Hugging Face's logo

Buckets:
hf-doc-build
/
doc-dev

Files
xet
hf-doc-build/doc-dev / course /pr_1149 /te /chapter2 /5.html
rtrm's picture
rtrm
download
raw
54.3 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;బహుళ సీక్వెన్స్‌ల నిర్వహణ&quot;,&quot;local&quot;:&quot;handling-multiple-sequences&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;మోడళ్లు batch రూపంలో ఇన్‌పుట్‌లను ఆశిస్తాయి&quot;,&quot;local&quot;:&quot;models-expect-a-batch-of-inputs&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;ఇన్‌పుట్‌లను padding చేయడం&quot;,&quot;local&quot;:&quot;padding-the-inputs&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Attention masks&quot;,&quot;local&quot;:&quot;attention-masks&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;పొడవైన సీక్వెన్స్‌లు&quot;,&quot;local&quot;:&quot;longer-sequences&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
 <link href="/docs/course/pr_1149/te/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/entry/start.eee4c1e0.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/scheduler.cc52f4b9.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/singletons.69904a63.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/index.5033808a.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/paths.e8418a83.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/entry/app.0f38558a.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/preload-helper.ae923e55.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/index.bd400c31.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/nodes/0.bd814f6d.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/each.e59479a4.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/nodes/18.62864160.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/Tip.272e6bd8.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/MermaidChart.svelte_svelte_type_style_lang.25fe1ea6.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/Youtube.4afc5ee9.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/CodeBlock.1794e33c.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/CourseFloatingBanner.6f780cc4.js">
 <link rel="modulepreload" href="/docs/course/pr_1149/te/_app/immutable/chunks/FrameworkSwitchCourse.e8e44fa0.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;బహుళ సీక్వెన్స్‌ల నిర్వహణ&quot;,&quot;local&quot;:&quot;handling-multiple-sequences&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;మోడళ్లు batch రూపంలో ఇన్‌పుట్‌లను ఆశిస్తాయి&quot;,&quot;local&quot;:&quot;models-expect-a-batch-of-inputs&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;ఇన్‌పుట్‌లను padding చేయడం&quot;,&quot;local&quot;:&quot;padding-the-inputs&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Attention masks&quot;,&quot;local&quot;:&quot;attention-masks&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;పొడవైన సీక్వెన్స్‌లు&quot;,&quot;local&quot;:&quot;longer-sequences&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <div class="bg-white leading-none border border-gray-100 rounded-lg flex p-0.5 w-56 text-sm mb-4"><a class="flex justify-center flex-1 py-1.5 px-2.5 focus:outline-none !no-underline rounded-l bg-red-50 