Hugging Face's logo

Buckets:
hf-doc-build
/
doc

Files
xet
hf-doc-build/doc / transformers /main /pt /tasks /image_classification.html
mishig's picture
mishig
download
raw
61.5 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Classificação de imagens&quot;,&quot;local&quot;:&quot;image-classification&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Carregar o conjunto de dados Food-101&quot;,&quot;local&quot;:&quot;load-food-101-dataset&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Pré-processar&quot;,&quot;local&quot;:&quot;preprocess&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Avaliar&quot;,&quot;local&quot;:&quot;evaluate&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Treinar&quot;,&quot;local&quot;:&quot;train&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Inferência&quot;,&quot;local&quot;:&quot;inference&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
<link href="/docs/transformers/main/en/_app/immutable/entry/start.ByZEd4qm.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/DHdg-_0Z.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/CBySwR1V.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/entry/app.CQqZS61p.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/Cdb1BouH.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/dUGfULTE.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/DOLBYqcr.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/nodes/0.CEy2b_T8.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/Conj1S5Z.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/nodes/736.Bu4H1KsX.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/DoQ7MZQh.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/B6o9tG7M.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/DnCf4Kx5.js" rel="modulepreload">
<link href="/docs/transformers/main/en/_app/immutable/chunks/BqE_oZe8.js" rel="modulepreload">
<!--1bkvgze--><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Image classification&quot;,&quot;local&quot;:&quot;image-classification&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Load Food-101 dataset&quot;,&quot;local&quot;:&quot;load-food-101-dataset&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Preprocess&quot;,&quot;local&quot;:&quot;preprocess&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Evaluate&quot;,&quot;local&quot;:&quot;evaluate&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Train&quot;,&quot;local&quot;:&quot;train&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Inference&quot;,&quot;local&quot;:&quot;inference&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/><!---->
<link href="/docs/transformers/main/en/_app/immutable/assets/0.tn0RQdqM.css" rel="modulepreload"> <!--[--><!--[0--><!--[--><!--[0--><!--[--><p></p> <div class="items-center shrink-0 min-w-[100px] max-sm:min-w-[50px] justify-end ml-auto flex" style="float: inline-end; margin-inline-start: 10px; display: inline-flex; position: relative; z-index: 10;"><div dir="ltr" class="inline-flex rounded-md max-sm:rounded-sm"><button class="inline-flex items-center gap-1 h-7 max-sm:h-7 px-2 max-sm:px-1.5 text-sm font-medium text-gray-800 border border-r-0 rounded-l-md max-sm:rounded-l-sm border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-live="polite"><span class="inline-flex items-center justify-center rounded-md p-0.5 max-sm:p-0 hover:text-gray-800 dark:hover:text-gray-200"><svg class="sm:size-3.5 size-3" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----></span> <span dir="auto">Copy page</span></button> <button class="inline-flex items-center justify-center w-6 max-sm:w-5 h-7 max-sm:h-7 disabled:pointer-events-none text-sm text-gray-500 hover:text-gray-700 dark:hover:text-white rounded-r-md max-sm:rounded-r-sm border border-l transition border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-haspopup="menu" aria-expanded="false" aria-label="Open copy menu"><svg class="transition-transform text-gray-400 overflow-visible sm:size-3.5 size-3 rotate-0" width="1em" height="1em" viewBox="0 0 12 7" fill="none" xmlns="http://www.w3.org/2000/svg"><path d="M1 1L6 6L11 1" stroke="currentColor"></path></svg><!----></button></div> <!--[-1--><!--]--></div><!----> <div class="flex space-x-1 " style="float: inline-end; margin-inline-start: 10px; display: inline-flex; position: relative; z-index: 10;" dir="ltr"><!----> <!--[1--><div class="relative colab-dropdown "><button class=" " type="button"><!--[0--><!--[-1--><img alt="Open In Colab" class="!m-0" src="https://colab.research.google.com/assets/colab-badge.svg"/><!