{"id":10297,"date":"2024-08-03T17:38:05","date_gmt":"2024-08-03T09:38:05","guid":{"rendered":"https:\/\/17aitech.com\/?p=10297"},"modified":"2024-10-08T15:12:06","modified_gmt":"2024-10-08T07:12:06","slug":"%e3%80%90%e8%af%be%e7%a8%8b%e6%80%bb%e7%bb%93%e3%80%91day19%ef%bc%88%e4%b8%8b%ef%bc%89%ef%bc%9atransformer%e6%ba%90%e7%a0%81%e6%b7%b1%e5%85%a5%e7%90%86%e8%a7%a3","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=10297","title":{"rendered":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b"},"content":{"rendered":"<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_85 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-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/17aitech.com\/?p=10297\/#%E5%89%8D%E8%A8%80\" >\u524d\u8a00<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/17aitech.com\/?p=10297\/#%E8%B5%84%E6%96%99\" >\u8d44\u6599<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/17aitech.com\/?p=10297\/#%E6%95%B4%E4%BD%93%E6%A1%86%E6%9E%B6\" >\u6574\u4f53\u6846\u67b6<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/17aitech.com\/?p=10297\/#%E4%BB%A3%E7%A0%81%E5%88%86%E6%9E%90%E7%90%86%E8%A7%A3\" >\u4ee3\u7801\u5206\u6790\u7406\u89e3<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/17aitech.com\/?p=10297\/#%E5%88%9D%E5%A7%8B%E5%8C%96%E6%B5%81%E7%A8%8B\" >\u521d\u59cb\u5316\u6d41\u7a0b<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/17aitech.com\/?p=10297\/#EncoderDecoder_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\" >EncoderDecoder \u521b\u5efa\u8fc7\u7a0b<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/17aitech.com\/?p=10297\/#Encoder_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\" >Encoder \u521b\u5efa\u8fc7\u7a0b<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/17aitech.com\/?p=10297\/#EncoderLayer_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\" >EncoderLayer \u521b\u5efa\u8fc7\u7a0b<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/17aitech.com\/?p=10297\/#Decoder_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\" >Decoder \u521b\u5efa\u8fc7\u7a0b<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-10\" href=\"https:\/\/17aitech.com\/?p=10297\/#DecoderLayer_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\" >DecoderLayer \u521b\u5efa\u8fc7\u7a0b<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-11\" href=\"https:\/\/17aitech.com\/?p=10297\/#MultiHeadedAttention_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\" >MultiHeadedAttention \u521b\u5efa\u8fc7\u7a0b<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-12\" href=\"https:\/\/17aitech.com\/?p=10297\/#PositionwiseFeedForward_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\" >PositionwiseFeedForward \u521b\u5efa\u8fc7\u7a0b<\/a><\/li><\/ul><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-13\" href=\"https:\/\/17aitech.com\/?p=10297\/#%E5%86%85%E5%AE%B9%E5%B0%8F%E7%BB%93\" >\u5185\u5bb9\u5c0f\u7ed3<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-14\" href=\"https:\/\/17aitech.com\/?p=10297\/#%E5%8F%82%E8%80%83%E8%B5%84%E6%96%99\" >\u53c2\u8003\u8d44\u6599<\/a><\/li><\/ul><\/nav><\/div>\n<h2><span class=\"ez-toc-section\" id=\"%E5%89%8D%E8%A8%80\"><\/span>\u524d\u8a00<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>\u5728\u4e0a\u4e00\u7ae0<a href=\"https:\/\/17aitech.com\/?p=9911\">\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u67b6\u6784\u53ca\u6ce8\u610f\u529b\u673a\u5236\u4e86\u89e3<\/a>\u603b\u7ed3\u4e2d\uff0c\u6211\u4eec\u5bf9Transformer\u67b6\u6784\u4ee5\u53ca\u6ce8\u610f\u529b\u673a\u5236\u6709\u4e86\u521d\u6b65\u4e86\u89e3\uff0c\u672c\u7ae0\u5c06\u7ed3\u5408<a