Buckets:

hf-doc-build
/

doc

Files

xet

hf-doc-build/doc / transformers /main /pt /tasks /token_classification.html

HuggingFaceDocBuilder

about 1 hour ago

download

raw

40.5 kB

	<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"Classificação de tokens","local":"classificação-de-tokens","sections":[{"title":"Carregando o conjunto de dados WNUT 17","local":"carregando-o-conjunto-de-dados-wnut-17","sections":[],"depth":2},{"title":"Pré-processamento","local":"pré-processamento","sections":[],"depth":2},{"title":"Treinamento","local":"treinamento","sections":[],"depth":2}],"depth":1}">
	<link href="/docs/transformers/main/pt/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/entry/start.4a508f6b.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/scheduler.6452e238.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/singletons.c252e54b.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/paths.cc541a7b.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/entry/app.deac39ac.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/preload-helper.9c9c815c.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/index.8a08b073.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/nodes/0.e26da176.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/each.e59479a4.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/nodes/13.0a5c7dcd.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/Tip.d0d2eb35.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/MermaidChart.svelte_svelte_type_style_lang.86f31263.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/Youtube.ec59a37f.js">
	<link rel="modulepreload" href="/docs/transformers/main/pt/_app/immutable/chunks/CodeBlock.159023b8.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{"title":"Classificação de tokens","local":"classificação-de-tokens","sections":[{"title":"Carregando o conjunto de dados WNUT 17","local":"carregando-o-conjunto-de-dados-wnut-17","sections":[],"depth":2},{"title":"Pré-processamento","local":"pré-processamento","sections":[],"depth":2},{"title":"Treinamento","local":"treinamento","sections":[],"depth":2}],"depth":1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <div class="items-center shrink-0 min-w-[100px] max-sm:min-w-[50px] justify-end ml-auto flex" style="float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"><div 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>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> </div> <h1 class="relative group"><a id="classificação-de-tokens" 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="#classificação-de-tokens"><span><svg class="" 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 tokens</span></h1> <iframe class="w-full xl:w-4/6 h-80" src="https://www.youtube-nocookie.com/embed/wVHdVlPScxA" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> <p data-svelte-h="svelte-2rm2e9">A classificação de tokens atribui um rótulo a tokens individuais em uma frase. Uma das tarefas de classificação de tokens mais comuns é o Reconhecimento de Entidade Nomeada, também chamada de NER (sigla em inglês para Named Entity Recognition). O NER tenta encontrar um rótulo para cada entidade em uma frase, como uma pessoa, local ou organização.</p> <p data-svelte-h="svelte-kmg05x">Este guia mostrará como realizar o fine-tuning do <a href="https://huggingface.co/distilbert/distilbert-base-uncased" rel="nofollow">DistilBERT</a> no conjunto de dados <a href="https://huggingface.co/datasets/wnut_17" rel="nofollow">WNUT 17</a> para detectar novas entidades.</p> <blockquote class="tip"><p data-svelte-h="svelte-dbkroz">Consulte a <a href="https://huggingface.co/tasks/token-classification" rel="nofollow">página de tarefas de classificação de tokens</a> para obter mais informações sobre outras formas de classificação de tokens e seus modelos, conjuntos de dados e métricas associadas.</p></blockquote> <h2 class="relative group"><a id="carregando-o-conjunto-de-dados-wnut-17" 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="#carregando-o-conjunto-de-dados-wnut-17"><span><svg class="" 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>Carregando o conjunto de dados WNUT 17</span></h2> <p data-svelte-h="svelte-3wpxsw">Carregue o conjunto de dados WNUT 17 da biblioteca 🤗 Datasets:</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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> load_dataset