--]--><!----><!--]--></button> <!--[-1--><!--]--></div><!--]--> <!--[1--><div class="relative colab-dropdown "><button class=" " type="button"><!--[0--><!--[-1--><img alt="Open In Studio Lab" class="!m-0" src="https://studiolab.sagemaker.aws/studiolab.svg"/><!--]--><!----><!--]--></button> <!--[-1--><!--]--></div><!--]--></div><!----> <!--[0--><h1 class="relative group"><a id="image-classification" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#image-classification"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span><!---->Classificação de imagens<!----></span></h1><!--]--><!----> <iframe class="w-full xl:w-4/6 h-80" src="https://www.youtube-nocookie.com/embed/tjAIM7BOYhw" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen=""></iframe><!----> <p>A classificação de imagens atribui um rótulo ou classe a uma imagem. Diferentemente da classificação de texto ou áudio, as entradas são os valores de pixels que compõem uma imagem. Existem muitas aplicações para a classificação de imagens, como detectar danos após um desastre natural, monitorar a saúde das culturas ou ajudar a triar imagens médicas em busca de sinais de doenças.</p> <p>Este guia ilustra como:</p> <ol><li>Ajuste fino de <a href="../model_doc/vit">ViT</a> no conjunto de dados <a href="https://huggingface.co/datasets/ethz/food101" rel="nofollow">Food-101</a> para classificar um item de comida em uma imagem.</li> <li>Use seu modelo ajustado para inferência.</li></ol> <blockquote class="tip"><p>Para ver todas as arquiteturas e checkpoints compatíveis com esta tarefa, recomendamos consultar a <a href="https://huggingface.co/tasks/image-classification" rel="nofollow">página da tarefa</a></p><!----></blockquote><!----> <p>Antes de começar, verifique se você tem todas as bibliotecas necessárias instaladas:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-bash "><!---->pip install transformers datasets evaluate accelerate pillow torchvision scikit-learn trackio<!----></pre></div><!----> <p>Recomendamos que você faça login na sua conta do Hugging Face para enviar e compartilhar seu modelo com a comunidade. Quando solicitado, insira seu token para fazer login:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> huggingface_hub <span class="hljs-keyword">import</span> notebook_login
<span class="hljs-meta">&gt;&gt;&gt; </span>notebook_login()<!----></pre></div><!----> <!--[1--><h2 class="relative group"><a id="load-food-101-dataset" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#load-food-101-dataset"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span><!---->Carregar o conjunto de dados Food-101<!----></span></h2><!--]--><!----> <p>Comece carregando um subconjunto menor do conjunto de dados Food-101 a partir da biblioteca 🤗 Datasets. Isso lhe dará a oportunidade de experimentar e garantir que tudo funcione antes de gastar mais tempo treinando no conjunto de dados completo.</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> load_dataset
<span class="hljs-meta">&gt;&gt;&gt; </span>food = load_dataset(<span class="hljs-string">&quot;ethz/food101&quot;</span>, split=<span class="hljs-string">&quot;train[:5000]&quot;</span>)<!----></pre></div><!----> <p>Divida o split <code>train</code> do conjunto de dados em um conjunto de treinamento e um conjunto de teste com o método <a href="https://huggingface.co/docs/datasets/main/en/package_reference/main_classes#datasets.Dataset.train_test_split" rel="nofollow">train_test_split</a>:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>food = food.train_test_split(test_size=<span class="hljs-number">0.2</span>)<!----></pre></div><!----> <p>Em seguida, veja um exemplo:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>food[<span class="hljs-string">&quot;train&quot;</span>][<span class="hljs-number">0</span>]
{<span class="hljs-string">&#x27;image&#x27;</span>: &lt;PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=512x512 at <span class="hljs-number">0x7F52AFC8AC50</span>&gt;,
<span class="hljs-string">&#x27;label&#x27;</span>: <span class="hljs-number">79</span>}<!----></pre></div><!----> <p>Cada exemplo no conjunto de dados tem dois campos:</p> <ul><li><code>image</code>: uma imagem PIL do item de comida</li> <li><code>label</code>: a classe de rótulo do item de comida</li></ul> <p>Para facilitar a obtenção do nome do rótulo a partir do id do rótulo pelo modelo, crie um dicionário que mapeia o nome do rótulo para um inteiro e vice-versa:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>labels = food[<span class="hljs-string">&quot;train&quot;</span>].features[<span class="hljs-string">&quot;label&quot;</span>].names
