8\u7406\u4efb\u52a1\u7684\u5e76\u884c\u84b8\u998f\u5bf9\u6bd4\u5206\u6790\u53d1\u73b0\uff1a<\/p>\n<p>\u57fa\u4e8e AM-Thinking-v1 \u84b8\u998f\u8bad\u7ec3\u51fa\u7684\u5b66\u751f\u6a21\u578b\u5728\u591a\u4e2a\u9ad8\u96be\u63a8\u7406\u4efb\u52a1\u4e0a\u63a5\u8fd1\u751a\u81f3\u8fbe\u5230\u5f53\u524d\u6700\u4f18\u6c34\u5e73\uff08SOTA\uff09\uff1b<\/p>\n<p>\u8fd9\u4efd\u5f00\u6e90\u6570\u636e\u96c6\u4e3a\u4f4e\u6210\u672c\u6784\u5efa\u5f3a\u63a8\u7406\u80fd\u529b\u6a21\u578b\u63d0\u4f9b\u4e86\u624e\u5b9e\u652f\u6491\uff1b<\/p>\n<p>\u7528\u540c\u6837\u7684\u5b66\u751f\u6a21\u578b\u3001\u8bad\u7ec3\u914d\u7f6e\u3001benchmark \u5957\u4ef6\u5bf9\u6bd4\u8bad\u7ec3\u7ed3\u679c\uff1b<\/p>\n<p>\u5f00\u6e90\u5176\u4e2d\u4e24\u4efd\uff08AM-Thinking-v1 \u548c Qwen3-235B-A22B\uff09\u7684\u6570\u636e\u4ee5\u4f9b\u793e\u533a\u9a8c\u8bc1\u3002<\/p>\n<p>\u6b63\u5982\u672c\u7bc7\u8bba\u6587\u4e00\u8bed\u53cc\u5173\u7684\u4e3b\u6807\u9898\uff0c\u201cNot All Correct Answers Are Equal\u201d\u2014\u2014\u4e0d\u540c\u7684\u6a21\u578b\uff0c\u5982\u679c\u751f\u6210\u76f8\u4f3c\u6b63\u786e\u7684\u7b54\u6848\uff0c\u4f46\u5e76\u4e0d\u4ee3\u8868\u4ed6\u4eec\u90fd\u6709\u76f8\u540c\u7684\u4ef7\u503c\u3002\u56e0\u4e3a\u57fa\u4e8e\u8fd9\u4e2a\u6b63\u786e\u7b54\u6848\u80cc\u540e\u7684\u6570\u636e\u6e90\u8d28\u91cf\u3001\u7ed3\u6784\u5b58\u5728\u5dee\u5f02\uff0c\u81ea\u7136\u5bf9\u540e\u7eed\u8bad\u7ec3\u7684\u8d21\u732e\u53ef\u80fd\u5b58\u5728\u663e\u8457\u5dee\u5f02\u3002<\/p>\n<p>a-m-team\u7814\u7a76\u4ee5\u81ea\u5bb6\u6700\u65b0\u53d1\u5e03\u7684AM-Thinking-v1\u4e3a\u4f8b\uff0c\u6839\u636e\u5176\u56e2\u961f\u7814\u7a76\uff0c\u4ee5AM-Thinking-v1\u4e3a\u6559\u5e08\u6a21\u578b\u57f9\u8bad\u51fa\u6765\u7684\u7eaf\u84b8\u998f\u6a21\u578b\uff0c\u76f8\u6bd4\u4e8eQwen3-235B-A22\u548cDeepSeek-R1\u7684\u84b8\u998f\u6a21\u578b\u5728\u6027\u80fd\u4e0a\u6709\u7740\u663e\u8457\u7684\u9886\u5148\u4f18\u52bf\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-b27f623b64fcd2bb075da22b040b1fdc.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-b27f623b64fcd2bb075da22b040b1fdc.png\"><\/a><\/p>\n<p>\u4e0d\u4ec5\u5982\u6b64\uff0cAM\u84b8\u998f\u8fc7\u7a0b\u4e2d\u7684\u635f\u5931\u66f2\u7ebf\u4e5f\u662f\u6700\u4f4e\u7684\uff0c\u8fd9\u70b9\u5176\u5b9e\u4ece\u56fe\u4e00\u7684\u8bc4\u5206\u6392\u5217\u4e2d\u4e5f\u53ef\u89c1\u4e00\u6591\u3002\u5982\u4e0b\u56fe\u6240\u793a\uff0cAM\u957f\u671f\u4fdd\u6301\u4e86\u8fdc\u4f4e\u4e8e\u5176\u4ed6\u6a21\u578b\u7684\u57fa\u51c6\u635f\u8017\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-8925ea48bbdf13e0e0fdf49c2d81a600.