href=\"https:\/\/nlp.seas.harvard.edu\/2018\/04\/03\/attention.html\">\u300aThe Annotated Transformer\u300b<\/a>\u7684\u8d44\u6599\u4ee5\u53caTransfomer_demo\u7684\u793a\u4f8b\u4ee3\u7801\uff0c\u5bf9Transformer\u7684\u67b6\u6784\u8fdb\u884c\u6df1\u5165\u7406\u89e3\u3002<\/p>\n<h2><span class=\"ez-toc-section\" id=\"%E8%B5%84%E6%96%99\"><\/span>\u8d44\u6599<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Transformer_demo\u662f\u4e00\u4e2a\u4f7f\u7528Transformer\u67b6\u6784\u8fdb\u884c\u7ffb\u8bd1\u7684\u793a\u4f8b\u4ee3\u7801\uff0c\u8be5\u5de5\u7a0b\u4e0e<a href=\"https:\/\/17aitech.com\/?p=9534\">\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day18\uff1aSeq2Seq\u7684\u6df1\u5165\u4e86\u89e3<\/a>\u4e2d\u7684Seq2Seq\u5b9e\u73b0\u975e\u5e38\u7c7b\u4f3c\uff0c\u53ef\u4ee5\u65b9\u4fbf\u6211\u4eec\u901a\u8fc7\u5355\u6b65\u8c03\u8bd5\u7684\u65b9\u5f0f\uff0c\u52a8\u6001\u5730\u4e86\u89e3Transformer\u67b6\u6784\u3002<\/p>\n<ul>\n<li>\u4ee3\u7801\u5730\u5740\uff1a<a href=\"https:\/\/github.com\/domonic18\/transformer_demo.git\">https:\/\/github.com\/domonic18\/transformer_demo.git<\/a><\/li>\n<\/ul>\n<p>\u5176\u4ed6\u8d44\u6599\uff1a\u300aThe Annotated Transformer\u300b<\/p>\n<ul>\n<li>\u535a\u5ba2\u5730\u5740\uff1a<a href=\"https:\/\/nlp.seas.harvard.edu\/2018\/04\/03\/attention.html\">https:\/\/nlp.seas.harvard.edu\/2018\/04\/03\/attention.html<\/a><\/li>\n<li>Github\u4ed3\u5e93\u5730\u5740\uff1a<a href=\"https:\/\/github.com\/harvardnlp\/annotated-transformer\">https:\/\/github.com\/harvardnlp\/annotated-transformer<\/a><\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"%E6%95%B4%E4%BD%93%E6%A1%86%E6%9E%B6\"><\/span>\u6574\u4f53\u6846\u67b6<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Transformer\u7684\u7406\u89e3.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Transformer\u7684\u7406\u89e3.png\" alt=\"\" \/><\/a><\/p>\n<h2><span class=\"ez-toc-section\" id=\"%E4%BB%A3%E7%A0%81%E5%88%86%E6%9E%90%E7%90%86%E8%A7%A3\"><\/span>\u4ee3\u7801\u5206\u6790\u7406\u89e3<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<h3><span class=\"ez-toc-section\" id=\"%E5%88%9D%E5%A7%8B%E5%8C%96%E6%B5%81%E7%A8%8B\"><\/span>\u521d\u59cb\u5316\u6d41\u7a0b<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u542f\u52a8\u6d41\u7a0b\u65f6\u5e8f\u56fe.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u542f\u52a8\u6d41\u7a0b\u65f6\u5e8f\u56fe.png\" alt=\"\" \/><\/a><\/p>\n<p>Transformer_demo\u7684\u542f\u52a8\u6d41\u7a0b\u4e2d\uff1a<\/p>\n<ul>\n<li>\u7b2c\u4e00\u6b65\uff0c\u5728 <code>main.py<\/code> \u4e2d\u8fdb\u884c <code>Translation()<\/code> \u521d\u59cb\u5316\uff1b<\/li>\n<li>\u7b2c\u4e8c\u6b65\uff0c\u8c03\u7528\u5206\u8bcd\u5668 <code>tokenizer<\/code> \u63d0\u4f9b\u7684\u63a5\u53e3 <code>get_tokenizer()<\/code> \u8fdb\u884c\u5b57\u5178\u7684\u521b\u5efa\uff1b(\u8fd9\u90e8\u5206\u903b\u8f91\u5b9e\u73b0\u4e0e<a href=\"https:\/\/17aitech.com\/?p=9534#toc-5\">\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day18\uff1aSeq2Seq\u7684\u6df1\u5165\u4e86\u89e3<\/a>\u4e00\u81f4\uff0c\u6240\u4ee5\u8fd9\u91cc\u4e0d\u518d\u8d58\u8ff0)<\/li>\n<li>\u7b2c\u4e09\u6b65\uff0c\u8c03\u7528 <code>get_model()<\/code> \u521b\u5efaTransformer\u6a21\u578b\uff1b<\/li>\n<li>\u7b2c\u56db\u6b65\uff0c\u521b\u5efa\u8fc7\u7a0b\u4e2d\uff0c\u4e3b\u8981\u903b\u8f91\u4e3a\uff1a\u5b9e\u4f8b\u5316\u521b\u5efa<code>Encoder<\/code>\u5bf9\u8c61\u548c<code>Decoder<\/code>\u5bf9\u8c61\u3002(\u5b9e\u4f8b\u5316\u4e4b\u524d\uff0c\u63d0\u524d\u51c6\u5907\u4e86\u591a\u5934\u6ce8\u610f\u529b\u5bf9\u8c61<code>attn<\/code>\u3001\u524d\u9988\u7f51\u7edc\u5bf9\u8c61<code>ff<\/code>\u4ee5\u53ca\u4f4d\u7f6e\u7f16\u7801\u5bf9\u8c61<code>position<\/code>)<\/li>\n<li>\u7b2c\u4e94\u6b65\uff0c\u6700\u540e\u521b\u5efa\u751f\u6210\u5668<code>Genearator<\/code>\u5bf9\u8c61\u3002<\/li>\n<\/ul>\n<p>\u5bf9\u5e94\u6e90\u7801\uff1a<\/p>\n<pre><code