Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"Безусловная генерация изображений","local":"unconditional-image-generation","sections":[{"title":"Параметры скрипта","local":"script-parameters","sections":[],"depth":2},{"title":"Скрипт обучения","local":"training-script","sections":[],"depth":2},{"title":"Запустите скрипт","local":"launch-the-script","sections":[],"depth":2}],"depth":1}"/> | |
| <link href="/docs/diffusers/main/en/_app/immutable/entry/start.0S0_iEoG.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/I9eooN77.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/Bupkjr9f.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/entry/app.CPdOxCbw.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/CyvZ3FS5.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/l0QnnOED.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/DnVIJfC5.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/nodes/0.DteTvmQZ.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/CoSrdQax.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/nodes/381.D9tpXcXv.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/DSPDl1gz.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/D5eijdnr.js" rel="modulepreload"> | |
| <!--eur7zo--><meta name="hf:doc:metadata" content="{"title":"Unconditional image generation","local":"unconditional-image-generation","sections":[{"title":"Script parameters","local":"script-parameters","sections":[],"depth":2},{"title":"Training script","local":"training-script","sections":[],"depth":2},{"title":"Launch the script","local":"launch-the-script","sections":[],"depth":2}],"depth":1}"/><!----> | |
| <link href="/docs/diffusers/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><!----> <!--[0--><h1 class="relative group"><a id="unconditional-image-generation" 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="#unconditional-image-generation"><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><!---->Безусловная генерация изображений<!----></span></h1><!--]--><!----> <p>Модели безусловной генерации изображений не используют в качестве условий текст или изображения во время обучения. Они генерируют только изображения, распределение которых сходно с распределением обучающих данных.</p> <p>В этом руководстве мы рассмотрим скрипт обучения <a href="https://github.com/huggingface/diffusers/blob/main/examples/unconditional_image_generation/train_unconditional.py" rel="nofollow">train_unconditional.py</a>, чтобы вы могли ознакомиться с ним и понять, как адаптировать его под свои задачи.</p> <p>Перед запуском скрипта убедитесь, что вы установили библиотеку из исходного кода:</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> Скопировано</div><!----></button><!----></div> <pre class="language-bash "><!---->git <span class="hljs-built_in">clone</span> https://github.com/huggingface/diffusers | |
| <span class="hljs-built_in">cd</span> diffusers | |
| pip install .<!----></pre></div><!----> <p>Затем перейдите в папку с примером, содержащую скрипт обучения, и установите необходимые зависимости:</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> Скопировано</div><!----></button><!----></div> <pre class="language-bash "><!----><span class="hljs-built_in">cd</span> examples/unconditional_image_generation | |
| pip install -r requirements.txt<!----></pre></div><!----> <blockquote class="tip"><p>🤗 Accelerate — это библиотека, которая помогает обучать модели на нескольких GPU/TPU или с использованием смешанной точности. Она автоматически настраивает среду обучения в зависимости от вашего оборудования и окружения. Подробнее можно узнать в <a href="https://huggingface.co/docs/accelerate/quicktour" rel="nofollow">Кратком обзоре</a> 🤗 Accelerate.</p></blockquote> <p>Инициализируйте среду 🤗 Accelerate:</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> Скопировано</div><!----></button><!----></div> <pre class="language-bash "><!---->accelerate config<!----></pre></div><!----> <p>Чтобы настроить среду по умолчанию 🤗 Accelerate без выбора каких-либо конфигураций:</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> Скопировано</div><!----></button><!----></div> <pre class="language-bash "><!---->accelerate config default<!----></pre></div><!----> <p>Или, если ваша среда не поддерживает интерактивную оболочку, например, ноутбук, вы можете использовать:</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> Скопировано</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-keyword">from</span> accelerate.utils <span class="hljs-keyword">import</span> write_basic_config | |
