{"id":14803,"date":"2023-09-22T13:35:22","date_gmt":"2023-09-22T05:35:22","guid":{"rendered":"https:\/\/aif.amtbbs.org\/?p=14803"},"modified":"2023-09-22T13:35:22","modified_gmt":"2023-09-22T05:35:22","slug":"34b%e5%8f%82%e6%95%b0%e9%87%8f%e8%b6%85%e8%b6%8agpt-4%ef%bc%81%e3%80%8c%e6%95%b0%e5%ad%a6%e9%80%9a%e7%94%a8%e5%a4%a7%e6%a8%a1%e5%9e%8b%e3%80%8dmammoth%e5%bc%80%e6%ba%90%ef%bc%9a%e5%b9%b3%e5%9d%87","status":"publish","type":"post","link":"https:\/\/aif.amtbbs.org\/index.php\/2023\/09\/22\/14803\/","title":{"rendered":"34B\u53c2\u6570\u91cf\u8d85\u8d8aGPT-4\uff01\u300c\u6570\u5b66\u901a\u7528\u5927\u6a21\u578b\u300dMAmmoTH\u5f00\u6e90\uff1a\u5e73\u5747\u51c6\u786e\u7387\u6700\u9ad8\u63d0\u534729%"},"content":{"rendered":"<p>\u6570\u5b66\u63a8\u7406\u95ee\u9898\u662f\u8bed\u8a00\u6a21\u578b\u7ed5\u4e0d\u8fc7\u7684\u75db\u70b9\uff0c\u5728\u5404\u79cd\u9ed1\u79d1\u6280\u7684\u52a0\u6301\u4e0b\uff0c\u5f00\u6e90\u6a21\u578b\u7684\u63a8\u7406\u6027\u80fd\u4f9d\u7136\u4e0d\u591f\u770b\u3002<\/p>\n<p>\u6700\u8fd1\uff0c\u6ed1\u94c1\u5362\u5927\u5b66\u3001\u4fc4\u4ea5\u4fc4\u5dde\u7acb\u5927\u5b66\u3001\u9999\u6e2f\u79d1\u6280\u5927\u5b66\u3001\u7231\u4e01\u5821\u5927\u5b66\u7684\u7814\u7a76\u4eba\u5458\u8054\u5408\u5f00\u6e90\u4e86\u4e00\u4e2a\u4e13\u4e3a\u300c\u901a\u7528\u6570\u5b66\u95ee\u9898\u300d\u5b9a\u5236\u7684\u5927\u6a21\u578bMAmmoTH\u548c\u4e00\u4e2a\u6307\u4ee4\u8c03\u4f18\u6570\u636e\u96c6MathInstruct.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-14804\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-52.png\" width=\"1080\" height=\"352\" alt=\"\u56fe\u7247\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-52.png 1080w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-52-300x98.png 300w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-52-1024x334.png 1024w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-52-768x250.png 768w\" sizes=\"auto, (max-width: 1080px) 100vw, 1080px\" \/><\/p>\n<p>\u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/pdf\/2309.05653.pdf<\/p>\n<p>\u9879\u76ee\u94fe\u63a5\uff1ahttps:\/\/tiger-ai-lab.github.io\/MAmmoTH\/<\/p>\n<p>MathInstruct\u753113\u4e2a\u5177\u6709\u4e2d\u95f4\u539f\u7406\u7684\u6570\u5b66\u6570\u636e\u96c6\u7f16\u8bd1\u800c\u6210\uff0c\u5176\u4e2d6\u4e2a\u4e3a\u65b0\u6570\u636e\u96c6\uff0c\u6df7\u5408\u4e86\u601d\u60f3\u94fe\uff08CoT\uff09\u548c\u601d\u60f3\u7a0b\u5e8f\uff08PoT\uff09\uff0c\u5e76\u786e\u4fdd\u8986\u76d6\u4e86\u5e7f\u6cdb\u7684\u6570\u5b66\u9886\u57df\u3002<\/p>\n<p>CoT\u548cPoT\u7684\u6df7\u5408\u4e0d\u4ec5\u53ef\u4ee5\u91ca\u653e\u5de5\u5177\u4f7f\u7528\u7684\u6f5c\u529b\uff0c\u800c\u4e14\u8fd8\u5141\u8bb8\u6a21\u578b\u9488\u5bf9\u4e0d\u540c\u7684\u6570\u5b66\u95ee\u9898\u8fdb\u884c\u4e0d\u540c\u7684\u601d\u7ef4\u8fc7\u7a0b\u3002<\/p>\n<p>\u56e0\u6b64\uff0cMAmmoTH\u7cfb\u5217\u5728\u6240\u6709\u5c3a\u5ea6\u4e0a\u76849\u4e2a\u6570\u5b66\u63a8\u7406\u6570\u636e\u96c6\u4e0a\u7684\u8868\u73b0\u5927\u5927\u4f18\u4e8e\u73b0\u6709\u7684\u5f00\u6e90\u6a21\u578b\uff0c\u5e73\u5747\u51c6\u786e\u7387\u63d0\u9ad8\u4e8612%\u81f329%\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-14805\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-53.png\" width=\"1080\" height=\"666\" alt=\"\u56fe\u7247\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-53.png 