dark:bg-transparent text-red-600" href="?fw=pt"><svg class="mr-1.5" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><defs><clipPath id="a"><rect x="3.05" y="0.5" width="25.73" height="31" fill="none"></rect></clipPath></defs><g clip-path="url(#a)"><path d="M24.94,9.51a12.81,12.81,0,0,1,0,18.16,12.68,12.68,0,0,1-18,0,12.81,12.81,0,0,1,0-18.16l9-9V5l-.84.83-6,6a9.58,9.58,0,1,0,13.55,0ZM20.44,9a1.68,1.68,0,1,1,1.67-1.67A1.68,1.68,0,0,1,20.44,9Z" fill="#ee4c2c"></path></g></svg> Pytorch </a><a class="flex justify-center flex-1 py-1.5 px-2.5 focus:outline-none !no-underline rounded-r text-gray-500 filter grayscale" href="?fw=tf"><svg class="mr-1.5" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="0.94em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 274"><path d="M145.726 42.065v42.07l72.861 42.07v-42.07l-72.86-42.07zM0 84.135v42.07l36.43 21.03V105.17L0 84.135zm109.291 21.035l-36.43 21.034v126.2l36.43 21.035v-84.135l36.435 21.035v-42.07l-36.435-21.034V105.17z" fill="#E55B2D"></path><path d="M145.726 42.065L36.43 105.17v42.065l72.861-42.065v42.065l36.435-21.03v-84.14zM255.022 63.1l-36.435 21.035v42.07l36.435-21.035V63.1zm-72.865 84.135l-36.43 21.035v42.07l36.43-21.036v-42.07zm-36.43 63.104l-36.436-21.035v84.135l36.435-21.035V210.34z" fill="#ED8E24"></path><path d="M145.726 0L0 84.135l36.43 21.035l109.296-63.105l72.861 42.07L255.022 63.1L145.726 0zm0 126.204l-36.435 21.03l36.435 21.036l36.43-21.035l-36.43-21.03z" fill="#F8BF3C"></path></svg> TensorFlow </a></div> <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 max-sm:gap-0.5 h-6 max-sm:h-5 px-2 max-sm:px-1.5 text-[11px] max-sm:text-[9px] 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"><svg class="w-3 h-3 max-sm:w-2.5 max-sm:h-2.5" 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-6 max-sm:h-5 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 w-3 h-3 max-sm:w-2.5 max-sm:h-2.5 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="handling-multiple-sequences" 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="#handling-multiple-sequences"><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>బహుళ సీక్వెన్స్‌ల నిర్వహణ</span></h1> <div class="flex space-x-1 absolute z-10 right-0 top-0" style=""><a href="https://discuss.huggingface.co/t/chapter-2-questions" target="_blank"><img alt="Ask a Question" class="!m-0" src="https://img.shields.io/badge/Ask%20a%20question-ffcb4c.svg?logo=data:image/svg+xml;base64,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"></a> <a href="https://colab.research.google.com/github/huggingface/notebooks/blob/master/course/en/chapter2/section5_pt.ipynb" target="_blank"><img alt="Open In Colab" class="!m-0" src="https://colab.research.google.com/assets/colab-badge.svg"></a> <a href="https://studiolab.sagemaker.aws/import/github/huggingface/notebooks/blob/master/course/en/chapter2/section5_pt.ipynb" target="_blank"><img alt="Open In Studio Lab" class="!m-0" src="https://studiolab.sagemaker.aws/studiolab.svg"></a></div> <iframe class="w-full xl:w-4/6 h-80" src="https://www.youtube-nocookie.com/embed/M6adb1j2jPI" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> <p data-svelte-h="svelte-linlqc">మునుపటి విభాగంలో, చిన్న పొడవు ఉన్న ఒకే సీక్వెన్స్‌పై inference చేయడం లాంటి అత్యంత సరళమైన ఉపయోగం చూశాం. కానీ అక్కడే కొన్ని ప్రశ్నలు వస్తాయి:</p> <ul data-svelte-h="svelte-1t3zal8"><li>బహుళ సీక్వెన్స్‌లను ఎలా హ్యాండిల్ చేస్తాం?