Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"훑어보기","local":"훑어보기","sections":[{"title":"DiffusionPipeline","local":"diffusionpipeline","sections":[{"title":"로컬 파이프라인","local":"로컬-파이프라인","sections":[],"depth":3},{"title":"스케줄러 교체","local":"스케줄러-교체","sections":[],"depth":3}],"depth":2},{"title":"모델","local":"모델","sections":[],"depth":2},{"title":"스케줄러","local":"스케줄러","sections":[],"depth":2},{"title":"다음 단계","local":"다음-단계","sections":[],"depth":2}],"depth":1}"> | |
| <link href="/docs/diffusers/pr_12820/ko/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/entry/start.3d235f8e.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/scheduler.23542ac5.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/singletons.90d76c2a.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/index.7166da64.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/paths.5789b09f.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/entry/app.86e25fc8.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/preload-helper.b51cdadd.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/index.9b1f405b.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/nodes/0.49c759ef.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/each.e59479a4.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/nodes/19.1ee74317.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/MermaidChart.svelte_svelte_type_style_lang.af5bd6bd.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/CodeBlock.dc8eaa1c.js"> | |
| <link rel="modulepreload" href="/docs/diffusers/pr_12820/ko/_app/immutable/chunks/DocNotebookDropdown.68a629d2.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{"title":"훑어보기","local":"훑어보기","sections":[{"title":"DiffusionPipeline","local":"diffusionpipeline","sections":[{"title":"로컬 파이프라인","local":"로컬-파이프라인","sections":[],"depth":3},{"title":"스케줄러 교체","local":"스케줄러-교체","sections":[],"depth":3}],"depth":2},{"title":"모델","local":"모델","sections":[],"depth":2},{"title":"스케줄러","local":"스케줄러","sections":[],"depth":2},{"title":"다음 단계","local":"다음-단계","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> <div class="flex space-x-1 " style="float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"> <div class="relative colab-dropdown "> <button class=" " type="button"> <img alt="Open In Colab" class="!m-0" src="https://colab.research.google.com/assets/colab-badge.svg"> </button> </div> <div class="relative colab-dropdown "> <button class=" " type="button"> <img alt="Open In Studio Lab" class="!m-0" src="https://studiolab.sagemaker.aws/studiolab.svg"> </button> </div></div> <h1 class="relative group"><a id="훑어보기" 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="#훑어보기"><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>훑어보기</span></h1> <p data-svelte-h="svelte-12ydspm">Diffusion 모델은 이미지나 오디오와 같은 관심 샘플들을 생성하기 위해 랜덤 가우시안 노이즈를 단계별로 제거하도록 학습됩니다. 이로 인해 생성 AI에 대한 관심이 매우 높아졌으며, 인터넷에서 diffusion 생성 이미지의 예를 본 적이 있을 것입니다. 🧨 Diffusers는 누구나 diffusion 모델들을 널리 이용할 수 있도록 하기 위한 라이브러리입니다.</p> <p data-svelte-h="svelte-fidl65">개발자든 일반 사용자든 이 훑어보기를 통해 🧨 Diffusers를 소개하고 빠르게 생성할 수 있도록 도와드립니다! 알아야 할 라이브러리의 주요 구성 요소는 크게 세 가지입니다:</p> <ul data-svelte-h="svelte-1hg1a9y"><li><code>DiffusionPipeline</code>은 추론을 위해 사전 학습된 diffusion 모델에서 샘플을 빠르게 생성하도록 설계된 높은 수준의 엔드투엔드 클래스입니다.</li> <li>Diffusion 시스템 생성을 위한 빌딩 블록으로 사용할 수 있는 널리 사용되는 사전 학습된 <a href="./api/models">model</a> 아키텍처 및 모듈.</li> <li>다양한 <a href="./api/schedulers/overview">schedulers</a> - 학습을 위해 노이즈를 추가하는 방법과 추론 중에 노이즈 제거된 이미지를 생성하는 방법을 제어하는 알고리즘입니다.</li></ul> <p data-svelte-h="svelte-kc7anl">훑어보기에서는 추론을 위해 <code>DiffusionPipeline</code>을 사용하는 방법을 보여준 다음, 모델과 스케줄러를 결합하여 <code>DiffusionPipeline</code> 내부에서 일어나는 일을 복제하는 방법을 안내합니다.</p> <blockquote class="tip" data-svelte-h="svelte-1n890cv"><p>훑어보기는 간결한 버전의 🧨 Diffusers 소개로서 <a href="https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/diffusers_intro.ipynb" rel="nofollow">노트북</a> 빠르게 시작할 수 있도록 도와드립니다. 디퓨저의 목표, 디자인 철학, 핵심 API에 대한 추가 세부 정보를 자세히 알아보려면 노트북을 확인하세요!</p></blockquote> <p data-svelte-h="svelte-1k0z9pm">시작하기 전에 필요한 라이브러리가 모두 설치되어 있는지 확인하세요:</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-comment"># 주석 풀어서 Colab에 필요한 라이브러리 설치하기.</span> | |