1080w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-53-300x185.png 300w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-53-1024x631.png 1024w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-53-768x474.png 768w\" sizes=\"auto, (max-width: 1080px) 100vw, 1080px\" \/><\/p>\n<p>\u5176\u4e2dMAmmoTH-7B\u6a21\u578b\u5728MATH\uff08\u7ade\u8d5b\u7ea7\u6570\u636e\u96c6\uff09\u4e0a\u7684\u51c6\u786e\u7387\u8fbe\u5230\u4e8635%\uff0c\u8d85\u8fc7\u4e86\u6700\u597d\u7684\u5f00\u6e907B\u6a21\u578b\uff08WizardMath\uff0925%\uff0cMAmmoTH-34B\u6a21\u578b\u5728MATH\u4e0a\u7684\u51c6\u786e\u7387\u8fbe\u5230\u4e8646%\uff0c\u751a\u81f3\u8d85\u8fc7\u4e86GPT-4\u7684CoT\u7ed3\u679c\u3002<\/p>\n<section data-style-type=\"0\">\n<section>\u6570\u5b66\u63a8\u7406\u9886\u57df\u65b0\u738b\uff1aMAmmoTH<\/p>\n<\/section>\n<\/section>\n<p>\u5728\u6570\u5b66\u63a8\u7406\u4efb\u52a1\u4e0a\uff0c\u5f00\u6e90\u548c\u95ed\u6e90\u7684\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u4e4b\u95f4\u5b58\u5728\u5de8\u5927\u7684\u6027\u80fd\u5dee\u8ddd\uff0c\u76ee\u524d\u57fa\u51c6\u6570\u636e\u96c6\u4e0a\u7684sota\u4ecd\u7136\u662fGPT-4\uff0cPaLM-2\u548cClaude\u7b49\u95ed\u6e90\u6a21\u578b\uff0c\u5176\u4ed6\u5f00\u6e90\u6a21\u578b\u5982Llama\uff0cFalcon\u548cOPT\u7b49\u4ecd\u7136\u8fdc\u8fdc\u843d\u540e\u3002<\/p>\n<p>\u4e3a\u4e86\u5f25\u8865\u6027\u80fd\u5dee\u8ddd\uff0c\u4e3b\u8981\u7684\u7814\u7a76\u65b9\u6cd5\u6709\u4e24\u7c7b\uff1a<\/p>\n<p>1. \u5982Galactica\uff0cMINERVA\u7b49\u6a21\u578b\uff0c\u7ee7\u7eed\u4f7f\u7528\u6570\u5b66\u76f8\u5173\u7684\u7f51\u7edc\u6570\u636e\u5bf9\u8bed\u8a00\u6a21\u578b\u8fdb\u884c\u8bad\u7ec3\uff0c\u53ef\u4ee5\u63d0\u9ad8\u6a21\u578b\u7684\u901a\u7528\u79d1\u5b66\u63a8\u7406\u80fd\u529b\uff0c\u4f46\u8ba1\u7b97\u6210\u672c\u4f1a\u66f4\u9ad8\uff1b<\/p>\n<p>2. \u5982\u62d2\u7edd\u91c7\u6837\u5fae\u8c03\uff08RFT\uff09\u548cWizardMath\u7b49\uff0c\u4f7f\u7528\u7279\u5b9a\u9886\u57df\u6570\u636e\u96c6\u5bf9\u6a21\u578b\u8fdb\u884c\u5fae\u8c03\uff0c\u867d\u7136\u53ef\u4ee5\u63d0\u9ad8\u9886\u57df\u5185\u6027\u80fd\uff0c\u4f46\u65e0\u6cd5\u9002\u7528\u4e8e\u66f4\u5e7f\u6cdb\u7684\u6570\u5b66\u63a8\u7406\u4efb\u52a1\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-14807\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-54.png\" width=\"720\" height=\"405\" alt=\"\u56fe\u7247\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-54.png 720w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-54-300x169.png 300w\" sizes=\"auto, (max-width: 720px) 100vw, 720px\" \/><\/p>\n<p>\u5728\u89e3\u51b3\u6570\u5b66\u95ee\u9898\u65f6\uff0c\u73b0\u6709\u65b9\u6cd5\u901a\u5e38\u4f1a\u91c7\u7528\u601d\u7ef4\u94fe\uff08CoT\uff09\u65b9\u6cd5\u5f15\u5bfc\u8bed\u8a00\u6a21\u578b\u5faa\u5e8f\u6e10\u8fdb\u5730\u7528\u81ea\u7136\u8bed\u8a00\u63cf\u8ff0\u6765\u89e3\u51b3\u6570\u5b66\u95ee\u9898\u3002<\/p>\n<p>\u867d\u7136\u5728\u5927\u591a\u6570\u6570\u5b66\u4e3b\u9898\u4e0b\u8868\u73b0\u51fa\u5f88\u597d\u7684\u901a\u7528\u6027\uff0c\u4f46\u5728\u9700\u8981\u7cbe\u786e\u6216\u590d\u6742\u7684\u6570\u5b66\u8ba1\u7b97\u3001\u7b97\u6cd5\u63