<span class="hljs-meta">&gt;&gt;&gt; </span>label2id, id2label = <span class="hljs-built_in">dict</span>(), <span class="hljs-built_in">dict</span>()
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">for</span> i, label <span class="hljs-keyword">in</span> <span class="hljs-built_in">enumerate</span>(labels):
<span class="hljs-meta">... </span> label2id[label] = <span class="hljs-built_in">str</span>(i)
<span class="hljs-meta">... </span> id2label[<span class="hljs-built_in">str</span>(i)] = label<!----></pre></div><!----> <p>Agora você pode converter o id do rótulo em um nome de rótulo:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>id2label[<span class="hljs-built_in">str</span>(<span class="hljs-number">79</span>)]
<span class="hljs-string">&#x27;prime_rib&#x27;</span><!----></pre></div><!----> <!--[1--><h2 class="relative group"><a id="preprocess" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#preprocess"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span><!---->Pré-processar<!----></span></h2><!--]--><!----> <p>O próximo passo é carregar um processador de imagens ViT para processar a imagem em um tensor:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoImageProcessor
<span class="hljs-meta">&gt;&gt;&gt; </span>checkpoint = <span class="hljs-string">&quot;google/vit-base-patch16-224-in21k&quot;</span>
<span class="hljs-meta">&gt;&gt;&gt; </span>image_processor = AutoImageProcessor.from_pretrained(checkpoint)<!----></pre></div><!----> <p>Aplique algumas transformações de imagem às imagens para tornar o modelo mais robusto contra overfitting. Aqui, você usará o módulo <a href="https://pytorch.org/vision/stable/transforms.html" rel="nofollow"><code>transforms</code></a> do torchvision, mas também pode usar qualquer biblioteca de imagem que preferir.</p> <p>Recorte uma parte aleatória da imagem, redimensione-a e normalize-a com a média e o desvio padrão da imagem:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> torchvision.transforms <span class="hljs-keyword">import</span> RandomResizedCrop, Compose, Normalize, ToTensor
<span class="hljs-meta">&gt;&gt;&gt; </span>normalize = Normalize(mean=image_processor.image_mean, std=image_processor.image_std)
<span class="hljs-meta">&gt;&gt;&gt; </span>size = (
<span class="hljs-meta">... </span> image_processor.size[<span class="hljs-string">&quot;shortest_edge&quot;</span>]
<span class="hljs-meta">... </span> <span class="hljs-keyword">if</span> <span class="hljs-string">&quot;shortest_edge&quot;</span> <span class="hljs-keyword">in</span> image_processor.size
<span class="hljs-meta">... </span> <span class="hljs-keyword">else</span> (image_processor.size[<span class="hljs-string">&quot;height&quot;</span>], image_processor.size[<span class="hljs-string">&quot;width&quot;</span>])
<span class="hljs-meta">... </span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>_transforms = Compose([RandomResizedCrop(size), ToTensor(), normalize])<!----></pre></div><!----> <p>Em seguida, crie uma função de pré-processamento para aplicar as transformações e retornar os <code>pixel_values</code> — as entradas do modelo — da imagem:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">def</span> <span class="hljs-title function_">transforms</span>(<span class="hljs-params">examples</span>):
<span class="hljs-meta">... </span> examples[<span class="hljs-string">&quot;pixel_values&quot;</span>] = [_transforms(img.convert(<span class="hljs-string">&quot;RGB&quot;</span>)) <span class="hljs-keyword">for</span> img <span class="hljs-keyword">in</span> examples[<span class="hljs-string">&quot;image&quot;</span>]]
<span class="hljs-meta">... </span> <span class="hljs-keyword">del</span> examples[<span class="hljs-string">&quot;image&quot;</span>]
<span class="hljs-meta">... </span> <span class="hljs-keyword">return</span> examples<!----></pre></div><!----> <p>Para aplicar a função de pré-processamento em todo o conjunto de dados, use o método <a href="https://huggingface.co/docs/datasets/main/en/package_reference/main_classes#datasets.Dataset.with_transform" rel="nofollow">with_transform</a> do 🤗 Datasets. As transformações são aplicadas em tempo de execução ao carregar um elemento do conjunto de dados:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>food = food.with_transform(transforms)<!----></pre></div><!----> <p>Agora, crie um lote de exemplos usando <a href="/docs/transformers/main/pt/main_classes/data_collator#transformers.DefaultDataCollator">DefaultDataCollator</a>. Diferentemente de outros data collators no 🤗 Transformers, o <code>DefaultDataCollator</code> não aplica pré-processamento adicional, como padding.</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> DefaultDataCollator