| <span class="hljs-comment">#!pip install --upgrade diffusers accelerate transformers</span><!-- HTML_TAG_END --></pre></div> <ul data-svelte-h="svelte-8l98oh"><li><a href="https://huggingface.co/docs/accelerate/index" rel="nofollow">🤗 Accelerate</a>는 추론 및 학습을 위한 모델 로딩 속도를 높여줍니다.</li> <li><a href="https://huggingface.co/docs/transformers/index" rel="nofollow">🤗 Transformers</a>는 <a href="https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion/overview" rel="nofollow">Stable Diffusion</a>과 같이 가장 많이 사용되는 diffusion 모델을 실행하는 데 필요합니다.</li></ul> <h2 class="relative group"><a id="diffusionpipeline" 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="#diffusionpipeline"><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>DiffusionPipeline</span></h2> <p data-svelte-h="svelte-1szdpxj"><code>DiffusionPipeline</code> 은 추론을 위해 사전 학습된 diffusion 시스템을 사용하는 가장 쉬운 방법입니다. 모델과 스케줄러를 포함하는 엔드 투 엔드 시스템입니다. 다양한 작업에 <code>DiffusionPipeline</code>을 바로 사용할 수 있습니다. 아래 표에서 지원되는 몇 가지 작업을 살펴보고, 지원되는 작업의 전체 목록은 <a href="./api/pipelines/overview#diffusers-summary">🧨 Diffusers Summary</a> 표에서 확인할 수 있습니다.</p> <table data-svelte-h="svelte-1jwscsx"><thead><tr><th><strong>Task</strong></th> <th><strong>Description</strong></th> <th><strong>Pipeline</strong></th></tr></thead> <tbody><tr><td>Unconditional Image Generation</td> <td>generate an image from Gaussian noise</td> <td><a href="./using-diffusers/unconditional_image_generation">unconditional_image_generation</a></td></tr> <tr><td>Text-Guided Image Generation</td> <td>generate an image given a text prompt</td> <td><a href="./using-diffusers/conditional_image_generation">conditional_image_generation</a></td></tr> <tr><td>Text-Guided Image-to-Image Translation</td> <td>adapt an image guided by a text prompt</td> <td><a href="./using-diffusers/img2img">img2img</a></td></tr> <tr><td>Text-Guided Image-Inpainting</td> <td>fill the masked part of an image given the image, the mask and a text prompt</td> <td><a href="./using-diffusers/inpaint">inpaint</a></td></tr> <tr><td>Text-Guided Depth-to-Image Translation</td> <td>adapt parts of an image guided by a text prompt while preserving structure via depth estimation</td> <td><a href="./using-diffusers/depth2img">depth2img</a></td></tr></tbody></table> <p data-svelte-h="svelte-y8tkc8">먼저 <code>DiffusionPipeline</code>의 인스턴스를 생성하고 다운로드할 파이프라인 체크포인트를 지정합니다. | |
| 허깅페이스 허브에 저장된 모든 <a href="https://huggingface.co/models?library=diffusers&sort=downloads" rel="nofollow">checkpoint</a>에 대해 <code>DiffusionPipeline</code>을 사용할 수 있습니다. | |
| 이 훑어보기에서는 text-to-image 생성을 위한 <a href="https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5" rel="nofollow"><code>stable-diffusion-v1-5</code></a> 체크포인트를 로드합니다.</p> <blockquote data-svelte-h="svelte-1sqdffb"><p>[!WARNING][Stable Diffusion](<a href="https://huggingface.co/CompVis/stable-diffusion" rel="nofollow">https://huggingface.co/CompVis/stable-diffusion</a>) 모델의 경우, 모델을 실행하기 전에 <a href="https://huggingface.co/spaces/CompVis/stable-diffusion-license" rel="nofollow">라이선스</a>를 먼저 주의 깊게 읽어주세요. 🧨 Diffusers는 불쾌하거나 유해한 콘텐츠를 방지하기 위해 <a href="https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/safety_checker.py" rel="nofollow"><code>safety_checker</code></a>를 구현하고 있지만, 모델의 향상된 이미지 생성 기능으로 인해 여전히 잠재적으로 유해한 콘텐츠가 생성될 수 있습니다.</p></blockquote> <p data-svelte-h="svelte-28n0vb"><code>from_pretrained()</code> 방법으로 모델 로드하기:</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-meta">>>> </span><span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> DiffusionPipeline | |
