{"id":3287,"date":"2025-05-15T21:29:27","date_gmt":"2025-05-15T13:29:27","guid":{"rendered":"https:\/\/www.15zhi.net\/blog\/?p=3287"},"modified":"2025-05-15T21:29:27","modified_gmt":"2025-05-15T13:29:27","slug":"202505%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-the-alchemist-automated-labeling-500x-cheaper-than-llm-data-annotators","status":"publish","type":"post","link":"https:\/\/www.15zhi.net\/blog\/202505%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-the-alchemist-automated-labeling-500x-cheaper-than-llm-data-annotators\/","title":{"rendered":"202505\u8bba\u6587\u7814\u8bfb-The ALCHEmist: Automated Labeling 500x CHEaper Than LLM Data Annotators"},"content":{"rendered":"\n<p>\u4f5c\u8005\uff1aTzu-Heng Huang, Catherine Cao, Vaishnavi Bhargava\u7b49<\/p>\n\n\n\n<p>\u5355\u4f4d\uff1aUniversity of Wisconsin-Madison<br>\u6765\u6e90\uff1aarXiv<br>\u53d1\u8868\u65e5\u671f\uff1a2025.2<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u4e00\u3001\u80cc\u666f\u52a8\u673a<\/h2>\n\n\n\n<p>1.\u4f20\u7edf\u4f7f\u7528\u5927\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u76f4\u63a5\u751f\u6210\u6807\u7b7e\u7684\u65b9\u5f0f\u6210\u672c\u9ad8\u6602\uff0c\u6bcf\u6761\u6570\u636e\u90fd\u9700\u8c03\u7528\u4e00\u6b21\u6a21\u578b\u63a5\u53e3\uff0c\u96be\u4ee5\u9002\u5e94\u5927\u89c4\u6a21\u6807\u6ce8\u9700\u6c42\u3002<\/p>\n\n\n\n<p>2.\u73b0\u6709\u6a21\u578b\u751f\u6210\u7684\u6807\u7b7e\u7ed3\u679c\u4e0d\u53ef\u89e3\u91ca\u3001\u4e0d\u53ef\u8ffd\u8e2a\u4e14\u4e0d\u6613\u6269\u5c55\uff0c\u4e00\u65e6\u6807\u7b7e\u6807\u51c6\u6216\u4efb\u52a1\u53d8\u5316\uff0c\u9700\u8981\u91cd\u65b0\u8c03\u7528\u6a21\u578b\u91cd\u65b0\u6807\u6ce8\u3002<\/p>\n\n\n\n<p>3.\u7a0b\u5e8f\u5316\u6807\u6ce8\u903b\u8f91\u5c1a\u672a\u88ab\u5145\u5206\u5229\u7528\uff0c\u5c06\u6a21\u578b\u751f\u6210\u7684\u601d\u8def\u8f6c\u5316\u4e3a\u53ef\u590d\u7528\u3001\u53ef\u5ba1\u8ba1\u7684\u4ee3\u7801\u7a0b\u5e8f\uff0c\u6709\u6f5c\u529b\u5728\u6210\u672c\u63a7\u5236\u3001\u53ef\u7ef4\u62a4\u6027\u548c\u6807\u6ce8\u4e00\u81f4\u6027\u4e0a\u5e26\u6765\u91cd\u5927\u6539\u8fdb\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u4e8c\u3001\u6838\u5fc3\u5185\u5bb9<\/h2>\n\n\n\n<p>\u7cfb\u7edf\u6574\u4f53\u67b6\u6784\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"268\" src=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-89-1024x268.png\" alt=\"\" class=\"wp-image-3288\" srcset=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-89-1024x268.png 1024w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-89-300x78.png 300w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-89-768x201.png 768w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-89.png 1156w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p><strong>1.LLM\u751f\u6210\u6807\u6ce8\u7a0b\u5e8f<\/strong><\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"567\" src=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-90-1024x567.png\" alt=\"\" class=\"wp-image-3289\" srcset=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-90-1024x567.png 1024w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-90-300x166.png 300w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-90-768x425.png 768w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-90.png 1370w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>2.