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-8925ea48bbdf13e0e0fdf49c2d81a600.png\"><\/a><\/p>\n<p>\u635f\u8017\u4f4e\u7684\u80cc\u540e\u662f AM \u56e2\u961f\u672c\u8eab\u63d0\u4f9b\u4e86\u66f4\u9002\u5408\u5b66\u4e60\u7684\u9ad8\u8d28\u91cf\u7684\u6570\u636e\u3002<\/p>\n<p>a-m-team \u56e2\u961f\u5bf9\u4ece\u4e09\u4e2a\u4e0d\u540c\u7684\u5927\u89c4\u6a21\u6a21\u578b AM-Thinking-v1\u3001Qwen3-235B-A22B\u548c DeepSeek-R1\u7684\u8bad\u7ec3\u6570\u636e\u8fdb\u884c\u4e86\u8be6\u7ec6\u5206\u6790\u540e\uff0c\u8ba4\u4e3a\u8fd9\u79cd\u5206\u6570\u5dee\u5f02\u6765\u6216\u8bb8\u81ea\u4e8e\u4ed6\u4eec\u5728\u6570\u636e\u7ed3\u6784\u4e0a\u7684\u4f18\u52bf\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-8869919ea1b97a6ab2f25c93dc63a06f.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-8869919ea1b97a6ab2f25c93dc63a06f.png\"><\/a><\/p>\n<p>\u9996\u5148\uff0cAM\u7684token \u957f\u5ea6\u5206\u5e03\u66f4\u52a0\u591a\u6837\u3002<\/p>\n<p>\u5982\u4e0b\u56fe\u4e13\u95e8\u5c55\u793a\u4e86\u6570\u5b66\u5b9e\u4f8b\u7684\u8bcd\u5143\u8de8\u5ea6\u5206\u5e03\u60c5\u51b5\u3002\u7ed3\u679c\u8868\u660e\uff0cAM-Thinking-v1 \u7684\u6570\u5b66\u6570\u636e\u5448\u73b0\u51fa\u9ad8\u5ea6\u5206\u6563\u7684\u5206\u5e03\u72b6\u6001\uff0c\u5448\u73b0\u4e86\u66f4\u591a\u7684\u77ed\u5e8f\u5217\u3002<\/p>\n<p>\u8fd9\u610f\u5473\u7740\uff0cAM\u7684\u54cd\u5e94\u8de8\u5ea6\u5e7f\u2014\u2014\u5b83\u65e2\u80fd\u751f\u6210\u7b80\u6d01\u76841024 token\u4ee5\u5185\u56de\u590d\uff0c\u4e5f\u80fd\u8f93\u51fa\u8d85\u8fc7 10240 token \u7684\u590d\u6742\u63a8\u7406\u94fe\uff0c\u8fd9\u79cd\u201c\u957f\u77ed\u7ed3\u5408\u201d\u7684\u5206\u5e03\u4e3a\u6a21\u578b\u7684\u81ea\u9002\u5e94\u80fd\u529b\u63d0\u4f9b\u4e86\u6570\u636e\u652f\u6491\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-ff6834151233b07bcb67e45986536e56.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-ff6834151233b07bcb67e45986536e56.png\"><\/a><\/p>\n<p>\u5176\u6b21\uff0cAM \u6a21\u578b\u6570\u636e\u6e90\u7684\u56f0\u60d1\u5ea6\u66f4\u4f4e\uff0c\u6570\u636e\u8d28\u91cf\u66f4\u9ad8\u3002<\/p>\n<p>\u7814\u7a76\u79f0\uff0cAM \u7684\u5e73\u5747Perplexity(PPL)\u6570\u503c\u4e3a2.5\uff0c\u4f4e\u4e8eQwen3.0\u548cDeepSeek R1 2.9\uff0c\u8868\u660e AM \u7684\u6570\u636e\u5728\u8bed\u8a00\u6d41\u7545\u6027\u3001\u8fde\u8d2f\u6027\u7b49\u65b9\u9762\u66f4\u4f18\uff0c\u8fd9\u6216\u8bb8\u4ee3\u8868\u5176\u9002\u5408\u7528\u4e8e\u8bad\u7ec3\u7ed3\u6784\u5316 reasoning \u8fc7\u7a0b\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3e5a5e9d49ba67bffa3f93e98b96c18f.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3e5a5e9d49ba67bffa3f93e98b96c18f.png\"><\/a><\/p>\n<p>\u800c\u7531\u4e8e\u66f4\u597d\u7684\u6570\u636e\u7ed3\u6784\uff0c\u9664\u4e86\u8dd1\u5206\u9886\u5148\u5916\uff0c\u751f\u6210\u884c\u4e3a\u4e5f\u66f4\u7b26\u5408\u201c\u4eba\u7c7b\u63a8\u7406\u6a21\u5f0f\u201d\u3002<\/p>\n<p>\u7814\u7a76\u8005\u58f0\u79f0\uff0c\u5728\u8f83\u96be\u7684\u4efb\u52a1\u4e2d\uff08\u5982\u4e0b\u56feAIME20242025\uff09\uff0cAM \u84b8\u998f\u6a21\u578b\u4f1a\u81ea\u52a8\u751f\u6210\u66f4\u957f reasoning \u8fc7\u7a0b\uff0c\u800c\u5728\u8f83\u7b80\u5355\u4efb\u52a1\uff08\u5982\u4e0b\u56fe\u4e2d MATH500\uff09\u4e2d\u5219\u4fdd\u6301\u7b80\u660e\u3002<\/p>\n<p>\u8fd9\u610f\u5473\u7740\u6a21\u578b\u4fdd\u6301\u4e86\u4e00\u4e2a\u8f83\u597d\u7684\u957f\u8c03\u5ea6\u80fd\u529b\u3002<\/p>\n<p style=\"text-align: center\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-85ee14e1698b49f7f9d9761bc4793290.