{"id":24642,"date":"2024-12-07T09:55:23","date_gmt":"2024-12-07T01:55:23","guid":{"rendered":"https:\/\/www.1ai.net\/?p=24642"},"modified":"2024-12-07T09:55:23","modified_gmt":"2024-12-07T01:55:23","slug":"openai-%e6%b4%bb%e5%8a%a8%e7%ac%ac%e4%ba%8c%e5%bc%b9%ef%bc%9a%e5%bc%ba%e5%8c%96%e5%be%ae%e8%b0%83%e6%89%93%e9%80%a0%e9%a2%86%e5%9f%9f%e4%b8%93%e5%ae%b6-ai-%e6%a8%a1%e5%9e%8b%ef%bc%8c","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/24642.html","title":{"rendered":"OpenAI Event #2: \"Enhanced Fine-Tuning\" Builds Domain Expert AI Models, Altman Calls It the Biggest Surprise of the Year"},"content":{"rendered":"

12 \u6708 7 \u65e5\u6d88\u606f\uff0cOpenAI<\/a> \u542f\u52a8\u4e86\u4e3a\u671f 12 \u5929\u7684\u201cshipmas\u201d\u65b0\u54c1\u53d1\u5e03\u5468\u671f\uff0c\u5c06\u63a8\u51fa\u4e00\u7cfb\u5217\u65b0\u529f\u80fd\u3001\u65b0\u4ea7\u54c1\u4ee5\u53ca\u76f8\u5173\u6f14\u793a\u3002\u672c\u6b21\u6d3b\u52a8\u7b2c\u4e8c\u65e5\uff0cOpenAI \u63a8\u51fa\u4e86\u5f3a\u5316\u5fae\u8c03\uff08Reinforcement Fine-Tuning\uff09\uff0c\u5e2e\u52a9\u5f00\u53d1\u8005\u548c\u673a\u5668\u5b66\u4e60\u5de5\u7a0b\u5e08\u6253\u9020\u9488\u5bf9\u7279\u5b9a\u590d\u6742\u9886\u57df\u4efb\u52a1\u7684\u4e13\u5bb6\u6a21\u578b\u3002<\/strong><\/p>\n

\"055884d5j00so3pb40049d000qy00evp\"<\/p>\n

\u8be5\u9879\u76ee\u901a\u8fc7\u5168\u65b0\u7684\u6a21\u578b\u5b9a\u5236\u6280\u672f\uff0c\u8ba9\u5f00\u53d1\u8005\u53ef\u4ee5\u4f7f\u7528\u9ad8\u8d28\u91cf\u4efb\u52a1\u96c6\u5bf9\u6a21\u578b\u8fdb\u884c\u5fae\u8c03\uff0c\u5e76\u5229\u7528\u53c2\u8003\u7b54\u6848\u8bc4\u4f30\u6a21\u578b\u7684\u54cd\u5e94\uff0c\u4ece\u800c\u63d0\u5347\u6a21\u578b\u5728\u7279\u5b9a\u9886\u57df\u4efb\u52a1\u4e2d\u7684\u63a8\u7406\u80fd\u529b\u548c\u51c6\u786e\u6027\u3002<\/p>\n

\u5f3a\u5316\u5fae\u8c03\u7b80\u4ecb<\/strong><\/p>\n

1AI\u9644\u4e0a\u5b98\u65b9\u4ecb\u7ecd\uff1a\u5f00\u53d1\u4eba\u5458\u80fd\u591f\u4f7f\u7528\u6570\u5341\u5230\u6570\u5343\u4e2a\u9ad8\u8d28\u91cf\u4efb\u52a1\uff0c\u5b9a\u5236 OpenAI \u7684\u6a21\u578b\uff0c\u5e76\u4f7f\u7528\u63d0\u4f9b\u7684\u53c2\u8003\u7b54\u6848\u5bf9\u6a21\u578b\u7684\u54cd\u5e94\u8fdb\u884c\u8bc4\u5206\u3002\u5b98\u65b9\u8868\u793a\u8fd9\u9879\u6280\u672f\u5f3a\u5316\u4e86\u6a21\u578b\u63a8\u7406\u7c7b\u4f3c\u95ee\u9898\u7684\u65b9\u5f0f\uff0c\u5e76\u63d0\u9ad8\u4e86\u5176\u5728\u8be5\u9886\u57df\u7279\u5b9a\u4efb\u52a1\u4e0a\u7684\u51c6\u786e\u6027\u3002<\/p>\n

\u4e0e\u6807\u51c6\u5fae\u8c03\u4e0d\u540c\uff0cRFT \u5229\u7528\u5f3a\u5316\u5b66\u4e60\u7b97\u6cd5\uff0c\u53ef\u4ee5\u5c06\u6a21\u578b\u6027\u80fd\u4ece\u9ad8\u4e2d\u6c34\u5e73\u63d0\u5347\u5230\u4e13\u5bb6\u535a\u58eb\u6c34\u5e73\u3002<\/p>\n