| write_basic_config()<!----></pre></div><!----> <p>Наконец, если вы хотите обучить модель на собственном наборе данных, ознакомьтесь с руководством <a href="create_dataset">Create a dataset for training</a>, чтобы узнать, как создать набор данных, совместимый со скриптом обучения.</p> <!--[1--><h2 class="relative group"><a id="script-parameters" 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="#script-parameters"><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><!---->Параметры скрипта<!----></span></h2><!--]--><!----> <blockquote class="tip"><p>В следующих разделах выделены части скрипта обучения, важные для понимания того, как его модифицировать, но не все аспекты скрипта описаны подробно. Если вы хотите узнать больше, не стесняйтесь прочитать <a href="https://github.com/huggingface/diffusers/blob/main/examples/unconditional_image_generation/train_unconditional.py" rel="nofollow">скрипт</a> и сообщить нам, если у вас возникнут вопросы или замечания.</p></blockquote> <p>Скрипт обучения предоставляет множество параметров, которые помогают настроить процесс обучения. Все параметры и их описания находятся в функции <a href="https://github.com/huggingface/diffusers/blob/096f84b05f9514fae9f185cbec0a4d38fbad9919/examples/unconditional_image_generation/train_unconditional.py#L55" rel="nofollow"><code>parse_args()</code></a>. Для каждого параметра заданы значения по умолчанию, например, размер обучающей выборки и скорость обучения, но при необходимости вы можете указать свои значения в команде запуска обучения.</p> <p>Например, чтобы ускорить обучение с использованием смешанной точности в формате bf16, добавьте параметр <code>--mixed_precision</code> к команде обучения:</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> Скопировано</div><!----></button><!----></div> <pre class="language-bash "><!---->accelerate launch train_unconditional.py \ | |
| --mixed_precision=<span class="hljs-string">"bf16"</span><!----></pre></div><!----> <p>К числу основных и важных параметров, которые необходимо указать, относятся:</p> <ul><li><code>--dataset_name</code>: название набора данных на Hub или локальный путь к набору данных для обучения</li> <li><code>--output_dir</code>: место для сохранения обученной модели</li> <li><code>--push_to_hub</code>: нужно ли отправить обученную модель в Hub</li> <li><code>--checkpointing_steps</code>: частота сохранения контрольных точек в процессе обучения модели; это полезно, если обучение прерывается, вы можете продолжить обучение с этой контрольной точки, добавив <code>--resume_from_checkpoint</code> к команде обучения</li></ul> <p>Принесите свой набор данных, а остальное скрипт обучения сделает сам!</p> <!--[1--><h2 class="relative group"><a id="training-script" 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="#training-script"><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><!---->Скрипт обучения<!----></span></h2><!--]--><!----> <p>Код для предобработки набора данных и цикла обучения находится в функции <a href="https://github.com/huggingface/diffusers/blob/096f84b05f9514fae9f185cbec0a4d38fbad9919/examples/unconditional_image_generation/train_unconditional.py#L275" rel="nofollow"><code>main()</code></a>. Если вам нужно адаптировать скрипт обучения, именно здесь следует внести изменения.</p> <p>Скрипт <code>train_unconditional</code> <a href="https://github.com/huggingface/diffusers/blob/096f84b05f9514fae9f185cbec0a4d38fbad9919/examples/unconditional_image_generation/train_unconditional.py#L356" rel="nofollow">инициализирует <code>UNet2DModel</code></a>, если вы не предоставили конфигурацию модели. При необходимости вы можете настроить UNet здесь:</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> Скопировано</div><!----></button><!----></div> <pre class="language-py "><!---->model = UNet2DModel( | |
| sample_size=args.resolution, | |