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-85ee14e1698b49f7f9d9761bc4793290.png\"><\/a><\/p>\n<h1 style=\"font-size: 32px;font-weight: bold;padding: 0px 4px 0px 0px;text-align: center;margin: 0px 0px 20px\"><span class=\"ez-toc-section\" id=\"%E4%BB%8E%E4%B8%8D%E6%9C%8D%E8%B7%91%E4%B8%AA%E5%88%86%EF%BC%8C%E5%88%B0%E4%B8%8D%E6%9C%8D%E8%92%B8%E4%B8%AA%E9%A6%8F%EF%BC%9F\"><\/span>\n<br \/>\u4ece\u4e0d\u670d\u8dd1\u4e2a\u5206\uff0c\u5230\u4e0d\u670d\u84b8\u4e2a\u998f\uff1f<span class=\"ez-toc-section-end\"><\/span><\/h1>\n<p><\/p>\n<p>a-m-team\u5ba3\u5e03\u5f00\u6e90\u4e86\u5b83\u7684\u84b8\u998f\u6570\u636e\uff0c\u4e3b\u8981\u5305\u62ecAM-Thinking-v1 Distilled Dataset\u548cQwen3-235B-A22B Distilled Dataset\u3002\u5176\u4e2d\u6bcf\u6761\u6570\u636e\u90fd\u5305\u542b\uff1a<\/p>\n<p>\u6807\u51c6\u5316 prompt\uff1b<\/p>\n<p>\u7b26\u5408\u9a8c\u8bc1\u6807\u51c6\u7684 \u00a0+ \u00a0\u683c\u5f0f\u63a8\u7406\u94fe\uff1b<\/p>\n<p>\u7cbe\u51c6\u5206\u4efb\u52a1\u5206\u7c7b\uff08\u6570\u5b66\u3001\u4ee3\u7801\u3001\u79d1\u5b66\u3001\u6307\u4ee4\u3001\u591a\u8f6e\u5bf9\u8bdd\u7b49\uff09\uff1b<\/p>\n<p>\u9a8c\u8bc1\u5206\u6570\uff08\u5747\u9ad8\u4e8e 0.9\uff09\u53ca\u81ea\u52a8\u8bc4\u5206\u4fe1\u606f\u3002<\/p>\n<p>\u7814\u7a76\u8005\u5e0c\u671b\uff0c\u8fd9\u4efd\u6570\u636e\u96c6\u4e0d\u4ec5\u80fd\u52a9\u529b\u793e\u533a\u4f4e\u6210\u672c\u8fbe\u6210\u5f3a\u63a8\u7406\u80fd\u529b\uff0c\u66f4\u80fd\u591f\u4e3a\u540e\u7eed RLHF\u3001CoT \u5fae\u8c03\u3001\u53ef\u63a7\u751f\u6210\u7b49\u9ad8\u7ea7\u80fd\u529b\u6784\u5efa\u63d0\u4f9b\u57fa\u7840\u3002\u540c\u65f6\uff0cAM\u56e2\u961f\u4e5f\u6b63\u5728\u8bd5\u6c34\u66f4\u5e7f\u6cdb\u7684\u6a21\u578b\u6539\u8fdb\u548c\u5e94\u7528\u573a\u666f\u3002<\/p>\n<p>\u6b63\u5982AM\u56e2\u961f\u5728\u4e00\u7bc7\u5206\u6790\u6587\u7ae0\u4e2d\u6307\u51fa\uff0c\u201c\u84b8\u998f\u6570\u636e\u7684\u9009\u62e9\u4e0d\u518d\u662f\u6280\u672f\u7ec6\u8282\uff0c\u800c\u662f\u51b3\u5b9a\u6027\u80fd\u4e0a\u9650\u7684\u5173\u952e\u56e0\u7d20\u3002\u201d\u4f18\u8d28\u7684\u84b8\u998f\u6570\u636e\u6e90\u672a\u6765\u6216\u8bb8\u5c06\u76f4\u63a5\u5f71\u54cd\u8bad\u7ec3\u6548\u7387\u548c\u6027\u80fd\u4e0a\u9650\u3002<\/p>\n<p>\u800c\u5bf9\u4e8e\u6a21\u578b\u89c2\u5bdf\u8005\u6765\u8bf4\uff0cAM\u56e2\u961f\u5728\u672c\u6b21\u8bba\u6587\u4e2d\u65d7\u5e1c\u9c9c\u660e\u5730\u63d0\u51fa\u4e86\u201cNot All Correct Answers Are