{"id":138,"date":"2023-06-15T17:18:01","date_gmt":"2023-06-15T09:18:01","guid":{"rendered":"https:\/\/myhelpfulai.com\/?p=138"},"modified":"2023-06-15T17:21:44","modified_gmt":"2023-06-15T09:21:44","slug":"%e7%99%be%e5%b7%9d%e6%a8%a1%e5%9e%8b%ef%bc%8c%e4%bb%8a%e6%97%a5%e6%94%be%e5%87%ba","status":"publish","type":"post","link":"https:\/\/aitimes.link\/index.php\/2023\/06\/15\/%e7%99%be%e5%b7%9d%e6%a8%a1%e5%9e%8b%ef%bc%8c%e4%bb%8a%e6%97%a5%e6%94%be%e5%87%ba\/","title":{"rendered":"\u767e\u5ddd\u6a21\u578b\uff0c\u4eca\u65e5\u653e\u51fa"},"content":{"rendered":"\n<p><a href=\"https:\/\/github.com\/baichuan-inc\/baichuan-7B\">baichuan-inc\/baichuan-7B: A large-scale 7B pretraining language model developed by Baichuan (github.com)<\/a><\/p>\n\n\n\n<p>baichuan-7B \u662f\u7531\u767e\u5ddd\u667a\u80fd\u5f00\u53d1\u7684\u4e00\u4e2a\u5f00\u6e90\u53ef\u5546\u7528\u7684\u5927\u89c4\u6a21\u9884\u8bad\u7ec3\u8bed\u8a00\u6a21\u578b\u3002\u57fa\u4e8e Transformer \u7ed3\u6784\uff0c\u5728\u5927\u7ea61.2\u4e07\u4ebf tokens \u4e0a\u8bad\u7ec3\u768470\u4ebf\u53c2\u6570\u6a21\u578b\uff0c\u652f\u6301\u4e2d\u82f1\u53cc\u8bed\uff0c\u4e0a\u4e0b\u6587\u7a97\u53e3\u957f\u5ea6\u4e3a4096\u3002\u5728\u6807\u51c6\u7684\u4e2d\u6587\u548c\u82f1\u6587\u6743\u5a01 benchmark\uff08C-EVAL\/MMLU\uff09\u4e0a\u5747\u53d6\u5f97\u540c\u5c3a\u5bf8\u6700\u597d\u7684\u6548\u679c\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u5206\u8bcd<\/h2>\n\n\n\n<p>\u6211\u4eec\u53c2\u8003\u5b66\u672f\u754c\u65b9\u6848\u4f7f\u7528 SentencePiece \u4e2d\u7684 byte pair encoding (BPE)\u4f5c\u4e3a\u5206\u8bcd\u7b97\u6cd5\uff0c\u5e76\u4e14\u8fdb\u884c\u4e86\u4ee5\u4e0b\u7684\u4f18\u5316\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u76ee\u524d\u5927\u90e8\u5206\u5f00\u6e90\u6a21\u578b\u4e3b\u8981\u57fa\u4e8e\u82f1\u6587\u4f18\u5316\uff0c\u56e0\u6b64\u5bf9\u4e2d\u6587\u8bed\u6599\u5b58\u5728\u6548\u7387\u8f83\u4f4e\u7684\u95ee\u9898\u3002\u6211\u4eec\u4f7f\u75282000\u4e07\u6761\u4ee5\u4e2d\u82f1\u4e3a\u4e3b\u7684\u591a\u8bed\u8a00\u8bed\u6599\u8bad\u7ec3\u5206\u8bcd\u6a21\u578b\uff0c\u663e\u8457\u63d0\u5347\u5bf9\u4e8e\u4e2d\u6587\u7684\u538b\u7f29\u7387\u3002<\/li>\n\n\n\n<li>\u5bf9\u4e8e\u6570\u5b66\u9886\u57df\uff0c\u6211\u4eec\u53c2\u8003\u4e86 LLaMA \u548c Galactica \u4e2d\u7684\u65b9\u6848\uff0c\u5bf9\u6570\u5b57\u7684\u6bcf\u4e00\u4f4d\u5355\u72ec\u5206\u5f00\uff0c\u907f\u514d\u51fa\u73b0\u6570\u5b57\u4e0d\u4e00\u81f4\u7684\u95ee\u9898\uff0c\u5bf9\u4e8e\u63d0\u5347\u6570\u5b66\u80fd\u529b\u6709\u91cd\u8981\u5e2e\u52a9\u3002<\/li>\n\n\n\n<li>\u5bf9\u4e8e\u7f55\u89c1\u5b57\u8bcd\uff08\u5982\u7279\u6b8a\u7b26\u53f7\u7b49\uff09\uff0c\u652f\u6301 UTF-8-characters \u7684 byte \u7f16\u7801\uff0c\u56e0\u6b64\u505a\u5230\u672a\u77e5\u5b57\u8bcd\u7684\u5168\u8986\u76d6\u3002<\/li>\n\n\n\n<li>\u6211\u4eec\u5206\u6790\u4e86\u4e0d\u540c\u5206\u8bcd\u5668\u5bf9\u8bed\u6599\u7684\u538b\u7f29\u7387\uff0c\u5982\u4e0b\u8868\uff0c\u53ef\u89c1\u6211\u4eec\u7684\u5206\u8bcd\u5668\u660e\u663e\u4f18\u4e8e LLaMA, Falcon \u7b49\u5f00\u6e90\u6a21\u578b\uff0c\u5e76\u4e14\u5bf9\u6bd4\u5176\u4ed6\u4e2d\u6587\u5206\u8bcd\u5668\u5728\u538b\u7f29\u7387\u76f8\u5f53\u7684\u60c5\u51b5\u4e0b\uff0c\u8bad\u7ec3\u548c\u63a8\u7406\u6548\u7387\u66f4\u9ad8\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>baichuan-inc\/baichuan-7B: A large-scale 7B pretraining  [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[],"class_list":["post-138","post","type-post","status-publish","format-standard","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/posts\/138","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/comments?post=138"}],"version-history":[{"count":4,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/posts\/138\/revisions"}],"predecessor-version":[{"id":143,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/posts\/138\/revisions\/143"}],"wp:attachment":[{"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/media?parent=138"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/categories?post=138"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/tags?post=138"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}