\u5f31\u76d1\u7763\u805a\u5408<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"557\" src=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-91-1024x557.png\" alt=\"\" class=\"wp-image-3290\" srcset=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-91-1024x557.png 1024w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-91-300x163.png 300w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-91-768x418.png 768w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-91.png 1434w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>3.\u6a21\u6001\u6269\u5c55\u4e0e\u6cdb\u5316\u80fd\u529b<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"852\" height=\"439\" src=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-92.png\" alt=\"\" class=\"wp-image-3291\" srcset=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-92.png 852w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-92-300x155.png 300w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-92-768x396.png 768w\" sizes=\"auto, (max-width: 852px) 100vw, 852px\" \/><\/figure>\n\n\n\n<p>Alchemist \u5728\u56fe\u50cf\u4efb\u52a1\u4e2d\u7684\u5904\u7406\u6d41\u7a0b\u5982\u4e0a\u6240\u793a\uff0c\u901a\u8fc7\u8bed\u8a00\u6a21\u578b\u63d0\u53d6\u8bed\u4e49\u6982\u5ff5\uff0c\u7528\u56fe\u50cf\u6a21\u578b\u63d0\u53d6\u56fe\u50cf\u7279\u5f81\uff0c\u8ba1\u7b97\u76f8\u4f3c\u5ea6\u540e\u4f5c\u4e3a\u7a0b\u5e8f\u8f93\u5165\uff0c\u5b9e\u73b0\u4f4e\u6210\u672c\u7684\u590d\u6742\u6a21\u6001\u6807\u6ce8<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u4e09\u3001\u5b9e\u9a8c\u8bc4\u4f30<\/h2>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1014\" height=\"364\" src=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-93.png\" alt=\"\" class=\"wp-image-3292\" srcset=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-93.png 1014w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-93-300x108.png 300w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-93-768x276.png 768w\" sizes=\"auto, (max-width: 1014px) 100vw, 1014px\" \/><\/figure>\n\n\n\n<p>\u4e0d\u540c\u63d0\u793a\u7b56\u7565\u4e0b\uff0c\u4f7f\u7528 GPT-3.5\u3001GPT-4 \u548c Claude 3 \u6240\u751f\u6210\u7684\u6807\u6ce8\u7a0b\u5e8f\u5728\u5404\u6570\u636e\u96c6\u4e0a\u7684\u6027\u80fd\u5bf9\u6bd4<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"208\" src=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-94-1024x208.png\" alt=\"\" class=\"wp-image-3293\" srcset=\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-94-1024x208.png 1024w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-94-300x61.png 300w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-94-768x156.png 768w, https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2025\/05\/image-94.png 1068w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>\u5728\u4e0d\u540c\u6570\u636e\u96c6\u4e0a\uff0c\u968f\u7740\u6240\u6536\u96c6\u7684\u6807\u6ce8\u7a0b\u5e8f\u6570\u91cf\u589e\u52a0\uff0c\u6807\u7b7e\u6a21\u578b\u6027\u80fd\uff08F1-score \u6216 Accuracy\uff09\u9010\u6b65\u63d0\u5347\uff0c\u9a8c\u8bc1\u591a\u6837\u6027\u7b56\u7565\u7684\u6709\u6548\u6027\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u56db\u3001\u603b\u7ed3\u601d\u8003<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">\u8bba\u6587\u603b\u7ed3\uff1a<\/h4>\n\n\n\n<p><\/p>\n\n\n\n<p><strong>\u6838\u5fc3\u65b9\u6cd5\uff1a<\/strong>\u672c\u6587\u63d0\u51fa\u4e86\u4e00\u79cd\u4f4e\u6210\u672c\u3001\u9ad8\u53ef\u63a7\u6027\u7684\u81ea\u52a8\u6807\u6ce8\u7cfb\u7edf Alchemist\uff0c\u5c06\u6807\u7b7e\u751f\u6210\u4efb\u52a1\u8f6c\u5316\u4e3a\u7a0b\u5e8f\u751f\u6210\u4efb\u52a1\u3002\u901a\u8fc7 