class=\"language-python\">def get_model(src_vocab,\n              tgt_vocab,\n              N=6,\n              d_model=512,\n              d_ff=2048,\n              h=8,\n              dropout=0.1):\n    &quot;&quot;&quot;\n        \u6784\u5efatransformer\u6a21\u578b\n        src_vocab: \u6e90\u8bed\u8a00\u8bcd\u5178\u5927\u5c0f\n        tgt_vocab: \u76ee\u6807\u8bed\u8a00\u8bcd\u5178\u5927\u5c0f\n        N: \u7f16\u7801\u89e3\u7801\u5c42\u6570\n        d_model: \u6a21\u578b\u7ef4\u5ea6\n        d_ff: \u524d\u5411\u4f20\u64ad\u5c42\u7ef4\u5ea6\n        h: \u591a\u5934\u6ce8\u610f\u529b\u7684\u5934\u6570\n        dropout: \u968f\u673a\u5931\u6d3b\u6982\u7387\n    &quot;&quot;&quot;\n    # \u6df1\u62f7\u8d1d\u51fd\u6570\n    c = copy.deepcopy\n    attn = MultiHeadedAttention(h, d_model, dropout)\n    ff = PositionwiseFeedForward(d_model, d_ff, dropout)\n    position = PositionalEncoding(d_model, dropout)\n    model = EncoderDecoder(\n        Encoder(EncoderLayer(d_model, c(attn), c(ff), dropout), N),\n        Decoder(DecoderLayer(d_model, c(attn), c(attn), c(ff), dropout), N),\n        nn.Sequential(Embeddings(d_model, src_vocab), c(position)),\n        nn.Sequential(Embeddings(d_model, tgt_vocab), c(position)),\n        Generator(d_model, tgt_vocab),\n    )\n\n    # This was important from their code.\n    # Initialize parameters with Glorot \/ fan_avg.\n    for p in model.parameters():\n        if p.dim() &gt; 1:\n            nn.init.xavier_uniform_(p)\n    return model<\/code><\/pre>\n<h4><span class=\"ez-toc-section\" id=\"EncoderDecoder_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\"><\/span>EncoderDecoder \u521b\u5efa\u8fc7\u7a0b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u9996\u5148\uff0c\u6211\u4eec\u4eceEncoderDecoder\u7c7b\u5206\u6790\u4ee3\u7801\uff1a<\/p>\n<pre><code class=\"language-python\">class EncoderDecoder(nn.Module):\n    &quot;&quot;&quot;\n    A standard Encoder-Decoder architecture. Base for this and many\n    other models.\n    &quot;&quot;&quot;\n\n    def __init__(self, encoder, decoder, src_embed, tgt_embed, generator):\n        super(EncoderDecoder, self).__init__()\n        self.encoder = encoder\n        self.decoder = decoder\n        self.src_embed = src_embed\n        self.tgt_embed = tgt_embed\n        self.generator = generator\n\n    # (\u4e0b\u9762\u4ee3\u7801\u6682\u7565)<\/code><\/pre>\n<p>\u4ee3\u7801\u7406\u89e3\uff1a<\/p>\n<ul>\n<li>Transformer\u67b6\u6784\u603b\u4f53\u7531\u4e00\u4e2a <code>EncoderDecoder<\/code> \u7c7b\u6784\u6210\u3002<\/li>\n<li><code>EncoderDecoder<\/code> \u7c7b\u7684\u6210\u5458\u53d8\u91cf\u5305\u542b\u4e94\u90e8\u5206\uff1a\n<ul>\n<li><code>encoder<\/code>\uff1a\u7f16\u7801\u5668\uff0c\u8d1f\u8d23\u5c06\u8f93\u5165\u5e8f\u5217\u7f16\u7801\u4e3a\u56fa\u5b9a\u957f\u5ea6\u7684\u5411\u91cf\u3002<\/li>\n<li><code>decoder<\/code>\uff1a\u89e3\u7801\u5668\uff0c\u8d1f\u8d23\u5c06\u7f16\u7801\u540e\u7684\u5411\u91cf\u89e3\u7801\u4e3a\u8f93\u51fa\u5e8f\u5217\u3002<\/li>\n<li><code>src_embed<\/code>\uff1a\u8f93\u5165\u5e8f\u5217\u7684\u5d4c\u5165\u5c42\uff0c\u5c06\u8f93\u5165\u5e8f\u5217\u8f6c\u6362\u4e3a\u56fa\u5b9a\u7ef4\u5ea6\u7684\u5411\u91cf\u3002<\/li>\n<li><code>tgt_embed<\/code>\uff1a\u8f93\u51fa\u5e8f\u5217\u7684\u5d4c\u5165\u5c42\uff0c\u5c06\u8f93\u51fa\u5e8f\u5217\u8f6c\u6362\u4e3a\u56fa\u5b9a\u7ef4\u5ea6\u7684\u5411\u91cf\u3002<\/li>\n<li><code>generator<\/code>\uff1a\u751f\u6210\u5668\uff0c\u5c06\u89e3\u7801\u540e\u7684\u5411\u91cf\u8f6c\u6362\u4e3a\u8f93\u51fa\u5e8f\u5217\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"Encoder_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\"><\/span>Encoder \u521b\u5efa\u8fc7\u7a0b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u67e5\u770b <code>Encoder<\/code> \u7c7b\u7684\u7ed3\u6784\uff0c\u6e90\u7801\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-python\">class Encoder(nn.Module):\n    &quot;Core encoder is a stack of N layers&quot;\n\n    def __init__(self, layer, N):\n        super(Encoder, self).__init__()\n        self.layers = clones(layer, N)\n        self.norm = LayerNorm(layer.size)\n\n    # (\u4e0b\u9762\u4ee3\u7801\u6682\u7565)<\/code><\/pre>\n<p>\u4ee3\u7801\u7406\u89e3\uff1a<\/p>\n<ul>\n<li><code>Encoder<\/code> \u7c7b\u7684\u521d\u59cb\u5316\u51fd\u6570\u4e2d\uff0c\u5206\u522b\u521b\u5efa\u4e86<code>self.layers<\/code> \u548c <code>self.norm<\/code>\u3002<\/li>\n<li><code>self.layers<\/code> \u7684\u521b\u5efa\u65b9\u5f0f\u662f\u4f7f\u7528<code>clones<\/code>\u51fd\u6570\u8fdb\u884cdeepcopy\u7684\u6279\u91cf\u5316\u521b\u5efa\n<ul>\n<li><code>clones<\/code> \u51fd\u6570\u53ef\u4ee5\u521b\u5efa<strong>\u6307\u5b9a\u6570\u91cf<\/strong>\u7684\u76f8\u540c\u5bf9\u8c61\u7684\u5217\u8868\u3002(Transformer\u7684\u591a\u5c42\u7f51\u7edc\u80fd\u529b\u5373\u7531\u8be5\u51fd\u6570\u4f53\u73b0)<\/li>\n<li><code>self.layers<\/code> \u521b\u5efa\u5e76\u5b9e\u4f8b\u5316\u7684<code>layer<\/code>\u5bf9\u8c61\uff0c\u5176\u7c7b\u578b\u4e3a<code>EncoderLayer<\/code>\u3002<\/li>\n<li><code>Encoder<\/code>\u7684\u524d\u5411\u4f20\u64ad <code>forward<\/code> \u51fd\u6570\u4e2d\uff0c\u4f1a\u4f9d\u6b21\u7ed9\u6bcf\u4e2a<code>EncoderLayer<\/code>\u5bf9\u8c61\u4f20\u5165mask\u4ee5\u4fbf\u8fdb\u884cpad\u63a9\u7801\u64cd\u4f5c\u3002<\/li>\n<\/ul>\n<\/li>\n<li><code>self.norm<\/code> \u5bf9\u5e94\u662f<code>LayerNorm<\/code>\u7c7b\uff0c\u8be5\u7c7b\u7528\u4e8e\u5bf9\u8f93\u5165\u5e8f\u5217\u8fdb\u884c\u5f52\u4e00\u5316\u5904\u7406\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"EncoderLayer_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\"><\/span>EncoderLayer \u521b\u5efa\u8fc7\u7a0b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u56e0\u4e3a <code>Encoder<\/code> \u7c7b\u662f\u7531\u591a\u4e2a <code>EncoderLayer<\/code>\u6784\u6210\uff0c\u6240\u4ee5\u63a5\u7740\u4e86\u89e3<code>EncoderLayer<\/code>\u7c7b\u3002<br \/>\n\u6e90\u7801\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-python\">class EncoderLayer(nn.Module):\n    &quot;Encoder is made up of self-attn and feed forward (defined below)&quot;\n\n    def __init__(self, size, self_attn, feed_forward, dropout):\n        super(EncoderLayer, self).__init__()\n        self.self_attn = self_attn\n        self.feed_forward = feed_forward\n        self.sublayer = clones(SublayerConnection(size, dropout), 2)\n        self.size = size\n\n    # (\u4e0b\u9762\u4ee3\u7801\u6682\u7565)<\/code><\/pre>\n<p>\u4ee3\u7801\u89e3\u6790\uff1a<\/p>\n<ul>\n<li><code>EncoderLayer<\/code> \u7c7b\u4e2d\u5305\u542b4\u4e2a\u6210\u5458\u53d8\u91cf\uff1a\n<ul>\n<li><code>self_attn<\/code>\uff1a\u81ea\u6ce8\u610f\u529b\u673a\u5236\uff0c\u7528\u4e8e\u5bf9\u8f93\u5165\u5e8f\u5217\u8fdb\u884c\u81ea\u6ce8\u610f\u529b\uff0c\u8be5\u6210\u5458\u53d8\u91cf\u521b\u5efa\u65f6\u4f1a\u901a\u8fc7\u516c\u5171\u51fd\u6570attension\u51fd\u6570\u521b\u5efa\uff0c\u5e76\u4f5c\u4e3a\u53c2\u6570\u4f20\u5165\u7ed9<code>self_attn<\/code>\u6210\u5458\u53d8\u91cf\u3002<\/li>\n<li