</li> <li><em>భిన్నమైన పొడవు</em> ఉన్న బహుళ సీక్వెన్స్‌లను ఎలా హ్యాండిల్ చేస్తాం?</li> <li>vocabulary indices (input IDs) మాత్రమే మోడల్ బాగా పని చేయడానికి సరిపోతాయా?</li> <li>సీక్వెన్స్ చాలా ఎక్కువగా ఉండటం కూడా సమస్యగా మారుతుందా?</li></ul> <p data-svelte-h="svelte-1sadr0r">ఇప్పుడు ఈ ప్రశ్నలు ఏరకమైన సమస్యలను కలిగిస్తాయో, వాటిని 🤗 Transformers API తో ఎలా పరిష్కరించాలో చూద్దాం.</p> <h2 class="relative group"><a id="models-expect-a-batch-of-inputs" 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="#models-expect-a-batch-of-inputs"><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>మోడళ్లు batch రూపంలో ఇన్‌పుట్‌లను ఆశిస్తాయి</span></h2> <p data-svelte-h="svelte-1fwzm5n">మునుపటి exercise లో, సీక్వెన్స్‌లు సంఖ్యల జాబితాగా ఎలా మారతాయో చూశారు.<br>
ఇప్పుడు ఆ సంఖ్యల జాబితాను టెన్సర్‌గా మార్చి మోడల్‌కి పంపుదాం:</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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer, AutoModelForSequenceClassification
checkpoint = <span class="hljs-string">&quot;distilbert-base-uncased-finetuned-sst-2-english&quot;</span>
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForSequenceClassification.from_pretrained(checkpoint)
sequence = <span class="hljs-string">&quot;I&#x27;ve been waiting for a HuggingFace course my whole life.&quot;</span>
tokens = tokenizer.tokenize(sequence)
ids = tokenizer.convert_tokens_to_ids(tokens)
input_ids = torch.tensor(ids)
<span class="hljs-comment"># This line will fail.</span>
model(input_ids)<!-- HTML_TAG_END --></pre></div> <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=""><!-- HTML_TAG_START -->IndexError: Dimension out of <span class="hljs-built_in">range</span> (expected to be <span class="hljs-keyword">in</span> <span class="hljs-built_in">range</span> of [-<span class="hljs-number">1</span>, <span class="hljs-number">0</span>], but got <span class="hljs-number">1</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1ykopqi">అయ్యో! ఇది ఎందుకు fail అయింది? సెక్షన్ 2 లో pipeline ఉపయోగించిన స్టెప్పులనే ఇక్కడ కూడా పాటించాం కదా.</p> <p data-svelte-h="svelte-yhdcqa">సమస్య ఏమిటంటే, మనం మోడల్‌కి ఒక్క సీక్వెన్స్ మాత్రమే పంపాం, కానీ 🤗 Transformers మోడళ్లు default గా బహుళ వాక్యాలను (multiple sentences) ఆశిస్తాయి.<br>
ఇక్కడ మనం tokenizer <code>sequence</code> మీద వెనుకపట్లో చేసే పని మనమే చేతితో చేయడానికి ప్రయత్నించాం. కానీ బాగా గమనిస్తే, tokenizer input IDs లిస్ట్‌ను టెన్సర్‌గా మార్చడమే కాకుండా, దానిపై అదనంగా ఒక dimension కూడా వేసింది:</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=""><!-- HTML_TAG_START -->tokenized_inputs = tokenizer(sequence, return_tensors=<span class="hljs-string">&quot;pt&quot;</span>)