	<span class="hljs-meta">>>> </span>wnut = load_dataset(<span class="hljs-string">"wnut_17"</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1vbpr8z">E dê uma olhada em 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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span>wnut[<span class="hljs-string">"train"</span>][<span class="hljs-number">0</span>]
	{<span class="hljs-string">'id'</span>: <span class="hljs-string">'0'</span>,
	<span class="hljs-string">'ner_tags'</span>: [<span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">7</span>, <span class="hljs-number">8</span>, <span class="hljs-number">8</span>, <span class="hljs-number">0</span>, <span class="hljs-number">7</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>, <span class="hljs-number">0</span>],
	<span class="hljs-string">'tokens'</span>: [<span class="hljs-string">'@paulwalk'</span>, <span class="hljs-string">'It'</span>, <span class="hljs-string">"'s"</span>, <span class="hljs-string">'the'</span>, <span class="hljs-string">'view'</span>, <span class="hljs-string">'from'</span>, <span class="hljs-string">'where'</span>, <span class="hljs-string">'I'</span>, <span class="hljs-string">"'m"</span>, <span class="hljs-string">'living'</span>, <span class="hljs-string">'for'</span>, <span class="hljs-string">'two'</span>, <span class="hljs-string">'weeks'</span>, <span class="hljs-string">'.'</span>, <span class="hljs-string">'Empire'</span>, <span class="hljs-string">'State'</span>, <span class="hljs-string">'Building'</span>, <span class="hljs-string">'='</span>, <span class="hljs-string">'ESB'</span>, <span class="hljs-string">'.'</span>, <span class="hljs-string">'Pretty'</span>, <span class="hljs-string">'bad'</span>, <span class="hljs-string">'storm'</span>, <span class="hljs-string">'here'</span>, <span class="hljs-string">'last'</span>, <span class="hljs-string">'evening'</span>, <span class="hljs-string">'.'</span>]
	}<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1cz4103">Cada número em <code>ner_tags</code> representa uma entidade. Converta o número em um rótulo para obter mais informações:</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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span>label_list = wnut[<span class="hljs-string">"train"</span>].features[<span class="hljs-string">f"ner_tags"</span>].feature.names
	<span class="hljs-meta">>>> </span>label_list
	[
	<span class="hljs-string">"O"</span>,
	<span class="hljs-string">"B-corporation"</span>,
	<span class="hljs-string">"I-corporation"</span>,
	<span class="hljs-string">"B-creative-work"</span>,
	<span class="hljs-string">"I-creative-work"</span>,
	<span class="hljs-string">"B-group"</span>,
	<span class="hljs-string">"I-group"</span>,
	<span class="hljs-string">"B-location"</span>,
	<span class="hljs-string">"I-location"</span>,
	<span class="hljs-string">"B-person"</span>,
	<span class="hljs-string">"I-person"</span>,
	<span class="hljs-string">"B-product"</span>,
	<span class="hljs-string">"I-product"</span>,
	]<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-cqfv4y">O <code>ner_tag</code> descreve uma entidade, como uma organização, local ou pessoa. A letra que prefixa cada <code>ner_tag</code> indica a posição do token da entidade:</p> <ul data-svelte-h="svelte-13k6qsm"><li><code>B-</code> indica o início de uma entidade.</li> <li><code>I-</code> indica que um token está contido dentro da mesma entidade (por exemplo, o token <code>State</code> pode fazer parte de uma entidade como <code>Empire State Building</code>).</li> <li><code>0</code> indica que o token não corresponde a nenhuma entidade.</li></ul> <h2 class="relative group"><a id="pré-processamento" 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="#pré-processamento"><span><svg class="" 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-processamento</span></h2> <iframe class="w-full xl:w-4/6 h-80" src="https://www.youtube-nocookie.com/embed/iY2AZYdZAr0" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> <p data-svelte-h="svelte-1e5ex1w">Carregue o tokenizer do DistilBERT para processar os <code>tokens</code>:</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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer

	<span class="hljs-meta">>>> </span>tokenizer = AutoTokenizer.from_pretrained(<span class="hljs-string">"distilbert/distilbert-base-uncased"</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-g8rrel">Como a entrada já foi dividida em palavras, defina <code>is_split_into_words=True</code> para tokenizar as palavras em subpalavras:</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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span>tokenized_input = tokenizer(example[<span class="hljs-string">"tokens"</span>], is_split_into_words=<span class="hljs-literal">True</span>)
	<span class="hljs-meta">>>> </span>tokens = tokenizer.convert_ids_to_tokens(tokenized_input[<span class="hljs-string">"input_ids"</span>])
	<span class="hljs-meta">>>> </span>tokens
	[<span class="hljs-string">'[CLS]'</span>, <span class="hljs-string">'@'</span>, <span class="hljs-string">'paul'</span>, <span class="hljs-string">'##walk'</span>, <span class="hljs-string">'it'</span>, <span class="hljs-string">"'"</span>, <span class="hljs-string">'s'</span>, <span class="hljs-string">'the'</span>, <span class="hljs-string">'view'</span>, <span class="hljs-string">'from'</span>, <span class="hljs-string">'where'</span>, <span class="hljs-string">'i'</span>, <span class="hljs-string">"'"</span>, <span class="hljs-string">'m'</span>, <span class="hljs-string">'living'</span>, <span class="hljs-string">'for'</span>, <span class="hljs-string">'two'</span>, <span class="hljs-string">'weeks'</span>, <span class="hljs-string">'.'</span>, <span class="hljs-string">'empire'</span>, <span class="hljs-string">'state'</span>, <span class="hljs-string">'building'</span>, <span class="hljs-string">'='</span>, <span class="hljs-string">'es'</span>, <span class="hljs-string">'##b'</span>, <span class="hljs-string">'.'</span>, <span class="hljs-string">'pretty'</span>, <span class="hljs-string">'bad'</span>, <span class="hljs-string">'storm'</span>, <span class="hljs-string">'here'</span>, <span class="hljs-string">'last'</span>, <span class="hljs-string">'evening'</span>, <span class="hljs-string">'.'</span>, <span class="hljs-string">'[SEP]'</span>]<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-17o23dm">Ao adicionar os tokens especiais <code>[CLS]</code> e <code>[SEP]</code> e a tokenização de subpalavras uma incompatibilidade é gerada entre a entrada e os rótulos. Uma única palavra correspondente a um único rótulo pode ser dividida em duas subpalavras. Você precisará realinhar os tokens e os rótulos da seguinte forma:</p> <ol data-svelte-h="svelte-9yvtqn"><li>Mapeie todos os tokens para a palavra correspondente com o método <a href="https://huggingface.co/docs/tokenizers/python/latest/api/reference.html#tokenizers.Encoding.word_ids" rel="nofollow"><code>word_ids</code></a>.</li> <li>Atribuindo o rótulo <code>-100</code> aos tokens especiais <code>[CLS]</code> e <code>[SEP]</code> para que a função de loss do PyTorch ignore eles.</li> <li>Rotular apenas o primeiro token de uma determinada palavra. Atribuindo <code>-100</code> a outros subtokens da mesma palavra.</li></ol> <p data-svelte-h="svelte-576loj">Aqui está como você pode criar uma função para realinhar os tokens e rótulos e truncar sequências para não serem maiores que o comprimento máximo de entrada do DistilBERT:</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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span><span class="hljs-keyword">def</span> <span class="hljs-title function_">tokenize_and_align_labels</span>(<span class="hljs-params">examples</span>):
	<span class="hljs-meta">... </span> tokenized_inputs = tokenizer(examples[<span class="hljs-string">"tokens"</span>], truncation=<span class="hljs-literal">True</span>, is_split_into_words=<span class="hljs-literal">True</span>)

	<span class="hljs-meta">... </span> labels = []
	<span class="hljs-meta">... </span> <span class="hljs-keyword">for</span> i, label <span class="hljs-keyword">in</span> <span class="hljs-built_in">enumerate</span>(examples[<span class="hljs-string">f"ner_tags"</span>]):
	<span class="hljs-meta">... </span> word_ids = tokenized_inputs.word_ids(batch_index=i) <span class="hljs-comment"># Map tokens to their respective word.</span>
	<span class="hljs-meta">... </span> previous_word_idx = <span class="hljs-literal">None</span>
	<span class="hljs-meta">... </span> label_ids = []
	<span class="hljs-meta">... </span> <span class="hljs-keyword">for</span> word_idx <span class="hljs-keyword">in</span> word_ids: <span class="hljs-comment"># Set the special tokens to -100.</span>
	<span class="hljs-meta">... </span> <span class="hljs-keyword">if</span> word_idx <span class="hljs-keyword">is</span> <span class="hljs-literal">None</span>:
	<span class="hljs-meta">... </span> label_ids.append(-<span class="hljs-number">100</span>)
	<span class="hljs-meta">... </span> <span class="hljs-keyword">elif</span> word_idx != previous_word_idx: <span class="hljs-comment"># Only label the first token of a given word.</span>
	<span class="hljs-meta">... </span> label_ids.append(label[word_idx])
	<span class="hljs-meta">... </span> <span class="hljs-keyword">else</span>:
	<span class="hljs-meta">... </span> label_ids.append(-<span class="hljs-number">100</span>)
	<span class="hljs-meta">... </span> previous_word_idx = word_idx
	<span class="hljs-meta">... </span> labels.append(label_ids)

	<span class="hljs-meta">... </span> tokenized_inputs[<span class="hljs-string">"labels"</span>] = labels
	<span class="hljs-meta">... </span> <span class="hljs-keyword">return</span> tokenized_inputs<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1dnx1ri">Use a função <a href="https://huggingface.co/docs/datasets/process#map" rel="nofollow"><code>map</code></a> do 🤗 Datasets para tokenizar e alinhar os rótulos em todo o conjunto de dados. Você pode acelerar a função <code>map</code> configurando <code>batched=True</code> para processar vários elementos do conjunto de dados de uma só vez:</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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span>tokenized_wnut = wnut.<span class="hljs-built_in">map</span>(tokenize_and_align_labels, batched=<span class="hljs-literal">True</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-yw3ehb">Use o <code>DataCollatorForTokenClassification</code> para criar um batch de exemplos. Ele também <em>preencherá dinamicamente</em> seu texto e rótulos para o comprimento do elemento mais longo em seu batch, para que tenham um comprimento uniforme. Embora seja possível preencher seu texto na função <code>tokenizer</code> configurando <code>padding=True</code>, o preenchimento dinâmico é mais eficiente.</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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> DataCollatorForTokenClassification