><code>feed_forward<\/code>\uff1a\u524d\u9988\u7f51\u7edc\uff0c\u5bf9\u5e94<code>PositionwiseFeedForward<\/code>\u7c7b\u7684\u5bf9\u8c61\u3002<\/li>\n<li><code>sublayer<\/code>\uff1a\u5305\u542b\u4e24\u4e2a<code>SublayerConnection<\/code>\u7c7b\u7684\u5bf9\u8c61(\u5bf9\u5e94\u56fe\u793a\u4e2d\u7684Add&amp;Norm)\uff0c\u5176\u4f5c\u7528\u662f\u5bf9\u8f93\u5165\u5e8f\u5217\u8fdb\u884c\u5f52\u4e00\u5316\u5904\u7406\u3002<\/li>\n<li><code>size<\/code>\uff1a\u8f93\u5165\u5e8f\u5217\u7684\u7ef4\u5ea6\u5927\u5c0f\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"Decoder_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\"><\/span>Decoder \u521b\u5efa\u8fc7\u7a0b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u67e5\u770b <code>Decoder<\/code> \u7684\u5b9e\u73b0\uff0c\u6e90\u7801\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-python\">class Decoder(nn.Module):\n    &quot;Generic N layer decoder with masking.&quot;\n\n    def __init__(self, layer, N):\n        super(Decoder, self).__init__()\n        self.layers = clones(layer, N)\n        self.norm = LayerNorm(layer.size)\n\n    # (\u4e0b\u9762\u4ee3\u7801\u6682\u7565)<\/code><\/pre>\n<p>\u4ee3\u7801\u7406\u89e3\uff1a<\/p>\n<ul>\n<li><code>Decoder<\/code> \u7c7b\u7684\u521d\u59cb\u5316\u51fd\u6570\u4e2d\u4e0e<code>Encoder<\/code> \u7c7b\u7c7b\u4f3c\uff0c\u521b\u5efa\u4e86<code>self.layers<\/code> \u548c <code>self.norm<\/code>\u3002<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"DecoderLayer_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\"><\/span>DecoderLayer \u521b\u5efa\u8fc7\u7a0b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p><code>Decoder<\/code> \u4e5f\u662f\u7531\u591a\u4e2a<code>DecoderLayer<\/code>\u7ec4\u6210\uff0c\u6240\u4ee5\u67e5\u770b<code>DecoderLayer<\/code>\u7684\u5b9e\u73b0\u3002<br \/>\n\u6e90\u7801\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-python\">class DecoderLayer(nn.Module):\n    &quot;Decoder is made of self-attn, src-attn, and feed forward (defined below)&quot;\n\n    def __init__(self, size, self_attn, src_attn, feed_forward, dropout):\n        super(DecoderLayer, self).__init__()\n        self.size = size\n        self.self_attn = self_attn\n        self.src_attn = src_attn\n        self.feed_forward = feed_forward\n        self.sublayer = clones(SublayerConnection(size, dropout), 3)\n\n    # (\u4e0b\u9762\u4ee3\u7801\u6682\u7565)<\/code><\/pre>\n<p>\u4ee3\u7801\u7406\u89e3\uff1a(\u5982\u4e0b\u56fe\u6240\u793a)<\/p>\n<ul>\n<li><code>Decoder<\/code> \u76f8\u6bd4 <code>Encoder<\/code>\uff1a\n<ul>\n<li>\u591a\u4e861\u4e2a\u591a\u5934\u6ce8\u610f\u529b\u6a21\u5757\uff0c\u5bf9\u5e94\u4ee3\u7801\u4e2d\u7684 <code>src_attn<\/code>\u3002<\/li>\n<li>\u591a\u4e861\u4e2aAdd&amp;Norm\u6a21\u5757\uff0c\u5bf9\u5e94\u4ee3\u7801\u4e2d\u7684 <code>sublayer<\/code> \u7684\u521b\u5efa\u4e2a\u6570\u4e3a <code>3<\/code> \u3002<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Decoder\u7684\u4e09\u4e2alayer\u793a\u4f8b.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/Decoder\u7684\u4e09\u4e2alayer\u793a\u4f8b.png\" alt=\"\" \/><\/a><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"MultiHeadedAttention_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\"><\/span>MultiHeadedAttention \u521b\u5efa\u8fc7\u7a0b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u7531\u4e8eEncoder\u548cDecoder\u4e2d\u90fd\u4f7f\u7528\u4e86\u591a\u5934\u6ce8\u610f\u529b\u5bf9\u8c61\uff0c\u56e0\u6b64\u67e5\u770b\u5176\u6e90\u7801\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-python\">class MultiHeadedAttention(nn.Module):\n    def __init__(self, h, d_model, dropout=0.1):\n        &quot;Take in model size and number of heads.