| in_channels=<span class="hljs-number">3</span>, | |
| out_channels=<span class="hljs-number">3</span>, | |
| layers_per_block=<span class="hljs-number">2</span>, | |
| block_out_channels=(<span class="hljs-number">128</span>, <span class="hljs-number">128</span>, <span class="hljs-number">256</span>, <span class="hljs-number">256</span>, <span class="hljs-number">512</span>, <span class="hljs-number">512</span>), | |
| down_block_types=( | |
| <span class="hljs-string">"DownBlock2D"</span>, | |
| <span class="hljs-string">"DownBlock2D"</span>, | |
| <span class="hljs-string">"DownBlock2D"</span>, | |
| <span class="hljs-string">"DownBlock2D"</span>, | |
| <span class="hljs-string">"AttnDownBlock2D"</span>, | |
| <span class="hljs-string">"DownBlock2D"</span>, | |
| ), | |
| up_block_types=( | |
| <span class="hljs-string">"UpBlock2D"</span>, | |
| <span class="hljs-string">"AttnUpBlock2D"</span>, | |
| <span class="hljs-string">"UpBlock2D"</span>, | |
| <span class="hljs-string">"UpBlock2D"</span>, | |
| <span class="hljs-string">"UpBlock2D"</span>, | |
| <span class="hljs-string">"UpBlock2D"</span>, | |
| ), | |
| )<!----></pre></div><!----> <p>Далее скрипт инициализирует <a href="https://github.com/huggingface/diffusers/blob/096f84b05f9514fae9f185cbec0a4d38fbad9919/examples/unconditional_image_generation/train_unconditional.py#L418" rel="nofollow">scheduler</a> и <a href="https://github.com/huggingface/diffusers/blob/096f84b05f9514fae9f185cbec0a4d38fbad9919/examples/unconditional_image_generation/train_unconditional.py#L429" rel="nofollow">optimizer</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> Скопировано</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-comment"># Initialize the scheduler</span> | |
| accepts_prediction_type = <span class="hljs-string">"prediction_type"</span> <span class="hljs-keyword">in</span> <span class="hljs-built_in">set</span>(inspect.signature(DDPMScheduler.__init__).parameters.keys()) | |
| <span class="hljs-keyword">if</span> accepts_prediction_type: | |
| noise_scheduler = DDPMScheduler( | |
| num_train_timesteps=args.ddpm_num_steps, | |
| beta_schedule=args.ddpm_beta_schedule, | |
| prediction_type=args.prediction_type, | |
| ) | |
| <span class="hljs-keyword">else</span>: | |
| noise_scheduler = DDPMScheduler(num_train_timesteps=args.ddpm_num_steps, beta_schedule=args.ddpm_beta_schedule) | |
| <span class="hljs-comment"># Initialize the optimizer</span> | |
| optimizer = torch.optim.AdamW( | |
| model.parameters(), | |
| lr=args.learning_rate, | |
| betas=(args.adam_beta1, args.adam_beta2), | |
| weight_decay=args.adam_weight_decay, | |
| eps=args.adam_epsilon, | |
| )<!----></pre></div><!----> <p>Затем он <a href="https://github.com/huggingface/diffusers/blob/096f84b05f9514fae9f185cbec0a4d38fbad9919/examples/unconditional_image_generation/train_unconditional.py#L451" rel="nofollow">загружает набор данных</a>, и вы можете указать, как его <a href="https://github.com/huggingface/diffusers/blob/096f84b05f9514fae9f185cbec0a4d38fbad9919/examples/unconditional_image_generation/train_unconditional.py#L455" rel="nofollow">предобработать</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> Скопировано</div><!----></button><!----></div> <pre class="language-py "><!---->dataset = load_dataset(<span class="hljs-string">"imagefolder"</span>, data_dir=args.train_data_dir, cache_dir=args.cache_dir, split=<span class="hljs-string">"train"</span>) | |
| augmentations = transforms.Compose( | |
| [ | |
| transforms.Resize(args.resolution, interpolation=transforms.InterpolationMode.BILINEAR), | |
| transforms.CenterCrop(args.resolution) <span class="hljs-keyword">if</span> args.center_crop <span class="hljs-keyword">else</span> transforms.RandomCrop(args.resolution), | |
| transforms.RandomHorizontalFlip() <span class="hljs-keyword">if</span> args.random_flip <span class="hljs-keyword">else</span> transforms.Lambda(<span class="hljs-keyword">lambda</span> x: x), | |
| transforms.ToTensor(), | |
| transforms.Normalize([<span class="hljs-number">0.5</span>], [<span class="hljs-number">0.5</span>]), | |
| ] | |