<span class="hljs-built_in">print</span>(tokenized_inputs[<span class="hljs-string">&quot;input_ids&quot;</span>])<!-- HTML_TAG_END --></pre></div> <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=""><!-- HTML_TAG_START -->tensor([[ <span class="hljs-number">101</span>, <span class="hljs-number">1045</span>, <span class="hljs-number">1005</span>, <span class="hljs-number">2310</span>, <span class="hljs-number">2042</span>, <span class="hljs-number">3403</span>, <span class="hljs-number">2005</span>, <span class="hljs-number">1037</span>, <span class="hljs-number">17662</span>, <span class="hljs-number">12172</span>,
 <span class="hljs-number">2607</span>, <span class="hljs-number">2026</span>, <span class="hljs-number">2878</span>, <span class="hljs-number">2166</span>, <span class="hljs-number">1012</span>, <span class="hljs-number">102</span>]])<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-15mtz0y">ఇప్పుడు మనమే ఆ కొత్త dimension ను జోడించి మళ్లీ ప్రయత్నిద్దాం:</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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer, AutoModelForSequenceClassification
checkpoint = <span class="hljs-string">&quot;distilbert-base-uncased-finetuned-sst-2-english&quot;</span>
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForSequenceClassification.from_pretrained(checkpoint)
sequence = <span class="hljs-string">&quot;I&#x27;ve been waiting for a HuggingFace course my whole life.&quot;</span>
tokens = tokenizer.tokenize(sequence)
ids = tokenizer.convert_tokens_to_ids(tokens)
input_ids = torch.tensor([ids])
<span class="hljs-built_in">print</span>(<span class="hljs-string">&quot;Input IDs:&quot;</span>, input_ids)
output = model(input_ids)
<span class="hljs-built_in">print</span>(<span class="hljs-string">&quot;Logits:&quot;</span>, output.logits)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-wix95o">ఇప్పుడు input IDs ని, అలాగే లభించిన logits ని ప్రింట్ చేస్తాం — అవి ఈ విధంగా ఉంటాయి:</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=""><!-- HTML_TAG_START -->Input IDs: [[ <span class="hljs-number">1045</span>, <span class="hljs-number">1005</span>, <span class="hljs-number">2310</span>, <span class="hljs-number">2042</span>, <span class="hljs-number">3403</span>, <span class="hljs-number">2005</span>, <span class="hljs-number">1037</span>, <span class="hljs-number">17662</span>, <span class="hljs-number">12172</span>, <span class="hljs-number">2607</span>, <span class="hljs-number">2026</span>, <span class="hljs-number">2878</span>, <span class="hljs-number">2166</span>, <span class="hljs-number">1012</span>]]
Logits: [[-<span class="hljs-number">2.7276</span>, <span class="hljs-number">2.8789</span>]]<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-5gmkto"><em>Batching</em> అంటే, ఒకేసారి బహుళ వాక్యాలను (multiple sentences) మోడల్‌కి పంపడం.<br>
మీ దగ్గర ఒక్క sentence మాత్రమే ఉన్నా, ఆ ఒక్క సీక్వెన్స్‌తోనే ఒక batch నిర్మించవచ్చు:</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=""><!-- HTML_TAG_START --><span class="hljs-attr">batched_ids</span> = [ids, ids]<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1xkts23">ఇది ఇప్పుడు రెండు ఒకే సీక్వెన్స్‌లను కలిగిన ఒక batch!</p> <blockquote class="tip"><p data-svelte-h="svelte-5uszdr">✏️ <strong>ప్రయత్నించండి!</strong> ఈ <code>batched_ids</code> లిస్ట్‌ను టెన్సర్‌గా మార్చి మీ మోడల్‌లోకి పంపండి.<br>