a8\u7406\u7684\u95ee\u9898\u4e0b\uff08\u5982\u6c42\u89e3\u4e8c\u6b21\u65b9\u7a0b\u6839\uff0c\u8ba1\u7b97\u77e9\u9635\u7279\u5f81\u503c\uff09\u8868\u73b0\u4e0d\u4f73\u3002<\/p>\n<p>\u76f8\u6bd4\u4e4b\u4e0b\uff0c\u601d\u7ef4\u7a0b\u5e8f\uff08PoT, Program-of-Thought\uff09\u65b9\u6cd5\u548cPAL\u5229\u7528\u5916\u90e8\u5de5\u5177\uff08\u5373Python\u89e3\u91ca\u5668\uff09\u5927\u5927\u7b80\u5316\u4e86\u6570\u5b66\u6c42\u89e3\u8fc7\u7a0b\uff0c\u5c06\u8ba1\u7b97\u8fc7\u7a0b\u5378\u8f7d\u5230\u5916\u90e8Python\u89e3\u91ca\u5668\uff0c\u4ee5\u89e3\u51b3\u590d\u6742\u7684\u6570\u5b66\u548c\u7b97\u6cd5\u63a8\u7406\u8fc7\u7a0b\uff08\u4f8b\u5982\uff0c\u7528sympy\u6c42\u89e3\u4e8c\u6b21\u65b9\u7a0b\u6216\u7528numpy\u8ba1\u7b97\u77e9\u9635\u7279\u5f81\u503c\uff09\u3002<\/p>\n<p>\u7136\u800c\uff0cPoT\u5728\u5904\u7406\u66f4\u62bd\u8c61\u7684\u63a8\u7406\u573a\u666f\u65b9\u9762\u6709\u6240\u6b20\u7f3a\uff0c\u5c24\u5176\u662f\u5728\u6ca1\u6709\u5185\u7f6eAPI\u7684\u60c5\u51b5\u4e0b\uff0c\u5e38\u8bc6\u63a8\u7406\u3001\u5f62\u5f0f\u903b\u8f91\u548c\u62bd\u8c61\u4ee3\u6570\u7684\u63a8\u7406\u80fd\u529b\u4f1a\u66f4\u5dee\u3002<\/p>\n<section data-style-type=\"0\">\n<section>\u65b9\u6cd5\u6982\u8ff0<\/p>\n<\/section>\n<\/section>\n<p>\u7814\u7a76\u4eba\u5458\u7684\u76ee\u6807\u662f\u7f16\u5236\u4e00\u4e2a\u9ad8\u8d28\u91cf\u3001\u591a\u6837\u5316\u7684\u6570\u5b66\u6307\u4ee4\u8c03\u6574\uff08instruction-tuning\uff09\u6570\u636e\u96c6\u5217\u8868\u3002<\/p>\n<p><strong>1. \u8986\u76d6\u4e0d\u540c\u6570\u5b66\u9886\u57df\u548c\u590d\u6742\u5ea6<\/strong><\/p>\n<p>\u66f4\u5168\u9762\u7684\u6570\u636e\u96c6\u53ef\u4ee5\u8ba9\u6a21\u578b\u63a5\u89e6\u5230\u591a\u6837\u5316\u7684\u6570\u5b66\u77e5\u8bc6\uff0c\u63d0\u5347\u6a21\u578b\u7684\u591a\u529f\u80fd\u6027\u3002<\/p>\n<p>\u7814\u7a76\u4eba\u5458\u5c06\u9009\u62e9\u8303\u56f4\u7f29\u5c0f\u5230\u51e0\u4e2a\u88ab\u5e7f\u6cdb\u91c7\u7528\u7684\u9ad8\u8d28\u91cf\u6570\u636e\u96c6\uff0c\u5305\u62ecGSM8K\u3001math\u3001AQuA\u3001Camel\u548cTheoremQA.<\/p>\n<p>\u8fd8\u53ef\u4ee5\u6ce8\u610f\u5230\uff0c\u73b0\u6709\u7684\u6570\u636e\u96c6\u7f3a\u4e4f\u5bf9\u5927\u5b66\u6c34\u5e73\u7684\u6570\u5b66\u77e5\u8bc6\u7684\u8986\u76d6\uff0c\u5982\u62bd\u8c61\u4ee3\u6570\u548c\u5f62\u5f0f\u903b\u8f91\uff0c\u6240\u4ee5\u7814\u7a76\u4eba\u5458\u9009\u62e9\u4f7f\u7528GPT-4\u6765\u5408\u6210TheoremQA\u95ee\u9898\u4e2d\u7684\u601d\u7ef4\u94fe\uff08CoT\uff09\u539f\u7406\uff0c\u5229\u7528\u7f51\u7edc\u4e0a\u627e\u5230\u7684\u6570\u4e2a\u79cd\u5b50\u6837\u4f8b\uff0c\u901a\u8fc7\u81ea\u6211\u6307\u5bfc\uff08self-instruct\uff09\u521b\u5efa\u95ee\u9898\u548cCoT\u7684\u6570\u636e\u5bf9\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-14808\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-55.png\" width=\"585\" height=\"600\" alt=\"\u56fe\u7247\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-55.png 585w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-55-293x300.png 293w\" sizes=\"auto, (max-width: 585px) 100vw, 