<span class="hljs-meta">&gt;&gt;&gt; </span>data_collator = DefaultDataCollator()<!----></pre></div><!----> <!--[1--><h2 class="relative group"><a id="evaluate" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#evaluate"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span><!---->Avaliar<!----></span></h2><!--]--><!----> <p>Incluir uma métrica durante o treinamento é frequentemente útil para avaliar o desempenho do seu modelo. Você pode carregar rapidamente um método de avaliação com a biblioteca 🤗 <a href="https://huggingface.co/docs/evaluate/index" rel="nofollow">Evaluate</a>. Para esta tarefa, carregue a métrica <a href="https://huggingface.co/spaces/evaluate-metric/accuracy" rel="nofollow">accuracy</a> (veja o <a href="https://huggingface.co/docs/evaluate/a_quick_tour" rel="nofollow">quick tour</a> do 🤗 Evaluate para saber mais sobre como carregar e calcular uma métrica):</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">import</span> evaluate
<span class="hljs-meta">&gt;&gt;&gt; </span>accuracy = evaluate.load(<span class="hljs-string">&quot;accuracy&quot;</span>)<!----></pre></div><!----> <p>Em seguida, crie uma função que passa suas previsões e rótulos para <a href="https://huggingface.co/docs/evaluate/main/en/package_reference/main_classes#evaluate.EvaluationModule.compute" rel="nofollow">compute</a> para calcular a precisão:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">import</span> numpy <span class="hljs-keyword">as</span> np
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">def</span> <span class="hljs-title function_">compute_metrics</span>(<span class="hljs-params">eval_pred</span>):
<span class="hljs-meta">... </span> predictions, labels = eval_pred
<span class="hljs-meta">... </span> predictions = np.argmax(predictions, axis=<span class="hljs-number">1</span>)
<span class="hljs-meta">... </span> <span class="hljs-keyword">return</span> accuracy.compute(predictions=predictions, references=labels)<!----></pre></div><!----> <p>Sua função <code>compute_metrics</code> está pronta e você voltará a ela ao configurar o treinamento.</p> <!--[1--><h2 class="relative group"><a id="train" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#train"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span><!---->Treinar<!----></span></h2><!--]--><!----> <blockquote class="tip"><p>Se você não está familiarizado com o ajuste fino de um modelo usando o <a href="/docs/transformers/main/pt/main_classes/trainer#transformers.Trainer">Trainer</a>, confira o tutorial básico <a href="../training">aqui</a>!</p><!----></blockquote><!----> <p>Agora você está pronto para começar o treinamento do seu modelo! Carregue o ViT com <a href="/docs/transformers/main/pt/model_doc/auto#transformers.AutoModelForImageClassification">AutoModelForImageClassification</a>. Especifique o número de rótulos, junto com o número de rótulos esperados e os mapeamentos de rótulos:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModelForImageClassification, TrainingArguments, Trainer
<span class="hljs-meta">&gt;&gt;&gt; </span>model = AutoModelForImageClassification.from_pretrained(
<span class="hljs-meta">... </span> checkpoint,
<span class="hljs-meta">... </span> num_labels=<span class="hljs-built_in">len</span>(labels),
<span class="hljs-meta">... </span> id2label=id2label,
<span class="hljs-meta">... </span> label2id=label2id,
<span class="hljs-meta">... </span>)<!----></pre></div><!----> <p>Neste ponto, restam apenas três etapas:</p> <ol><li>Defina os hiperparâmetros de treinamento em <a href="/docs/transformers/main/pt/main_classes/trainer#transformers.TrainingArguments">TrainingArguments</a>. É importante não remover colunas não utilizadas, pois isso descartaria a coluna <code>image</code>. Sem a coluna <code>image</code>, não é possível criar <code>pixel_values</code>. Defina <code>remove_unused_columns=False</code> para evitar esse comportamento! O único outro parâmetro obrigatório é <code>output_dir</code>, que especifica onde salvar o modelo. Você enviará este modelo para o Hub definindo <code>push_to_hub=True</code> (é necessário estar conectado ao Hugging Face para enviar o modelo). No final de cada epoch, o <a href="/docs/transformers/main/pt/main_classes/trainer#transformers.Trainer">Trainer</a> avaliará a precisão e salvará o checkpoint de treinamento.</li> <li>Passe os argumentos de treinamento para o <a href="/docs/transformers/main/pt/main_classes/trainer#transformers.Trainer">Trainer</a> junto com o modelo, o conjunto de dados, o tokenizer, o data collator e a função <code>compute_metrics</code>.</li> <li>Chame <a href="/docs/transformers/main/pt/main_classes/trainer#transformers.Trainer.train">train()</a> para fazer o ajuste fino do seu modelo.</li></ol> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>training_args = TrainingArguments(