| )<!----></pre></div><!----> <p>Наконец, <a href="https://github.com/huggingface/diffusers/blob/096f84b05f9514fae9f185cbec0a4d38fbad9919/examples/unconditional_image_generation/train_unconditional.py#L540" rel="nofollow">training loop</a> выполняет все остальные задачи: добавление шума к изображениям, предсказание остатка шума, вычисление функции потерь, сохранение контрольных точек на указанных шагах, а также сохранение модели и её публикацию в Hub. Если вы хотите узнать больше о том, как работает цикл обучения, обратитесь к учебнику <a href="../using-diffusers/write_own_pipeline">Understanding pipelines, models and schedulers</a>, в котором подробно разбирается базовый паттерн процесса устранения шума.</p> <!--[1--><h2 class="relative group"><a id="launch-the-script" 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="#launch-the-script"><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><!---->Запустите скрипт<!----></span></h2><!--]--><!----> <p>Когда вы внесёте все изменения или решите оставить конфигурацию по умолчанию, можно запускать скрипт обучения! 🚀</p> <blockquote class="warning"><p>Полный цикл обучения занимает 2 часа на 4xV100 GPU.</p></blockquote> <div class="flex space-x-2 items-center my-1.5 mr-8 h-7 !pl-0 -mx-3 md:mx-0"><!--[--><div class="flex items-center border rounded-lg px-1.5 py-1 leading-none select-none text-smd border-gray-800 bg-black dark:bg-gray-700 text-white">один GPU</div><div class="flex items-center border rounded-lg px-1.5 py-1 leading-none select-none text-smd text-gray-500 cursor-pointer opacity-90 hover:text-gray-700 dark:hover:text-gray-200 hover:shadow-sm">multi-GPU</div><!--]--></div> <div class="language-select"><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 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> Скопировано</div><!----></button><!----></div> <pre class="language-bash "><!---->accelerate launch train_unconditional.py \ | |
| --dataset_name=<span class="hljs-string">"huggan/flowers-102-categories"</span> \ | |
| --output_dir=<span class="hljs-string">"ddpm-ema-flowers-64"</span> \ | |
| --mixed_precision=<span class="hljs-string">"fp16"</span> \ | |
| --push_to_hub<!----></pre></div><!----></div><!----> <div hidden=""><p>Если вы обучаете модель на нескольких GPU, добавьте параметр <code>--multi_gpu</code> в команду обучения:</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> Скопировано</div><!----></button><!----></div> <pre class="language-bash "><!---->accelerate launch --multi_gpu train_unconditional.py \ | |
| --dataset_name=<span class="hljs-string">"huggan/flowers-102-categories"</span> \ | |
| --output_dir=<span class="hljs-string">"ddpm-ema-flowers-64"</span> \ | |
| --mixed_precision=<span class="hljs-string">"fp16"</span> \ | |
| --push_to_hub<!----></pre></div><!----><!----></div><!----><!----></div><!----> <p>Скрипт обучения создаёт и сохраняет файл контрольной точки в вашем репозитории. Теперь вы можете загрузить и использовать обученную модель для инференса:</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> Скопировано</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> DiffusionPipeline | |
| <span class="hljs-keyword">import</span> torch | |
| pipeline = DiffusionPipeline.from_pretrained(<span class="hljs-string">"anton-l/ddpm-butterflies-128"</span>).to(<span class="hljs-string">"cuda"</span>) <span class="hljs-comment"># or "mps", "xpu", "cpu"</span> | |
| image = pipeline().images[<span class="hljs-number">0</span>]<!----></pre></div><!----> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/diffusers/blob/main/docs/source/en/training/unconditional_training.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_1qsdbwf = { | |
| base: "/docs/diffusers/main/en", | |
| assets: "/docs/diffusers/main/en" | |
| }; | |
| const element = document.currentScript.parentElement; | |
| Promise.all([ | |
| import("/docs/diffusers/main/en/_app/immutable/entry/start.0S0_iEoG.js"), | |
| import("/docs/diffusers/main/en/_app/immutable/entry/app.CPdOxCbw.js") | |
| ]).then(([kit, app]) => { | |
| kit.start(app, element, { | |
| node_ids: [0, 381], | |
| data: [null,null], | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 40.3 kB
- Xet hash:
- 47f8abef4ee33cc2b7bc9d1279cc89580ec11f7a24d39ff13b3c2b8a6d1e967e
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.