585px\" \/><\/p>\n<p><strong>2. \u6df7\u5408CoT\u548cPoT<\/strong><\/p>\n<p>\u73b0\u6709\u7684\u7814\u7a76\u65b9\u6cd5\u5927\u591a\u53ea\u5173\u6ce8CoT\uff0c\u5e76\u4e14\u6570\u636e\u96c6\u4e2d\u4e5f\u53ea\u5305\u542b\u6709\u9650\u7684\u89e3\u9898\u601d\u8def\uff0c\u5bfc\u81f4CoT\u548cPoT\u7684\u6570\u636e\u91cf\u5341\u5206\u4e0d\u5747\u8861\u3002<\/p>\n<p>\u4e3a\u4e86\u89e3\u51b3\u8be5\u95ee\u9898\uff0c\u7814\u7a76\u4eba\u5458\u5229\u7528GPT-4\u6765\u8865\u5145\u9009\u5b9a\u6570\u636e\u96c6\u7684PoT\u89e3\u9898\u601d\u8def\uff0c\u901a\u8fc7\u5bf9\u6bd4\u5408\u6210\u7a0b\u5e8f\u7684\u6267\u884c\u7ed3\u679c\u4ee5\u53ca\u4eba\u5de5\u6807\u6ce8\u7684\u7b54\u6848\u8fdb\u884c\u8fc7\u6ee4\uff0c\u786e\u4fdd\u751f\u6210\u6570\u636e\u7684\u9ad8\u8d28\u91cf\u3002<\/p>\n<p>\u9075\u5faa\u4e0a\u8ff0\u65b9\u6cd5\uff0c\u6700\u540e\u5f97\u5230\u4e8626\u4e07\u6761\u6307\u4ee4\u3001\u56de\u590d\u6570\u636e\u5bf9\uff0c\u6db5\u76d6\u4e86\u5e7f\u6cdb\u7684\u6838\u5fc3\u6570\u5b66\u9886\u57df\uff0c\u5982\u7b97\u672f\u3001\u4ee3\u6570\u3001\u6982\u7387\u3001\u5fae\u79ef\u5206\u548c\u51e0\u4f55\u7b49\uff0c\u6df7\u5408\u4e86CoT\u548cPoT\u57fa\u672c\u539f\u7406\uff0c\u5e76\u63d0\u4f9b\u591a\u79cd\u8bed\u8a00\u3001\u591a\u4e2a\u96be\u5ea6\u7ea7\u522b\u7684\u6570\u636e\uff0c\u8db3\u4ee5\u8bc1\u660e\u6570\u636e\u96c6\u7684\u9ad8\u54c1\u8d28\u548c\u72ec\u7279\u6027\u3002<\/p>\n<p><strong>\u8bad\u7ec3\u6b65\u9aa4<\/strong><\/p>\n<p>\u7814\u7a76\u4eba\u5458\u7edf\u4e00\u4e86MathInstruct\u4e2d\u7684\u6240\u6709\u5b50\u96c6\uff0c\u5c06\u6307\u4ee4\u6570\u636e\u96c6\u7684\u7ed3\u6784\u6807\u51c6\u5316\u4e3aAlpaca\u6a21\u578b\u7684\u683c\u5f0f\uff0c\u4f7f\u5f97\u6a21\u578b\u65e0\u9700\u8003\u8651\u539f\u59cb\u6570\u636e\u96c6\u7684\u683c\u5f0f\uff0c\u5728\u5fae\u8c03\u9636\u6bb5\u7edf\u4e00\u5904\u7406\u6570\u636e\u5373\u53ef\u3002<\/p>\n<p>\u7814\u7a76\u4eba\u5458\u9009\u62e9\u5f00\u6e90\u6a21\u578bLlama-2\u548cCode Llama\u4f5c\u4e3a\u57fa\u7840\u6a21\u578b\uff0c\u57287B\u300113B\u300134B\u548c70B\u5c3a\u5bf8\u7684\u6a21\u578b\u4e0a\u8fdb\u884c\u5fae\u8c03\u3002<\/p>\n<section data-style-type=\"0\">\n<section>\u5b9e\u9a8c\u90e8\u5206<\/p>\n<\/section>\n<\/section>\n<p><strong>\u8bc4\u4f30\u6570\u636e\u96c6<\/strong><\/p>\n<p>\u7814\u7a76\u4eba\u5458\u9009\u62e9\u4e86\u4e0d\u540c\u6570\u5b66\u9886\u57df\u4e0b\u7684\u6837\u672c\uff0c\u5bf9\u6a21\u578b\u7684\u901a\u7528\u6570\u5b66\u63a8\u7406\u80fd\u529b\u8fdb\u884c\u8bc4\u4f30\uff1a<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-14809\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-56.png\" width=\"1080\" height=\"454\" alt=\"\u56fe\u7247\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-56.png 1080w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-56-300x126.png 300w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-56-1024x430.png 1024w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-56-768x323.png 768w\" sizes=\"auto, (max-width: 1080px) 100vw, 