Equal\u201d\uff0c\u5176\u5b9e\u4e5f\u662f\u5bf9\u73b0\u9636\u6bb5\u6a21\u578b\u7684\u4e00\u79cd\u5168\u65b0\u53cd\u601d\u6846\u67b6\uff1a<\/p>\n<p>\u5c24\u5176\u5728\u8bc4\u5206\u5dee\u5f02\u4e0e\u7528\u6237\u611f\u77e5\u9010\u6e10\u8131\u8282\uff0c\u8bb8\u591a\u4ece\u4e1a\u8005\u5bf9\u8bc4\u5206\u5ba1\u7f8e\u75b2\u52b3\u7684\u4eca\u5929\uff0c\u4ec0\u4e48\u6837\u7684\u5f00\u6e90\u5927\u6a21\u578b\u624d\u66f4\u53ef\u4ee5\u88ab\u79f0\u4f5c\u9886\u5148\u7684\uff1f\u84b8\u998f\u80cc\u540e\u4f53\u73b0\u7684\u6570\u636e\u7ed3\u6784\u6216\u8bb8\u4f1a\u662f\u4e00\u4e2a\u5168\u65b0\u7684\u89c6\u89d2\u3002<\/p>\n<p>\u8fd1\u65e5\uff0c\u5927\u5bb6\u90fd\u5728\u611f\u53f9\uff1aDeepSeek \u7684 R1 \u4e4b\u540e\uff0c\u5927\u6a21\u578b\u5df2\u7ecf\u5f88\u4e45\u6ca1\u6709\u65b0\u7684\u6280\u672f\u7a81\u7834\u4e86\u3002\u4f46\u5b9e\u9645\u8bc1\u660e\uff0c\u9ed8\u9ed8\u5728\u8ba4\u771f\u7814\u7a76\u6280\u672f\u7684\u56e2\u961f\u8fd8\u5728\u4e0d\u65ad\u521b\u65b0\u548c\u53d6\u5f97\u8d21\u732e\u3002<\/p>\n<p>\uff08\u96f7\u5cf0\u7f51<span style=\"color:#FD5D3C\">(\u516c\u4f17\u53f7\uff1a\u96f7\u5cf0\u7f51)<\/span>\uff09<\/p>\n<p><\/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\/05\/frc-caffd5db911581c593515fca738917f2.jpeg\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-caffd5db911581c593515fca738917f2.jpeg\"><\/a><br \/>\n                        <\/a>\n                    <\/p>\n<p>                                            \u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.leiphone.com\/category\/academic\/y5eOuO8Wa5UPSqpv.html\" target=\"_blank\">\u7eaf\u84b8\u998f\u6a21\u578b SOTA \u51fa\u73b0\uff01\u76f4\u63a5 SFT \u6210\u672c\u76f4\u964d 50 \u500d\uff0c\u6570\u636e\u5df2\u5168\u90e8\u5f00\u6e90<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>a-m-team \u53c8\u53d1\u65b0\u8bba\u6587\u4e86\u3002 \u8fd9\u4e2a\u56e2 [&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":[68],"class_list":["post-40306","post","type-post","status-publish","format-standard","hentry","category-news","tag-68"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.4 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u7eaf\u84b8\u998f\u6a21\u578b SOTA \u51fa\u73b0\uff01\u76f4\u63a5 SFT \u6210\u672c\u76f4\u964d 50 \u500d\uff0c\u6570\u636e\u5df2\u5168\u90e8\u5f00\u6e90 - \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=40306\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/17aitech.com\/?p=40306\",\"url\":\"https:\/\/17aitech.com\/?p=40306\",\"name\":\"\u7eaf\u84b8\u998f\u6a21\u578b SOTA \u51fa\u73b0\uff01\u76f4\u63a5 SFT \u6210\u672c\u76f4\u964d 50 \u500d\uff0c\u6570\u636e\u5df2\u5168\u90e8\u5f00\u6e90 - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\/\/17aitech.com\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/17aitech.com\/?p=40306#primaryimage\"},\"image\":{\"@id\":\"https:\/\/17aitech.com\/?p=40306#primaryimage\"},\"thumbnailUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3387c20d40788ad9020506d9b326b2c1.png\",\"datePublished\":\"2025-05-22T13:02:59+00:00\",\"author\":{\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\"},\"breadcrumb\":{\"@id\":\"https:\/\/17aitech.com\/?p=40306#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/17aitech.com\/?p=40306\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/?p=40306#primaryimage\",\"url\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3387c20d40788ad9020506d9b326b2c1.png\",\"contentUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3387c20d40788ad9020506d9b326b2c1.png\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/17aitech.com\/?p=40306#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/17aitech.com\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u7eaf\u84b8\u998f\u6a21\u578b SOTA \u51fa\u73b0\uff01\u76f4\u63a5 SFT \u6210\u672c\u76f4\u964d 50 \u500d\uff0c\u6570\u636e\u5df2\u5168\u90e8\u5f00\u6e90\"}]},{\"@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":"\u7eaf\u84b8\u998f\u6a21\u578b SOTA \u51fa\u73b0\uff01\u76f4\u63a5 SFT \u6210\u672c\u76f4\u964d 50 \u500d\uff0c\u6570\u636e\u5df2\u5168\u90e8\u5f00\u6e90 - \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=40306","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=40306","url":"https:\/\/17aitech.com\/?p=40306","name":"\u7eaf\u84b8\u998f\u6a21\u578b SOTA \u51fa\u73b0\uff01\u76f4\u63a5 SFT \u6210\u672c\u76f4\u964d 50 \u500d\uff0c\u6570\u636e\u5df2\u5168\u90e8\u5f00\u6e90 - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=40306#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=40306#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3387c20d40788ad9020506d9b326b2c1.png","datePublished":"2025-05-22T13:02:59+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=40306#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=40306"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=40306#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3387c20d40788ad9020506d9b326b2c1.png","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2025\/05\/frc-3387c20d40788ad9020506d9b326b2c1.png"},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=40306#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"\u7eaf\u84b8\u998f\u6a21\u578b SOTA \u51fa\u73b0\uff01\u76f4\u63a5 SFT \u6210\u672c\u76f4\u964d 50 \u500d\uff0c\u6570\u636e\u5df2\u5168\u90e8\u5f00\u6e90"}]},{"@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\/40306","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=40306"}],"version-history":[{"count":0,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/40306\/revisions"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=40306"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=40306"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=40306"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}