&quot;\n        super(MultiHeadedAttention, self).__init__()\n        assert d_model % h == 0\n        # We assume d_v always equals d_k\n        self.d_k = d_model \/\/ h\n        self.h = h\n        self.linears = clones(nn.Linear(d_model, d_model), 4)\n        self.attn = None\n        self.dropout = nn.Dropout(p=dropout)\n\n    # (\u4e0b\u9762\u4ee3\u7801\u6682\u7565)<\/code><\/pre>\n<p>\u4ee3\u7801\u7406\u89e3\uff1a<br \/>\n\u591a\u5934\u6ce8\u610f\u529b\u5bf9\u8c61\u7684\u6784\u9020\u51fd\u6570\u4e2d\uff0c\u4e3b\u8981\u5305\u542b\u5982\u4e0b\u53c2\u6570\uff1a<\/p>\n<ul>\n<li>h\uff1a\u6ce8\u610f\u529b\u5934\u7684\u6570\u91cf\u3002<\/li>\n<li>d_model\uff1a\u6a21\u578b\u7684\u7ef4\u5ea6\uff08\u5373\u8f93\u5165\u548c\u8f93\u51fa\u7684\u7279\u5f81\u7ef4\u5ea6\uff09\u3002<\/li>\n<li>dropout\uff1a\u4e22\u5f03\u7387\uff0c\u7528\u4e8e\u9632\u6b62\u8fc7\u62df\u5408\u3002<\/li>\n<li>attn\uff1a\u6ce8\u610f\u529b\u5bf9\u8c61\uff0c\u7528\u4e8e\u8ba1\u7b97\u6ce8\u610f\u529b\u5206\u6570\u3002<br \/>\n<blockquote>\n<p>\u5176\u4e2datten\u5728\u6784\u9020\u51fd\u6570\u4e2d\u4e3aNone\uff0c\u5728\u524d\u5411forward\u51fd\u6570\u4e2d\u4f1a\u8c03\u7528attension\u51fd\u6570\u8fdb\u884c\u5b9e\u4f8b\u5316\u5bf9\u8c61\u521b\u5efa\uff0c\u8be5\u90e8\u5206\u5185\u5bb9\u6211\u4eec\u653e\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u68b3\u7406\u7406\u89e3\u3002<\/p>\n<\/blockquote>\n<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"PositionwiseFeedForward_%E5%88%9B%E5%BB%BA%E8%BF%87%E7%A8%8B\"><\/span>PositionwiseFeedForward \u521b\u5efa\u8fc7\u7a0b<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>PositionwiseFeedForward\u5bf9\u5e94Transformer\u6a21\u578b\u4e2d\u7684<code>Feed Forward<\/code>\u90e8\u5206\uff0c\u5176\u5b9e\u73b0\u6bd4\u8f83\u7b80\u5355\uff0c\u6e90\u7801\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-python\">class PositionwiseFeedForward(nn.Module):\n    &quot;Implements FFN equation.&quot;\n\n    def __init__(self, d_model, d_ff, dropout=0.1):\n        super(PositionwiseFeedForward, self).__init__()\n        self.w_1 = nn.Linear(d_model, d_ff)\n        self.w_2 = nn.Linear(d_ff, d_model)\n        self.dropout = nn.Dropout(dropout)\n\n    def forward(self, x):\n        return self.w_2(self.dropout(self.w_1(x).relu()))<\/code><\/pre>\n<p>\u4ee3\u7801\u89e3\u6790\uff1a<\/p>\n<ul>\n<li><code>Feed Forward<\/code>\u529f\u80fd\u6bd4\u8f83\u7b80\u5355\uff0c\u5176\u7ec4\u6210\u5c31\u662f\u4e24\u5c42\u5168\u8fde\u63a5 <code>w_1<\/code> \u548c <code>w_2<\/code>\uff0c\u6700\u540e\u901a\u8fc7 <code>dropout<\/code> \u8fdb\u884c\u968f\u673a\u4e22\u5f03\u4ee5\u63d0\u5347\u6cdb\u5316\u80fd\u529b\u3002<\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"%E5%86%85%E5%AE%B9%E5%B0%8F%E7%BB%93\"><\/span>\u5185\u5bb9\u5c0f\u7ed3<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<ul>\n<li>\u9996\u5148\uff0c\u4ee3\u7801\u4e2d\u5728<code>get_model()<\/code>\u51fd\u6570\u4e2d\u8c03\u7528 <code>EncoderDecoder<\/code> \u521d\u59cb\u5316\u65b9\u6cd5\uff0c\u6784\u5efa <code>Transformer<\/code> \u6a21\u578b\u3002<\/li>\n<li>\u5176\u6b21\uff0c<code>Transformer<\/code> \u6a21\u578b\u7531 <code>Encoder<\/code> \u548c <code>Decoder<\/code> \u7ec4\u6210\uff0c<code>Encoder<\/code> \u548c <code>Decoder<\/code> \u90fd\u7531\u591a\u4e2a <code>EncoderLayer<\/code> \u548c <code>DecoderLayer<\/code> \u7ec4\u6210\u3002<\/li>\n<li>\u7136\u540e\uff0c<code>EncoderLayer<\/code> \u548c <code>DecoderLayer<\/code> \u90fd\u7531 <code>MultiHeadedAttention<\/code> \u548c <code>PositionwiseFeedForward<\/code> \u7ec4\u6210\u3002<\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"%E5%8F%82%E8%80%83%E8%B5%84%E6%96%99\"><\/span>\u53c2\u8003\u8d44\u6599<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><a href=\"https:\/\/nlp.seas.harvard.edu\/2018\/04\/03\/attention.html\">The Annotated Transformer<\/a><\/p>\n<p align=\"center\">\u6b22\u8fce\u5173\u6ce8\u516c\u4f17\u53f7\u4ee5\u83b7\u5f97\u6700\u65b0\u7684\u6587\u7ae0\u548c\u65b0\u95fb<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/09\/\u626b\u7801_\u641c\u7d22\u8054\u5408\u4f20\u64ad\u6837\u5f0f-\u767d\u8272\u7248.bmp\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/09\/\u626b\u7801_\u641c\u7d22\u8054\u5408\u4f20\u64ad\u6837\u5f0f-\u767d\u8272\u7248.bmp\" alt=\"\" \/><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u524d\u8a00 \u5728\u4e0a\u4e00\u7ae0\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":10011,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"aside","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","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":"default","adv-header-id-meta":"","stick-header-meta":"default","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","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":[28],"tags":[68],"class_list":["post-10297","post","type-post","status-publish","format-aside","has-post-thumbnail","hentry","category-blog","tag-68","post_format-post-format-aside"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.9 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b - \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=10297\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297\"},\"author\":{\"name\":\"Dongming\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/#\\\/schema\\\/person\\\/3d23bb6f7f115fcefc9ae7803a691739\"},\"headline\":\"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b\",\"datePublished\":\"2024-08-03T09:38:05+00:00\",\"dateModified\":\"2024-10-08T07:12:06+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297\"},\"wordCount\":85,\"commentCount\":0,\"image\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/\u5c01\u9762_transformer.png\",\"keywords\":[\"\u5927\u6a21\u578b\"],\"articleSection\":[\"\u535a\u5ba2\u6587\u7ae0\"],\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/17aitech.com\\\/?p=10297#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297\",\"url\":\"https:\\\/\\\/17aitech.com\\\/?p=10297\",\"name\":\"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/\u5c01\u9762_transformer.png\",\"datePublished\":\"2024-08-03T09:38:05+00:00\",\"dateModified\":\"2024-10-08T07:12:06+00:00\",\"author\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/#\\\/schema\\\/person\\\/3d23bb6f7f115fcefc9ae7803a691739\"},\"breadcrumb\":{\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/17aitech.com\\\/?p=10297\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297#primaryimage\",\"url\":\"https:\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/\u5c01\u9762_transformer.png\",\"contentUrl\":\"https:\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/2024\\\/08\\\/\u5c01\u9762_transformer.png\",\"width\":1080,\"height\":608},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/17aitech.com\\\/?p=10297#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\\\/\\\/17aitech.com\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b\"}]},{\"@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\\\/3d23bb6f7f115fcefc9ae7803a691739\",\"name\":\"Dongming\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/member\\\/avatars\\\/238a0b923820dcc5.1732798681.jpg\",\"url\":\"\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/member\\\/avatars\\\/238a0b923820dcc5.1732798681.jpg\",\"contentUrl\":\"\\\/\\\/17aitech.com\\\/wp-content\\\/uploads\\\/member\\\/avatars\\\/238a0b923820dcc5.1732798681.jpg\",\"caption\":\"Dongming\"},\"description\":\"\u89c1\u5929\u5730\uff0c\u89c1\u4f17\u751f\uff0c\u89c1\u81ea\u5df1\u3002\",\"sameAs\":[\"http:\\\/\\\/17aitech.com\"],\"url\":\"https:\\\/\\\/17aitech.com\\\/?page_id=33738&user=1\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b - \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=10297","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/17aitech.com\/?p=10297#article","isPartOf":{"@id":"https:\/\/17aitech.com\/?p=10297"},"author":{"name":"Dongming","@id":"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739"},"headline":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b","datePublished":"2024-08-03T09:38:05+00:00","dateModified":"2024-10-08T07:12:06+00:00","mainEntityOfPage":{"@id":"https:\/\/17aitech.com\/?p=10297"},"wordCount":85,"commentCount":0,"image":{"@id":"https:\/\/17aitech.com\/?p=10297#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u5c01\u9762_transformer.png","keywords":["\u5927\u6a21\u578b"],"articleSection":["\u535a\u5ba2\u6587\u7ae0"],"inLanguage":"zh-Hans","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/17aitech.com\/?p=10297#respond"]}]},{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=10297","url":"https:\/\/17aitech.com\/?p=10297","name":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=10297#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=10297#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u5c01\u9762_transformer.png","datePublished":"2024-08-03T09:38:05+00:00","dateModified":"2024-10-08T07:12:06+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=10297#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=10297"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=10297#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u5c01\u9762_transformer.png","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/08\/\u5c01\u9762_transformer.png","width":1080,"height":608},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=10297#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011day19\uff08\u4e0b\uff09\uff1aTransformer\u6e90\u7801\u6df1\u5165\u7406\u89e3\u4e4b\u6784\u5efa\u6a21\u578b\u6d41\u7a0b"}]},{"@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\/3d23bb6f7f115fcefc9ae7803a691739","name":"Dongming","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg","url":"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg","contentUrl":"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg","caption":"Dongming"},"description":"\u89c1\u5929\u5730\uff0c\u89c1\u4f17\u751f\uff0c\u89c1\u81ea\u5df1\u3002","sameAs":["http:\/\/17aitech.com"],"url":"https:\/\/17aitech.com\/?page_id=33738&user=1"}]}},"_links":{"self":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/10297","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\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=10297"}],"version-history":[{"count":7,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/10297\/revisions"}],"predecessor-version":[{"id":33227,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/10297\/revisions\/33227"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/media\/10011"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=10297"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=10297"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=10297"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}