1080px\" \/><\/p>\n<p>\u9886\u57df\u5185\u6570\u636e\u96c6\u5305\u62ecGSM8K\uff0cMATH\uff0cAQuA-RAT\uff0cNumGLUE\uff1b\u9886\u57df\u5916\u6570\u636e\u96c6\u5305\u62ecSVAMP\uff0cMathematics\uff0cSimulEq\uff0cSAT-Math\u548cSimulEq\uff0c\u6db5\u76d6\u4e86\u5c0f\u5b66\u3001\u9ad8\u4e2d\u548c\u5927\u5b66\u6c34\u5e73\u7684\u6570\u5b66\u95ee\u9898\uff0c\u90e8\u5206\u6570\u636e\u96c6\u751a\u81f3\u5305\u62ec\u5f62\u5f0f\u903b\u8f91\u548c\u5e38\u8bc6\u63a8\u7406\u3002<\/p>\n<p>\u95ee\u9898\u7c7b\u578b\u4e3a\u5f00\u653e\u5f0f\u95ee\u9898\u548c\u591a\u9009\u9898\uff0c\u5176\u4e2d\u5f00\u653e\u5f0f\u95ee\u9898\uff08\u5982GSM8K\u3001\u6570\u5b66\uff09\u91c7\u7528PoT\u89e3\u7801\uff0c\u56e0\u4e3a\u5927\u591a\u6570\u95ee\u9898\u90fd\u53ef\u4ee5\u7531\u7a0b\u5e8f\u89e3\u51b3\uff1b\u591a\u9879\u9009\u62e9\u9898\uff08\u5982AQuA\u3001MMLU\uff09\u91c7\u7528CoT\u89e3\u7801\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-14810\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-57.png\" width=\"1080\" height=\"518\" alt=\"\u56fe\u7247\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-57.png 1080w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-57-300x144.png 300w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-57-1024x491.png 1024w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-57-768x368.png 768w\" sizes=\"auto, (max-width: 1080px) 100vw, 1080px\" \/><\/p>\n<p>CoT\u89e3\u7801\u4e0d\u9700\u8981\u89e6\u53d1\u8bcd\uff0cPoT\u9700\u8981\u89e6\u53d1\u77ed\u8bed\u300c\u8ba9\u6211\u4eec\u5199\u4e2a\u7a0b\u5e8f\u6765\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\u300d\uff08Let\u2019s write a program to solve the problem\uff09\u3002<\/p>\n<p><strong>\u5b9e\u9a8c\u7ed3\u679c<\/strong><\/p>\n<p>\u603b\u7684\u6765\u8bf4\uff0cMAmmoTH\u548cMAmmoTH-Coder\u5728\u4e0d\u540c\u7684\u6a21\u578b\u5c3a\u5bf8\u4e0a\u5747\u4f18\u4e8eSoTA\u6a21\u578b\uff0c\u5e76\u4e14\u5728\u9886\u57df\u5916\uff08OOD\uff09\u6570\u636e\u96c6\u4e0a\u7684\u589e\u76ca\u8981\u663e\u8457\u4f18\u4e8e\u9886\u57df\u5185\uff08IND\uff09\u6570\u636e\u96c6\uff0c\u5c55\u73b0\u51fa\u4e86\u8be5\u6a21\u578b\u4f5c\u4e3a\u6570\u5b66\u901a\u624d\u6a21\u578b\u7684\u6f5c\u529b\uff0c\u751a\u81f3\u5728\u51e0\u4e2a\u6570\u636e\u96c6\u4e0a\uff0cMAmmoTH-Coder-34B\u548cMAmmoTH-70B\u751a\u81f3\u8d85\u8fc7\u4e86\u95ed\u6e90\u6a21\u578b\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-14811\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-58.png\" width=\"851\" height=\"1280\" alt=\"\u56fe\u7247\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-58.png 851w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-58-199x300.png 199w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-58-681x1024.png 681w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-58-768x1155.png 768w\" sizes=\"auto, (max-width: 851px) 100vw, 851px\" \/><\/p>\n<p>&nbsp;<\/p>\n<p>\u5728\u9886\u57df\u5185\u6570\u636e\u7684\u8bc4\u4f30\uff0cMAmmoTH\u6a21\u578b\u7684\u4e3b\u8981\u7ade\u4e89\u5bf9\u624b\u662fWizardMath\u548cPlatypus\uff0c\u5176\u4e2dWizardMath\u7684\u8bad\u7ec3\u6df1\u5ea6\u4f9d\u8d56\u4e8eGSM8K\u548cMATH\u6570\u636e\u96c6\uff0cPlatypus\u5728\u66f4\u5e7f\u6cdb\u7684\u6587\u672c\u548c\u6570\u5b66\u63a8\u7406\u6570\u636e\u96c6\u4e0a\u5bf9LLM\u8fdb\u884c\u5fae\u8c03\u3002<\/p>\n<p>\u76f8\u6bd4\u4e4b\u4e0b\uff0cMAmmoTH\u5b9e\u73b0\u4e86\u5168\u9762\u7684\u6539\u8fdb\uff0c\u5e76\u4e14\u66f4\u64c5\u957f\u89e3\u51b3\u590d\u6742\u6570\u5b66\u95ee\u9898\uff0c\u76f8\u6bd4WizardMath\uff08MATH\u6570\u636e\u7684sota\uff09\u7684\u589e\u76ca\u6700\u9ad8\u8d85\u8fc7\u4e8625%<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-14812\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-59.png\" width=\"930\" height=\"1280\" alt=\"\u56fe\u7247\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-59.png 