1<!DOCTYPE html> 2<html> 3<head> 4 <meta charset="utf-8"> 5 <title> 6 ð DistilGPT-2 model checkpoint 7 </title> 8 <meta name="viewport" content="width=device-width, initial-scale=1.0, user-scalable=no"> 9 <link rel="stylesheet" href="/front/build/style.css?v4"> 10 <link href="https://fonts.googleapis.com/css?family=Source+Sans+Pro:400,600,700" rel="stylesheet"> 11 <meta property="og:title" content="ð DistilGPT-2 model checkpoint"> 12 <meta property="og:type" content="website"> 13 <meta property="og:url" content="https://transformer.huggingface.co/model/distil-gpt2"> 14 <meta property="og:image" content="https://transformer.huggingface.co/front/assets/thumbnail-large-distilgpt2.png"> 15 <meta property="fb:app_id" content="1321688464574422"> 16 <meta name="twitter:card" content="summary_large_image"> 17 <meta name="twitter:site" content="@huggingface"> 18 <meta name="description" content="See how a modern neural network auto-completes your text"> 19</head> 20<body class="model" data-editable=""> 21 22 23<div class="container"> 24 <a class="back" href="/"> 25 <img class="info" src="/front/assets/icon-back.svg"> 26 See all models and checkpoints 27 </a> 28 <div class="model-title"> 29 ð DistilGPT-2 model checkpoint 30 </div> 31 <div class="github-repo"> 32 <a 33 class="github-button" 34 href="https://github.com/huggingface/transformers" data-size="large" data-show-count="true" aria-label="Star huggingface/transformers on GitHub"> 35 Star 36 </a> 37 </div> 38 <div class="model-details"> 39 <p>The student of the now ubiquitous GPT-2 does not come short of its teacherâs expectations. 40 Obtained by distillation, DistilGPT-2 weighs 37% less, and is twice as fast as its OpenAI counterpart, while keeping the same generative power. 41 Runs smoothly on an iPhone 7. The dawn of lightweight generative transformers? ð¤¯</p> 42 43 <p>From the paper: <a target="_blank" href="https://arxiv.org/abs/1910.01108">DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter</a> by Victor Sanh, Lysandre Debut, Julien Chaumond and Thomas Wolf. 44 The same method was applied to distill GPT-2, and a Medium blogpost <a target="_blank" href="https://medium.com/huggingface/distilbert-8cf3380435b5">describes the process in detail</a>.</p> 45 46 47 48 </div> 49 <div class="model-bottom"> 50 <a class="btn btn-primary" href="/doc/distil-gpt2">Start writing</a> 51 </div> 52</div> 53 54 55
55<script src="/front/assets/github-buttons.js"></script>
55 56
56<script src="/front/node_modules/quill/dist/quill.js"></script>
56 57
57<script src="/front/build/b1775038951944.min.js"></script>
57 58
58<script> 59 (function() { 60 if (window.location.hostname === 'localhost') { 61 console.log('[livereload]', 'on'); 62 const s = document.createElement('script'); 63 s.setAttribute('src', '//localhost:35729/livereload.js'); 64 document.body.appendChild(s); 65 } 66 })(); 67 </script>
67 68 <!-- Google analytics (legacy) --> 69
69<script> 70 if (! ['localhost', 'huggingface.test'].includes(window.location.hostname)) {
vendor: 344 bytes, lines 70-75
70 71 (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){ 72 (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o), 73 m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m) 74 })(window,document,'script','https://www.google-analytics.com/analytics.js','ga'); 75 ga('create', '
75UA-83738774-5
vendor: 39 bytes, lines 75-77
75', 'auto'); 76 ga('send', 'pageview'); 77
77} 78 </script>
78 79 <!-- Google tag (gtag.js) -->
vendor: 65 bytes, lines 79-80
79 80 <script async src="https://www.googletagmanager.com/gtag/js?id=
80G-56Y0W3EPT3
vendor: 13 bytes, lines 80-81
80"></script> 81
81<script> 82 if (! ['localhost', 'huggingface.test'].includes(window.location.hostname)) { 83
vendor: 136 bytes, lines 83-86
83window.dataLayer = window.dataLayer || []; 84 function gtag(){dataLayer.push(arguments);} 85 gtag('js', new Date()); 86 gtag('config', '
86G-56Y0W3EPT3
vendor: 6 bytes, lines 86-87
86'); 87
87} 88 </script>
88 89
89<script type="importmap"> 90 { "imports": { "space-header": "https://esm.sh/@huggingface/space-header" 91 } } 92 </script>
92 93
93<script type="module"> 94 import {init} from 'space-header'; 95 init("banana-projects/transformer-autocomplete") 96 </script>
96 97</body> 98</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.