ముందు వచ్చిన logits మాదిరిగానే (కానీ రెండుసార్లు) వస్తున్నాయో చూడండి!</p></blockquote> <p data-svelte-h="svelte-9brv7k">Batching ద్వారా, మీరు బహుళ వాక్యాలను మోడల్‌కి పంపినప్పుడు అది సరిగ్గా పని చేస్తుంది.<br>
ఒక సీక్వెన్స్‌తో batch తయారుచేసినట్లే, బహుళ సీక్వెన్స్‌లతో కూడా batch తయారుచేయడం అంతే సులభం.</p> <p data-svelte-h="svelte-3urh72">కానీ ఇక్కడ రెండో సమస్య వస్తుంది: మీరు రెండు (లేదా అంతకంటే ఎక్కువ) వాక్యాలను batch లో కలపాలనుకుంటే, అవి ఒక్కోటి ఒక్కో పొడవు (length) లో ఉండవచ్చు.<br>
టెన్సర్‌లతో ముందే పని చేసి ఉంటే, అవి rectangular shape లో ఉండాల్సిందేనని మీకు తెలుసు; కాబట్టి input IDs జాబితాను నేరుగా టెన్సర్‌గా మార్చలేం.<br>
ఈ సమస్యను పరిష్కరించడానికి, సాధారణంగా మనం ఇన్‌పుట్‌లను <em>pad</em> చేస్తాం.</p> <h2 class="relative group"><a id="padding-the-inputs" 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="#padding-the-inputs"><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>ఇన్‌పుట్‌లను padding చేయడం</span></h2> <p data-svelte-h="svelte-cba0zh">క్రింద ఉన్న list of lists ను నేరుగా టెన్సర్‌గా మార్చలేం:</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=""><!-- HTML_TAG_START -->batched_ids = [
 [<span class="hljs-number">200</span>, <span class="hljs-number">200</span>, <span class="hljs-number">200</span>],
 [<span class="hljs-number">200</span>, <span class="hljs-number">200</span>]
]<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-15t8th7">దీనిని పరిష్కరించడానికి, మనం <em>padding</em> ఉపయోగిస్తాం, తద్వారా టెన్సర్‌లు rectangular shape లో ఉంటాయి.<br>
Padding ద్వారా, అన్ని వాక్యాలూ ఒకే పొడవు కలిగేలా, తక్కువ పదాలు ఉన్న వాక్యాలకు ప్రత్యేక పదమైన <em>padding token</em> ను చివరలో జోడిస్తాం.<br>
ఉదాహరణకు, 10 వాక్యాలు ఒక్కోటి 10 పదాలతో ఉండి, మరో ఒక వాక్యం 20 పదాలతో ఉందనుకోండి; padding చేసిన తర్వాత, అన్ని వాక్యాలూ 20 పదాలు కలిగి ఉంటాయి.<br>
మన ఉదాహరణలో, చివరికి వచ్చిన టెన్సర్ ఇలా కనిపిస్తుంది:</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=""><!-- HTML_TAG_START -->padding_id = <span class="hljs-number">100</span>
batched_ids = [
 [<span class="hljs-number">200</span>, <span class="hljs-number">200</span>, <span class="hljs-number">200</span>],
 [<span class="hljs-number">200</span>, <span class="hljs-number">200</span>, padding_id],
]<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1bqblqw">Padding token యొక్క ID ని <code>tokenizer.pad_token_id</code> లో చూడవచ్చు.<br>
ఇప్పుడిదాన్ని ఉపయోగించి, మన రెండు వాక్యాలను ఒక్కోసారి విడివిడిగా, అలాగే batch గా కలిపి మోడల్‌లోకి పంపుదాం:</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=""><!-- HTML_TAG_START -->model = AutoModelForSequenceClassification.from_pretrained(checkpoint)
sequence1_ids = [[<span class="hljs-number">200</span>, <span class="hljs-number">200</span>, <span class="hljs-number">200</span>]]
sequence2_ids = [[<span class="hljs-number">200</span>, <span class="hljs-number">200</span>]]
batched_ids = [
 [<span class="hljs-number">200</span>, <span class="hljs-number">200</span>, <span class="hljs-number">200</span>],
 [<span class="hljs-number">200</span>, <span class="hljs-number">200</span>, tokenizer.pad_token_id],
]
<span class="hljs-built_in">print</span>(model(torch.tensor(sequence1_ids)).logits)
<span class="hljs-built_in">print</span>(model(torch.tensor(sequence2_ids)).logits)