930w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-59-218x300.png 218w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-59-744x1024.png 744w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2023\/09\/4ffce04d92a4d6cb21c1494cdfcd6dc1-59-768x1057.png 768w\" sizes=\"auto, (max-width: 930px) 100vw, 930px\" \/><\/p>\n<p>&nbsp;<\/p>\n<p>\u5728\u9886\u57df\u5916\u6570\u636e\u8bc4\u4f30\u4e2d\uff0c\u4e3b\u8981\u7ade\u4e89\u6a21\u578b\u4f9d\u7136\u662fPlatypus\uff0c\u4e0d\u8fc7MAmmoTH\u53ef\u4ee5\u5b9e\u73b0\u6bd4\u9886\u57df\u5185\u6570\u636e\u66f4\u9ad8\u7684\u6027\u80fd\u63d0\u5347\uff0c\u5c55\u73b0\u51fa\u5bf9\u672a\u77e5\u6570\u5b66\u95ee\u9898\u7684\u901a\u7528\u80fd\u529b\u3002<\/p>\n<p>\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0cMAmmoTH-7B\u8fd8\u5c06WizardMath-7B\u5728MMLU-Math\u4e0a\u7684CoT\u6027\u80fd\u5927\u5e45\u63d0\u9ad8\u4e869%\uff0c\u5176\u4e2d\u5305\u542b\u5927\u91cf\u6ca1\u6709\u5728\u8bad\u7ec3\u6570\u636e\u96c6\u4e2d\u6db5\u76d6\u7684\u4e3b\u9898\u3002<\/p>\n<p><strong>\u4e0d\u540c\u57fa\u7840\u6a21\u578b\u4e4b\u95f4\u7684\u5bf9\u6bd4<\/strong><\/p>\n<p>\u53ef\u4ee5\u53d1\u73b0\uff0cCode-Llama\u4f5c\u4e3a\u57fa\u7840\u6a21\u578b\u65f6\u7684\u6548\u679c\u59cb\u7ec8\u4f18\u4e8eLlama-2\uff0c\u5c24\u5176\u662f\u5728\u9886\u57df\u5916\u6570\u636e\u96c6\u4e0a\uff0c\u4e8c\u8005\u4e4b\u95f4\u7684\u6027\u80fd\u5dee\u5f02\u751a\u81f3\u8fbe\u5230\u4e865%\uff0c\u5176\u4e2dMAmmoTH-Coder\uff0834B\uff09\u5728\u9886\u57df\u5916\u6570\u636e\u96c6\u4e0a\u7684\u5e73\u5747\u6027\u80fd\u5b9e\u9645\u4e0a\u9ad8\u4e8eMAmmoTH\uff0870B\uff09<\/p>\n<p>\u7814\u7a76\u4eba\u5458\u8ba4\u4e3a\uff0cMAmmoTH-Coder\u4eceCode-Llama\u7684\u6301\u7eed\u4ee3\u7801\u8bad\u7ec3\u4e2d\u53d7\u76ca\u532a\u6d45\uff0c\u4e0d\u4ec5\u589e\u5f3a\u4e86PoT\u80fd\u529b\uff0c\u8fd8\u63d0\u9ad8\u4e86Llama\u7684\u901a\u7528\u63a8\u7406\u6280\u80fd\u3002<\/p>\n<div class=\"pvc_clear\"><\/div>\n<p id=\"pvc_stats_14803\" class=\"pvc_stats total_only  \" data-element-id=\"14803\" style=\"\"><i class=\"pvc-stats-icon medium\" aria-hidden=\"true\"><svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" version=\"1.0\" viewBox=\"0 0 502 315\" preserveAspectRatio=\"xMidYMid meet\"><g transform=\"translate(0,332) scale(0.1,-0.1)\" fill=\"\" stroke=\"none\"><path d=\"M2394 3279 l-29 -30 -3 -207 c-2 -182 0 -211 15 -242 39 -76 157 -76 196 0 15 31 17 60 15 243 l-3 209 -33 29 c-26 23 -41 29 -80 29 -41 0 -53 -5 -78 -31z\"\/><path d=\"M3085 3251 c-45 -19 -58 -50 -96 -229 -47 -217 -49 -260 -13 -295 52 -53 146 -42 177 20 16 31 87 366 87 410 0 70 -86 122 -155 94z\"\/><path d=\"M1751 3234 c-13 -9 -29 -31 -37 -50 -12 -29 -10 -49 21 -204 19 -94 39 -189 45 -210 14 -50 54 -80 110 -80 34 0 48 6 76 34 21 21 34 44 34 59 0 14 -18 113 -40 219 -37 178 -43 195 -70 221 -36 32 -101 37 -139 11z\"\/><path d=\"M1163 3073 c-36 -7 -73 -59 -73 -102 0 -56 133 -378 171 -413 34 -32 83 -37 129 -13 70 36 67 87 -16 290 -86 209 -89 214 -129 231 -35 14 -42 15 -82 7z\"\/><path d=\"M3689 3066 c-15 -9 -33 -30 -42 -48 -48 -103 -147 -355 -147 -375 0 -98 