<span class="hljs-meta">... </span> output_dir=<span class="hljs-string">&quot;my_awesome_food_model&quot;</span>,
<span class="hljs-meta">... </span> remove_unused_columns=<span class="hljs-literal">False</span>,
<span class="hljs-meta">... </span> eval_strategy=<span class="hljs-string">&quot;epoch&quot;</span>,
<span class="hljs-meta">... </span> save_strategy=<span class="hljs-string">&quot;epoch&quot;</span>,
<span class="hljs-meta">... </span> learning_rate=<span class="hljs-number">5e-5</span>,
<span class="hljs-meta">... </span> per_device_train_batch_size=<span class="hljs-number">16</span>,
<span class="hljs-meta">... </span> gradient_accumulation_steps=<span class="hljs-number">4</span>,
<span class="hljs-meta">... </span> per_device_eval_batch_size=<span class="hljs-number">16</span>,
<span class="hljs-meta">... </span> num_train_epochs=<span class="hljs-number">3</span>,
<span class="hljs-meta">... </span> warmup_steps=<span class="hljs-number">0.1</span>,
<span class="hljs-meta">... </span> logging_steps=<span class="hljs-number">10</span>,
<span class="hljs-meta">... </span> report_to=<span class="hljs-string">&quot;trackio&quot;</span>,
<span class="hljs-meta">... </span> run_name=<span class="hljs-string">&quot;food101&quot;</span>,
<span class="hljs-meta">... </span> load_best_model_at_end=<span class="hljs-literal">True</span>,
<span class="hljs-meta">... </span> metric_for_best_model=<span class="hljs-string">&quot;accuracy&quot;</span>,
<span class="hljs-meta">... </span> push_to_hub=<span class="hljs-literal">True</span>,
<span class="hljs-meta">... </span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>trainer = Trainer(
<span class="hljs-meta">... </span> model=model,
<span class="hljs-meta">... </span> args=training_args,
<span class="hljs-meta">... </span> data_collator=data_collator,
<span class="hljs-meta">... </span> train_dataset=food[<span class="hljs-string">&quot;train&quot;</span>],
<span class="hljs-meta">... </span> eval_dataset=food[<span class="hljs-string">&quot;test&quot;</span>],
<span class="hljs-meta">... </span> processing_class=image_processor,
<span class="hljs-meta">... </span> compute_metrics=compute_metrics,
<span class="hljs-meta">... </span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>trainer.train()<!----></pre></div><!----> <p>Quando o treinamento for concluído, compartilhe seu modelo no Hub com o método <a href="/docs/transformers/main/pt/main_classes/trainer#transformers.Trainer.push_to_hub">push_to_hub()</a> para que todos possam usá-lo:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>trainer.push_to_hub()<!----></pre></div><!----> <blockquote class="tip"><p>Para um exemplo mais detalhado de como fazer o ajuste fino de um modelo para classificação de imagens, confira o <a href="https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/image_classification.ipynb" rel="nofollow">notebook do PyTorch</a> correspondente.</p><!----></blockquote><!----> <!--[1--><h2 class="relative group"><a id="inference" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#inference"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span><!---->Inferência<!----></span></h2><!--]--><!----> <p>Ótimo, agora que você ajustou um modelo, pode usá-lo para inferência!</p> <p>Carregue uma imagem na qual você deseja executar inferência:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>ds = load_dataset(<span class="hljs-string">&quot;ethz/food101&quot;</span>, split=<span class="hljs-string">&quot;validation[:10]&quot;</span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>image = ds[<span class="hljs-string">&quot;image&quot;</span>][<span class="hljs-number">0</span>]<!----></pre></div><!----> <div class="flex justify-center"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png" alt="image of beignets"/></div> <p>A maneira mais simples de testar seu modelo ajustado para inferência é usá-lo em um <a href="/docs/transformers/main/pt/main_classes/pipelines#transformers.pipeline">pipeline()</a>. Instancie um <code>pipeline</code> para classificação de imagem com seu modelo e passe sua imagem para ele:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> pipeline