<span class="hljs-built_in">print</span>(model(torch.tensor(batched_ids)).logits)<!-- HTML_TAG_END --></pre></div> <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=""><!-- HTML_TAG_START -->tensor([[ <span class="hljs-number">1.5694</span>, -<span class="hljs-number">1.3895</span>]], grad_fn=&lt;AddmmBackward&gt;)
tensor([[ <span class="hljs-number">0.5803</span>, -<span class="hljs-number">0.4125</span>]], grad_fn=&lt;AddmmBackward&gt;)
tensor([[ <span class="hljs-number">1.5694</span>, -<span class="hljs-number">1.3895</span>],
 [ <span class="hljs-number">1.3373</span>, -<span class="hljs-number">1.2163</span>]], grad_fn=&lt;AddmmBackward&gt;)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-6cwulo">ఇక్కడ batched predictions లో logits లో ఏదో తేడా ఉంది: రెండో row, రెండో వాక్యాన్ని విడిగా పంపినప్పుడు వచ్చిన logits మాదిరిగానే ఉండాలి, కానీ పూర్తిగా వేరు values వచ్చాయి!</p> <p data-svelte-h="svelte-19rz2m9">దీనికి కారణం, Transformer మోడళ్లలో ప్రధాన పాత్ర పోషించే attention layers — ఇవి ప్రతి token ను దాని <em>context</em> తో కలిపి చూడడానికి ప్రయత్నిస్తాయి.<br>
ఇవి padding tokens ను కూడా గమనిస్తాయి, ఎందుకంటే అవి సీక్వెన్స్‌లోని అన్ని tokens పై దృష్టి పెడతాయి.<br>
కాబట్టి, వేర్వేరు పొడవు ఉన్న వాక్యాలను విడిగా మోడల్‌కి పంపినా, అదే వాక్యాలను padding తో కూడిన batch గా పంపినా, ఒకే ఫలితాలు రావాలంటే, attention layers‌కు padding tokens ను <em>పట్టించుకోవద్దు</em> అని చెప్పాలి. ఇది <em>attention mask</em> ద్వారా చేస్తాం.</p> <h2 class="relative group"><a id="attention-masks" 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="#attention-masks"><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>Attention masks</span></h2> <p data-svelte-h="svelte-1sjkqry"><em>Attention mask</em> అనేది input IDs టెన్సర్‌కి అచ్చం అదే రూపంలో ఉండే టెన్సర్, కానీ 0లూ 1లతో నిండినది:<br>
1 ఉన్న స్థానాల్లోని tokens పై మోడల్ దృష్టి పెట్టాలి, 0 ఉన్న స్థానాల్లోని tokens ను దృష్టి పెట్టకూడదు (అంటే, ఆ tokens ను attention layers పూర్తిగా <em>ignore</em> చేయాలి).</p> <p data-svelte-h="svelte-p6831a">ఇప్పుడు ముందున్న ఉదాహరణకే ఒక attention mask జోడిద్దాం:</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=""><!-- HTML_TAG_START -->batched_ids = [
 [<span class="hljs-number">200</span>, <span class="hljs-number">200</span>, <span class="hljs-number">200</span>],
 [<span class="hljs-number">200</span>, <span class="hljs-number">200</span>, tokenizer.pad_token_id],
]
attention_mask = [
 [<span class="hljs-number">1</span>, <span class="hljs-number">1</span>, <span class="hljs-number">1</span>],
 [<span class="hljs-number">1</span>, <span class="hljs-number">1</span>, <span class="hljs-number">0</span>],
]
outputs = model(torch.tensor(batched_ids), attention_mask=torch.tensor(attention_mask))
<span class="hljs-built_in">print</span>(outputs.logits)<!-- HTML_TAG_END --></pre></div> <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=""><!-- HTML_TAG_START -->tensor([[ <span class="hljs-number">1.5694</span>, -<span class="hljs-number">1.3895</span>],