LLM \u751f\u6210\u53ef\u590d\u7528\u7684\u6807\u6ce8\u7a0b\u5e8f\uff0c\u5e76\u7ed3\u5408\u5f31\u76d1\u7763\u65b9\u6cd5\u878d\u5408\u591a\u7a0b\u5e8f\u8f93\u51fa\uff0c\u5b9e\u73b0\u4e86\u9ad8\u8d28\u91cf\u4f2a\u6807\u7b7e\u7684\u751f\u6210\u3002\u5b9e\u9a8c\u8868\u660e\uff0cAlchemist \u5728\u591a\u4e2a\u6587\u672c\u4e0e\u56fe\u50cf\u4efb\u52a1\u4e2d\u6027\u80fd\u4f18\u8d8a\uff0c\u5e73\u5747\u6210\u672c\u964d\u4f4e\u7ea6 500 \u500d\uff0c\u90e8\u5206\u4efb\u52a1\u51c6\u786e\u7387\u63d0\u5347 12.9%\u3002<\/p>\n\n\n\n<p><strong>\u4f18\u52bf\u603b\u7ed3\uff1a<\/strong><\/p>\n\n\n\n<p>1.\u5927\u5e45\u964d\u4f4e API \u8c03\u7528\u6210\u672c\uff0c\u9002\u7528\u4e8e\u5927\u89c4\u6a21\u6807\u6ce8\u4efb\u52a1<br>2.\u7a0b\u5e8f\u53ef\u590d\u7528\u3001\u53ef\u5ba1\u8ba1\uff0c\u589e\u5f3a\u6807\u7b7e\u751f\u6210\u7684\u53ef\u63a7\u6027<br>3.\u5f31\u76d1\u7763\u878d\u5408\u673a\u5236\u63d0\u5347\u6807\u7b7e\u8d28\u91cf\u4e0e\u4e00\u81f4\u6027<br>4.\u652f\u6301\u591a\u6a21\u6001\u4efb\u52a1\u6269\u5c55\uff0c\u5177\u5907\u826f\u597d\u7684\u6cdb\u5316\u80fd\u529b<\/p>\n\n\n\n<p><strong>\u542f\u53d1\u601d\u8003\uff1a<\/strong><\/p>\n\n\n\n<p>1.<strong>\u6807\u7b7e\u7a0b\u5e8f\u7ed3\u6784\u5316\uff1a<\/strong>Alchemist \u5c06\u6807\u7b7e\u903b\u8f91\u5c01\u88c5\u4e3a\u53ef\u6267\u884c\u7a0b\u5e8f\uff0c\u5b9e\u73b0\u4e86\u4ece\u201c\u6807\u7b7e\u7ed3\u679c\u201d\u5411\u201c\u6807\u7b7e\u903b\u8f91\u8868\u8fbe\u201d\u7684\u8303\u5f0f\u8f6c\u53d8\uff0c\u5177\u5907\u53ef\u89e3\u91ca\u3001\u53ef\u590d\u7528\u3001\u53ef\u7ec4\u5408\u7684\u4f18\u52bf\u3002<br>2.<strong>\u591a\u7a0b\u5e8f+\u5f31\u76d1\u7763\u673a\u5236\u501f\u9274\u4ef7\u503c<\/strong>\uff1a\u5229\u7528\u591a\u4e2a\u6807\u6ce8\u89c6\u89d2\u5efa\u6784\u6807\u7b7e\u5224\u65ad\u4f53\u7cfb\uff0c\u518d\u901a\u8fc7\u6807\u7b7e\u6a21\u578b\u8fdb\u884c\u878d\u5408\uff0c\u5951\u5408\u8f6f\u6807\u7b7e\u5bf9\u591a\u6e90\u5171\u8bc6\u548c\u7a33\u5b9a\u6027\u7684\u9700\u6c42\u3002<br>3.<strong>\u77e5\u8bc6\u8f93\u5165\u7ed3\u6784\u5316\u8d8b\u52bf\uff1a<\/strong>\u8be5\u6846\u67b6\u53cd\u6620\u4e86\u201c\u5148\u5efa\u6807\u7b7e\u903b\u8f91\uff0c\u518d\u63d0\u4f9b\u6807\u7b7e\u4fe1\u606f\u201d\u7684\u6d41\u7a0b\uff0c\u5bf9\u6784\u5efa\u7ed3\u6784\u5316\u3001\u53ef\u63a7\u7684\u77e5\u8bc6\u8f93\u5165\u63a5\u53e3\u5177\u6709\u542f\u53d1\u610f\u4e49\u3002<\/p>\n\n\n\n<p><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u4f5c\u8005\uff1aTzu-Heng Huang, Catherine Cao, Vaishnavi Bhargava\u7b49 \u5355 [&hellip;]<\/p>\n","protected":false},"author":67,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-3287","post","type-post","status-publish","format-standard","hentry","category-events"],"_links":{"self":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3287","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/users\/67"}],"replies":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/comments?post=3287"}],"version-history":[{"count":1,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3287\/revisions"}],"predecessor-version":[{"id":3294,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3287\/revisions\/3294"}],"wp:attachment":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/media?parent=3287"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/categories?post=3287"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/tags?post=3287"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}