<span class="hljs-meta">&gt;&gt;&gt; </span>classifier = pipeline(<span class="hljs-string">&quot;image-classification&quot;</span>, model=<span class="hljs-string">&quot;my_awesome_food_model&quot;</span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>classifier(image)
[{<span class="hljs-string">&#x27;score&#x27;</span>: <span class="hljs-number">0.31856709718704224</span>, <span class="hljs-string">&#x27;label&#x27;</span>: <span class="hljs-string">&#x27;beignets&#x27;</span>},
{<span class="hljs-string">&#x27;score&#x27;</span>: <span class="hljs-number">0.015232225880026817</span>, <span class="hljs-string">&#x27;label&#x27;</span>: <span class="hljs-string">&#x27;bruschetta&#x27;</span>},
{<span class="hljs-string">&#x27;score&#x27;</span>: <span class="hljs-number">0.01519392803311348</span>, <span class="hljs-string">&#x27;label&#x27;</span>: <span class="hljs-string">&#x27;chicken_wings&#x27;</span>},
{<span class="hljs-string">&#x27;score&#x27;</span>: <span class="hljs-number">0.013022331520915031</span>, <span class="hljs-string">&#x27;label&#x27;</span>: <span class="hljs-string">&#x27;pork_chop&#x27;</span>},
{<span class="hljs-string">&#x27;score&#x27;</span>: <span class="hljs-number">0.012728818692266941</span>, <span class="hljs-string">&#x27;label&#x27;</span>: <span class="hljs-string">&#x27;prime_rib&#x27;</span>}]<!----></pre></div><!----> <p>Você também pode replicar manualmente os resultados do <code>pipeline</code>, se desejar:</p> <p>Carregue um processador de imagem para pré-processar a imagem e retornar o <code>input</code> como tensores do PyTorch:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoImageProcessor
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">import</span> torch
<span class="hljs-meta">&gt;&gt;&gt; </span>image_processor = AutoImageProcessor.from_pretrained(<span class="hljs-string">&quot;my_awesome_food_model&quot;</span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>inputs = image_processor(image, return_tensors=<span class="hljs-string">&quot;pt&quot;</span>)<!----></pre></div><!----> <p>Passe suas entradas para o modelo e retorne os logits:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModelForImageClassification
<span class="hljs-meta">&gt;&gt;&gt; </span>model = AutoModelForImageClassification.from_pretrained(<span class="hljs-string">&quot;my_awesome_food_model&quot;</span>)
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">with</span> torch.no_grad():
<span class="hljs-meta">... </span> logits = model(**inputs).logits<!----></pre></div><!----> <p>Obtenha o rótulo previsto com a maior probabilidade e use o mapeamento <code>id2label</code> do modelo para convertê-lo em um rótulo:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copiado</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-meta">&gt;&gt;&gt; </span>predicted_label = logits.argmax(-<span class="hljs-number">1</span>).item()
<span class="hljs-meta">&gt;&gt;&gt; </span>model.config.id2label[predicted_label]
<span class="hljs-string">&#x27;beignets&#x27;</span><!----></pre></div><!----> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/transformers/blob/main/docs/source/en/tasks/image_classification.md" target="_blank"><svg class="mr-1" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M31,16l-7,7l-1.41-1.41L28.17,16l-5.58-5.59L24,9l7,7z"></path><path d="M1,16l7-7l1.41,1.41L3.83,16l5.58,5.59L8,23l-7-7z"></path><path d="M12.419,25.484L17.639,6.552l1.932,0.518L14.351,26.002z"></path></svg><!----> <span><span class="underline">Update</span> on GitHub</span></a><!----> <p></p><!--]--><!----><!--]--><!--]--><!--]--> <!--[-1--><!--]--><!--]-->
<script>
{
__sveltekit_1y3v6y9 = {
base: "/docs/transformers/main/en",
assets: "/docs/transformers/main/en"
};
const element = document.currentScript.parentElement;
Promise.all([
import("/docs/transformers/main/en/_app/immutable/entry/start.ByZEd4qm.js"),
import("/docs/transformers/main/en/_app/immutable/entry/app.CQqZS61p.js")
]).then(([kit, app]) => {
kit.start(app, element, {
node_ids: [0, 736],
data: [null,null],
form: null,
error: null
});
});
}
</script>

Xet Storage Details

Size:
61.5 kB
·
Xet hash:
25ff96befda42eecf84782b61274eaac072e7ff4a2c1e0a663df13ad70a7e7eb

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.