 [ <span class="hljs-number">0.5803</span>, -<span class="hljs-number">0.4125</span>]], grad_fn=&lt;AddmmBackward&gt;)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1xe1iop">ఇప్పుడు batch లో రెండో వాక్యానికి వచ్చిన logits, ఆ వాక్యాన్ని విడిగా పంపినప్పుడు వచ్చిన logits‌ల మాదిరిగానే ఉన్నాయి.</p> <p data-svelte-h="svelte-w71kb8">గమనించండి: రెండో సీక్వెన్స్ లో చివరి విలువ padding ID కి సంబంధించినది; దానికి attention mask లో 0 ఉంది.</p> <blockquote class="tip"><p data-svelte-h="svelte-dah7o3">✏️ <strong>ప్రయత్నించండి!</strong> సెక్షన్ 2 లో ఉపయోగించిన రెండు వాక్యాల (<code>&quot;I&#39;ve been waiting for a HuggingFace course my whole life.&quot;</code> మరియు <code>&quot;I hate this so much!&quot;</code>) మీద టోకనైజేషన్‌ను చేతితో చేయండి.<br>
వాటిని మోడల్‌కి విడివిడిగా పంపి, సెక్షన్ 2 లో వచ్చిన logits మాదిరిగానే వస్తున్నాయో చూసుకోండి.<br>
తర్వాత వాటిని padding token తో కలిపి ఒక batch చేయండి, సరైన attention mask నిర్మించండి, మళ్లీ మోడల్‌లోకి పంపి ఫలితాలు ఒకటేనా చూడండి!</p></blockquote> <h2 class="relative group"><a id="longer-sequences" 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="#longer-sequences"><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>పొడవైన సీక్వెన్స్‌లు</span></h2> <p data-svelte-h="svelte-1imsni9">Transformer మోడళ్లతో పని చేసేప్పుడు, మోడల్ అంగీకరించే సీక్వెన్స్‌ల పొడవుకు ఒక పరిమితి ఉంటుంది.<br>
చాలా మోడళ్లు గరిష్టంగా 512 లేదా 1024 tokens వరకు మాత్రమే సీక్వెన్స్‌లను హ్యాండిల్ చేయగలవు; అంతకంటే పొడవైన సీక్వెన్స్‌లను పంపితే, అవి క్రాష్ కావచ్చు.<br>
ఈ సమస్యకు రెండు పరిష్కారాలు ఉన్నాయి:</p> <ul data-svelte-h="svelte-md17h8"><li>పొడవైన సీక్వెన్స్‌లను సపోర్ట్ చేసే మోడల్ ఉపయోగించడం.</li> <li>మీ సీక్వెన్స్‌లను truncation ద్వారా చిన్నవిగా కత్తిరించడం.</li></ul> <p data-svelte-h="svelte-13cjnrn">వివిధ మోడళ్లకు వివిధ గరిష్ట సీక్వెన్స్ పొడవులు ఉంటాయి; కొన్ని మోడళ్లు చాలా పొడవైన సీక్వెన్స్‌ల కోసం ప్రత్యేకంగా రూపొందించబడ్డాయి.<br>
ఉదాహరణకు, <a href="https://huggingface.co/docs/transformers/model_doc/longformer" rel="nofollow">Longformer</a>, <a href="https://huggingface.co/docs/transformers/model_doc/led" rel="nofollow">LED</a> లాంటి మోడళ్లు.<br>
మీ పని చాలా పొడవైన సీక్వెన్స్‌లను అవసరం చేసుకుంటే, ఇటువంటి మోడళ్లపై ఒకసారి పరిశీలించమని సూచిస్తాము.</p> <p data-svelte-h="svelte-bpdf5i">లేదంటే, మీ సీక్వెన్స్‌లను truncate చేయడం మంచిది; దానికి <code>max_sequence_length</code> పరామీటర్‌ను ఇవ్వవచ్చు:</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=""><!-- HTML_TAG_START -->sequence = sequence[:max_sequence_length]<!-- HTML_TAG_END --></pre></div> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/course/blob/main/chapters/te/chapter2/5.mdx" 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_1e95xo9 = {
 assets: "/docs/course/pr_1149/te",
 base: "/docs/course/pr_1149/te",
 env: {}
 };
 const element = document.currentScript.parentElement;
 const data = [null,null];
 Promise.all([
 import("/docs/course/pr_1149/te/_app/immutable/entry/start.eee4c1e0.js"),
 import("/docs/course/pr_1149/te/_app/immutable/entry/app.0f38558a.js")
 ]).then(([kit, app]) => {
 kit.start(app, element, {
 node_ids: [0, 18],
 data,
 form: null,
 error: null
 });
 });
 }
 </script>

Xet Storage Details

Size:
54.3 kB
·
Xet hash:
4b5010a6a45eceb0a305a79005fd9947e3084e4cadf334ea36771257423bd1b3

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