131 -148 192 -74 13 15 57 108 97 206 80 196 84 226 37 273 -30 30 -99 39 -137 18z\"\/><path d=\"M583 2784 c-38 -19 -67 -74 -58 -113 9 -42 211 -354 242 -373 16 -10 45 -18 66 -18 51 0 107 52 107 100 0 39 -1 41 -124 234 -80 126 -108 162 -133 173 -41 17 -61 16 -100 -3z\"\/><path d=\"M4250 2784 c-14 -9 -74 -91 -133 -183 -95 -150 -107 -173 -107 -213 0 -55 33 -94 87 -104 67 -13 90 8 211 198 130 202 137 225 78 284 -27 27 -42 34 -72 34 -22 0 -50 -8 -64 -16z\"\/><path d=\"M2275 2693 c-553 -48 -1095 -270 -1585 -649 -135 -104 -459 -423 -483 -476 -23 -49 -22 -139 2 -186 73 -142 361 -457 571 -626 285 -228 642 -407 990 -497 242 -63 336 -73 660 -74 310 0 370 5 595 52 535 111 1045 392 1455 803 122 121 250 273 275 326 19 41 19 137 0 174 -41 79 -309 363 -465 492 -447 370 -946 591 -1479 653 -113 14 -422 18 -536 8z m395 -428 c171 -34 330 -124 456 -258 112 -119 167 -219 211 -378 27 -96 24 -300 -5 -401 -72 -255 -236 -447 -474 -557 -132 -62 -201 -76 -368 -76 -167 0 -236 14 -368 76 -213 98 -373 271 -451 485 -162 444 86 934 547 1084 153 49 292 57 452 25z m909 -232 c222 -123 408 -262 593 -441 76 -74 138 -139 138 -144 0 -16 -233 -242 -330 -319 -155 -123 -309 -223 -461 -299 l-81 -41 32 46 c18 26 49 83 70 128 143 306 141 649 -6 957 -25 52 -61 116 -79 142 l-34 47 45 -20 c26 -10 76 -36 113 -56z m-2057 25 c-40 -58 -105 -190 -130 -263 -110 -324 -59 -707 132 -981 25 -35 42 -64 37 -64 -19 0 -241 119 -326 174 -188 122 -406 314 -532 468 l-58 71 108 103 c185 178 428 349 672 473 66 33 121 60 123 61 2 0 -10 -19 -26 -42z\"\/><path d=\"M2375 1950 c-198 -44 -350 -190 -395 -379 -18 -76 -8 -221 19 -290 114 -284 457 -406 731 -260 98 52 188 154 231 260 27 69 37 214 19 290 -38 163 -166 304 -326 360 -67 23 -215 33 -279 19z\"\/><\/g><\/svg><\/i> <img loading=\"lazy\" decoding=\"async\" width=\"16\" height=\"16\" alt=\"Loading\" src=\"https:\/\/aif.amtbbs.org\/wp-content\/plugins\/page-views-count\/ajax-loader-2x.gif\" border=0 \/><\/p>\n<div class=\"pvc_clear\"><\/div>\n","protected":false},"excerpt":{"rendered":"<p>\u6570\u5b66\u63a8\u7406\u95ee\u9898\u662f\u8bed\u8a00\u6a21\u578b\u7ed5\u4e0d\u8fc7\u7684\u75db\u70b9\uff0c\u5728\u5404\u79cd\u9ed1\u79d1\u6280\u7684\u52a0\u6301\u4e0b\uff0c\u5f00\u6e90\u6a21\u578b\u7684\u63a8\u7406\u6027\u80fd\u4f9d\u7136\u4e0d\u591f\u770b\u3002 \u6700\u8fd1\uff0c\u6ed1\u94c1\u5362\u5927\u5b66\u3001\u4fc4 [&hellip;]<\/p>\n<div class=\"pvc_clear\"><\/div>\n<p id=\"pvc_stats_14803\" class=\"pvc_stats total_only  \" data-element-id=\"14803\" style=\"\"><i class=\"pvc-stats-icon medium\" aria-hidden=\"true\"><svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" version=\"1.0\" viewBox=\"0 0 502 315\" preserveAspectRatio=\"xMidYMid meet\"><g transform=\"translate(0,332) scale(0.1,-0.1)\" fill=\"\" stroke=\"none\"><path d=\"M2394 3279 l-29 -30 -3 -207 c-2 -182 0 -211 15 -242 39 -76 157 -76 196 0 15 31 17 60 15 243 l-3 209 -33 29 c-26 23 -41 29 -80 29 -41 0 -53 -5 -78 -31z\"\/><path d=\"M3085 3251 c-45 -19 -58 -50 -96 -229 -47 -217 -49 -260 -13 -295 52 -53 146 -42 177 20 16 31 87 366 87 410 0 70 -86 122 -155 94z\"\/><path d=\"M1751 3234 c-13 -9 -29 -31 -37 -50 -12 -29 -10 -49 21 -204 19 -94 39 -189 45 -210 14 -50 54 -80 110 -80 34 0 48 6 76 34 21 21 34 44 