RFT \u4e0e\u76d1\u7763\u5f0f\u5fae\u8c03\u4e0d\u540c\uff0c\u5b83\u4e0d\u662f\u8ba9\u6a21\u578b\u6a21\u4eff\u8f93\u5165\uff0c\u800c\u662f\u6559\u6a21\u578b\u4ee5\u5168\u65b0\u7684\u65b9\u5f0f\u8fdb\u884c\u63a8\u7406\uff0c\u901a\u8fc7\u5bf9\u6a21\u578b\u7b54\u6848\u8fdb\u884c\u8bc4\u5206\u5e76\u5f3a\u5316\u6b63\u786e\u7684\u63a8\u7406\u8def\u7ebf\uff0cRFT \u53ea\u9700\u5c11\u91cf\u793a\u4f8b\u5373\u53ef\u663e\u8457\u63d0\u9ad8\u6a21\u578b\u6027\u80fd\u3002<\/p>\n

RFT \u652f\u6301\u7528\u6237\u5229\u7528\u81ea\u5df1\u7684\u9ec4\u91d1\u6570\u636e\u96c6\u521b\u5efa\u72ec\u7279\u7684\u6a21\u578b\uff0c\u5e76\u5c06\u5176\u5e94\u7528\u4e8e\u6cd5\u5f8b\u3001\u91d1\u878d\u3001\u5de5\u7a0b\u3001\u4fdd\u9669\u7b49\u9700\u8981\u4e13\u4e1a\u77e5\u8bc6\u7684\u9886\u57df\u3002<\/p>\n

\u5f3a\u5316\u5fae\u8c03\u9762\u5411\u7fa4\u4f53<\/strong><\/p>\n

OpenAI \u52b1\u7814\u7a76\u673a\u6784\u3001\u9ad8\u6821\u548c\u4f01\u4e1a\u7533\u8bf7\uff0c\u7279\u522b\u662f\u90a3\u4e9b\u76ee\u524d\u7531\u4e13\u5bb6\u9886\u5bfc\u6267\u884c\u4e00\u7cfb\u5217\u72ed\u7a84\u590d\u6742\u4efb\u52a1\uff0c\u5e76\u4e14\u5c06\u53d7\u76ca\u4e8e\u4eba\u5de5\u667a\u80fd\u534f\u52a9\u7684\u673a\u6784\u3002<\/p>\n

OpenAI \u8868\u793a\u5f3a\u5316\u5fae\u8c03\u5728\u7ed3\u679c\u5177\u6709\u5ba2\u89c2\u201c\u6b63\u786e\u201d\u7b54\u6848\uff0c\u4e14\u5927\u591a\u6570\u4e13\u5bb6\u4f1a\u540c\u610f\u7684\u4efb\u52a1\u4e2d\u8868\u73b0\u51fa\u8272\uff0c\u56e0\u6b64\u8ba4\u4e3a\u5728\u6cd5\u5f8b\u3001\u4fdd\u9669\u3001\u533b\u7597\u3001\u91d1\u878d\u3001\u5de5\u7a0b\u7b49\u9886\u57df\u4f1a\u6709\u66f4\u597d\u7684\u8868\u73b0\u3002<\/p>\n

\u53c2\u4e0e\u8005\u53ef\u63d0\u524d\u8bbf\u95ee Alpha \u7248\u5f3a\u5316\u5fae\u8c03 API\uff0c\u5e76\u5728\u7279\u5b9a\u9886\u57df\u4efb\u52a1\u4e2d\u8fdb\u884c\u6d4b\u8bd5\uff0c\u6b64\u5916 OpenAI \u9f13\u52b1\u53c2\u4e0e\u8005\u5206\u4eab\u6570\u636e\u96c6\uff0c\u5171\u540c\u6539\u8fdb OpenAI \u6a21\u578b\u3002<\/p>\n

OpenAI \u9884\u8ba1 2025 \u5e74\u521d\u516c\u5f00\u53d1\u5e03\u5f3a\u5316\u5fae\u8c03\u529f\u80fd\u3002<\/p>\n

OpenAI\u00a0\u9996\u5e2d\u6267\u884c\u5b98\u5c71\u59c6\u30fb\u963f\u5c14\u7279\u66fc\uff08Sam\u00a0Altman\uff09\u8868\u793a\uff1a\u201c\u5f3a\u5316\u5fae\u8c03\uff0c\u6548\u679c\u51fa\u5947\u5730\u597d\uff1b\u5b83\u662f\u6211 2024 \u5e74\u6700\u5927\u7684\u60ca\u559c\u4e4b\u4e00\u3002\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"

December 7, 2012 - OpenAI has launched a 12-day \"shipmas\" release cycle, featuring a series of new features, products, and demos. On the second day of the event, OpenAI launched Reinforcement Fine-Tuning, a program that helps developers and machine learning engineers build expert models for complex domain-specific tasks. With a new model customization technique, the program allows developers to fine-tune models using a high-quality task set and evaluate the model's response using reference answers to improve the model's reasoning power and accuracy for domain-specific tasks. Introduction to Enhanced Fine-Tuning 1AI attached official description: developers are able to use tens to thousands of high-quality task<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[146],"tags":[167,190],"collection":[],"class_list":["post-24642","post","type-post","status-publish","format-standard","hentry","category-news","tag-ai","tag-openai"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/24642","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=24642"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/24642\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=24642"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=24642"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=24642"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=24642"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}