	<span class="hljs-meta">>>> </span>data_collator = DataCollatorForTokenClassification(tokenizer=tokenizer)<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="treinamento" 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="#treinamento"><span><svg class="" 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>Treinamento</span></h2> <p data-svelte-h="svelte-1vwp4j2">Carregue o DistilBERT com o <code>AutoModelForTokenClassification</code> junto com o número de rótulos esperados:</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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModelForTokenClassification, TrainingArguments, Trainer

	<span class="hljs-meta">>>> </span>model = AutoModelForTokenClassification.from_pretrained(<span class="hljs-string">"distilbert/distilbert-base-uncased"</span>, num_labels=<span class="hljs-number">14</span>)<!-- HTML_TAG_END --></pre></div> <blockquote class="tip"><p data-svelte-h="svelte-796e4s">Se você não estiver familiarizado com o fine-tuning de um modelo com o <code>Trainer</code>, dê uma olhada no tutorial básico <a href="../training#finetune-with-trainer">aqui</a>!</p></blockquote> <p data-svelte-h="svelte-1dd35iq">Nesse ponto, restam apenas três passos:</p> <ol data-svelte-h="svelte-1aiypan"><li>Definir seus hiperparâmetros de treinamento em <code>TrainingArguments</code>.</li> <li>Passar os argumentos de treinamento para o <code>Trainer</code> junto com o modelo, conjunto de dados, tokenizador e o data collator.</li> <li>Chamar a função <code>train()</code> para executar o fine-tuning 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 class="" 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> Copied</div></button></div> <pre class="language-py "><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span>training_args = TrainingArguments(
	<span class="hljs-meta">... </span> output_dir=<span class="hljs-string">"./results"</span>,
	<span class="hljs-meta">... </span> eval_strategy=<span class="hljs-string">"epoch"</span>,
	<span class="hljs-meta">... </span> learning_rate=<span class="hljs-number">2e-5</span>,
	<span class="hljs-meta">... </span> per_device_train_batch_size=<span class="hljs-number">16</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> weight_decay=<span class="hljs-number">0.01</span>,
	<span class="hljs-meta">... </span>)

	<span class="hljs-meta">>>> </span>trainer = Trainer(
	<span class="hljs-meta">... </span> model=model,
	<span class="hljs-meta">... </span> args=training_args,
	<span class="hljs-meta">... </span> train_dataset=tokenized_wnut[<span class="hljs-string">"train"</span>],
	<span class="hljs-meta">... </span> eval_dataset=tokenized_wnut[<span class="hljs-string">"test"</span>],
	<span class="hljs-meta">... </span> processing_class=tokenizer,
	<span class="hljs-meta">... </span> data_collator=data_collator,
	<span class="hljs-meta">... </span>)

	<span class="hljs-meta">>>> </span>trainer.train()<!-- HTML_TAG_END --></pre></div> <blockquote class="tip"><p data-svelte-h="svelte-1evwr8u">Para obter um exemplo mais aprofundado de como executar o fine-tuning de um modelo para classificação de tokens, dê uma olhada nesse <a href="https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/token_classification.ipynb" rel="nofollow">notebook utilizando PyTorch</a> ou nesse <a href="https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/token_classification-tf.ipynb" rel="nofollow">notebook utilizando TensorFlow</a>.</p></blockquote> <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/pt/tasks/token_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 data-svelte-h="svelte-zjs2n5"><span class="underline">Update</span> on GitHub</span></a> <p></p>

	<script>
	{
	__sveltekit_1eathqs = {
	assets: "/docs/transformers/main/pt",
	base: "/docs/transformers/main/pt",
	env: {}
	};

	const element = document.currentScript.parentElement;

	const data = [null,null];

	Promise.all([
	import("/docs/transformers/main/pt/_app/immutable/entry/start.4a508f6b.js"),
	import("/docs/transformers/main/pt/_app/immutable/entry/app.deac39ac.js")
	]).then(([kit, app]) => {
	kit.start(app, element, {
	node_ids: [0, 13],
	data,
	form: null,
	error: null
	});
	});
	}
	</script>

Xet Storage Details

Size:: 40.5 kB
Xet hash:: 65b6037f0fbd532ad2e16b4d3d050a3930bfba2eb6965e5184ae0090eef294e9

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