34 59 0 14 -18 113 -40 219 -37 178 -43 195 -70 221 -36 32 -101 37 -139 11z\"\/><path d=\"M1163 3073 c-36 -7 -73 -59 -73 -102 0 -56 133 -378 171 -413 34 -32 83 -37 129 -13 70 36 67 87 -16 290 -86 209 -89 214 -129 231 -35 14 -42 15 -82 7z\"\/><path d=\"M3689 3066 c-15 -9 -33 -30 -42 -48 -48 -103 -147 -355 -147 -375 0 -98 131 -148 192 -74 13 15 57 108 97 206 80 196 84 226 37 273 -30 30 -99 39 -137 18z\"\/><path d=\"M583 2784 c-38 -19 -67 -74 -58 -113 9 -42 211 -354 242 -373 16 -10 45 -18 66 -18 51 0 107 52 107 100 0 39 -1 41 -124 234 -80 126 -108 162 -133 173 -41 17 -61 16 -100 -3z\"\/><path d=\"M4250 2784 c-14 -9 -74 -91 -133 -183 -95 -150 -107 -173 -107 -213 0 -55 33 -94 87 -104 67 -13 90 8 211 198 130 202 137 225 78 284 -27 27 -42 34 -72 34 -22 0 -50 -8 -64 -16z\"\/><path d=\"M2275 2693 c-553 -48 -1095 -270 -1585 -649 -135 -104 -459 -423 -483 -476 -23 -49 -22 -139 2 -186 73 -142 361 -457 571 -626 285 -228 642 -407 990 -497 242 -63 336 -73 660 -74 310 0 370 5 595 52 535 111 1045 392 1455 803 122 121 250 273 275 326 19 41 19 137 0 174 -41 79 -309 363 -465 492 -447 370 -946 591 -1479 653 -113 14 -422 18 -536 8z m395 -428 c171 -34 330 -124 456 -258 112 -119 167 -219 211 -378 27 -96 24 -300 -5 -401 -72 -255 -236 -447 -474 -557 -132 -62 -201 -76 -368 -76 -167 0 -236 14 -368 76 -213 98 -373 271 -451 485 -162 444 86 934 547 1084 153 49 292 57 452 25z m909 -232 c222 -123 408 -262 593 -441 76 -74 138 -139 138 -144 0 -16 -233 -242 -330 -319 -155 -123 -309 -223 -461 -299 l-81 -41 32 46 c18 26 49 83 70 128 143 306 141 649 -6 957 -25 52 -61 116 -79 142 l-34 47 45 -20 c26 -10 76 -36 113 -56z m-2057 25 c-40 -58 -105 -190 -130 -263 -110 -324 -59 -707 132 -981 25 -35 42 -64 37 -64 -19 0 -241 119 -326 174 -188 122 -406 314 -532 468 l-58 71 108 103 c185 178 428 349 672 473 66 33 121 60 123 61 2 0 -10 -19 -26 -42z\"\/><path d=\"M2375 1950 c-198 -44 -350 -190 -395 -379 -18 -76 -8 -221 19 -290 114 -284 457 -406 731 -260 98 52 188 154 231 260 27 69 37 214 19 290 -38 163 -166 304 -326 360 -67 23 -215 33 -279 19z\"\/><\/g><\/svg><\/i> <img loading=\"lazy\" decoding=\"async\" width=\"16\" height=\"16\" alt=\"Loading\" src=\"https:\/\/aif.amtbbs.org\/wp-content\/plugins\/page-views-count\/ajax-loader-2x.gif\" border=0 \/><\/p>\n<div class=\"pvc_clear\"><\/div>\n","protected":false},"author":8,"featured_media":14804,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3],"tags":[],"class_list":["post-14803","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/14803","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\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/comments?post=14803"}],"version-history":[{"count":1,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/14803\/revisions"}],"predecessor-version":[{"id":14814,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/14803\/revisions\/14814"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/media\/14804"}],"wp:attachment":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/media?parent=14803"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/categories?post=14803"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/tags?post=14803"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}