| <span class="hljs-meta">>>> </span>pipeline = DiffusionPipeline.from_pretrained(<span class="hljs-string">"stable-diffusion-v1-5/stable-diffusion-v1-5"</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1puhto3">The <code>DiffusionPipeline</code>은 모든 모델링, 토큰화, 스케줄링 컴포넌트를 다운로드하고 캐시합니다. Stable Diffusion Pipeline은 무엇보다도 <code>UNet2DConditionModel</code>과 <code>PNDMScheduler</code>로 구성되어 있음을 알 수 있습니다:</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-meta">>>> </span>pipeline | |
| StableDiffusionPipeline { | |
| <span class="hljs-string">"_class_name"</span>: <span class="hljs-string">"StableDiffusionPipeline"</span>, | |
| <span class="hljs-string">"_diffusers_version"</span>: <span class="hljs-string">"0.13.1"</span>, | |
| ..., | |
| <span class="hljs-string">"scheduler"</span>: [ | |
| <span class="hljs-string">"diffusers"</span>, | |
| <span class="hljs-string">"PNDMScheduler"</span> | |
| ], | |
| ..., | |
| <span class="hljs-string">"unet"</span>: [ | |
| <span class="hljs-string">"diffusers"</span>, | |
| <span class="hljs-string">"UNet2DConditionModel"</span> | |
| ], | |
| <span class="hljs-string">"vae"</span>: [ | |
| <span class="hljs-string">"diffusers"</span>, | |
| <span class="hljs-string">"AutoencoderKL"</span> | |
| ] | |
| }<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-167uutc">이 모델은 약 14억 개의 파라미터로 구성되어 있으므로 GPU에서 파이프라인을 실행할 것을 강력히 권장합니다. | |
| PyTorch에서와 마찬가지로 제너레이터 객체를 GPU로 이동할 수 있습니다:</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-meta">>>> </span>pipeline.to(<span class="hljs-string">"cuda"</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-16e2uh9">이제 <code>파이프라인</code>에 텍스트 프롬프트를 전달하여 이미지를 생성한 다음 노이즈가 제거된 이미지에 액세스할 수 있습니다. 기본적으로 이미지 출력은 <a href="https://pillow.readthedocs.io/en/stable/reference/Image.html?highlight=image#the-image-class" rel="nofollow"><code>PIL.Image</code></a> 객체로 감싸집니다.</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-meta">>>> </span>image = pipeline(<span class="hljs-string">"An image of a squirrel in Picasso style"</span>).images[<span class="hljs-number">0</span>] | |
| <span class="hljs-meta">>>> </span>image<!-- HTML_TAG_END --></pre></div> <div class="flex justify-center" data-svelte-h="svelte-6cjtp1"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/image_of_squirrel_painting.png"></div> <p data-svelte-h="svelte-igu1m"><code>save</code>를 호출하여 이미지를 저장합니다:</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-meta">>>> </span>image.save(<span class="hljs-string">"image_of_squirrel_painting.png"</span>)<!-- HTML_TAG_END --></pre></div> <h3 class="relative group"><a id="로컬-파이프라인" 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="#로컬-파이프라인"><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>로컬 파이프라인</span></h3> <p data-svelte-h="svelte-7ma6tc">파이프라인을 로컬에서 사용할 수도 있습니다. 유일한 차이점은 가중치를 먼저 다운로드해야 한다는 점입니다:</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 -->!git lfs install | |
| !git <span class="hljs-built_in">clone</span> https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-vb7nw9">그런 다음 저장된 가중치를 파이프라인에 로드합니다:</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-meta">>>> </span>pipeline = DiffusionPipeline.from_pretrained(<span class="hljs-string">"./stable-diffusion-v1-5"</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1hsys9q">이제 위 섹션에서와 같이 파이프라인을 실행할 수 있습니다.</p> <h3 class="relative group"><a id="스케줄러-교체" 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="#스케줄러-교체"><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>스케줄러 교체</span></h3> <p data-svelte-h="svelte-lrbk5l">스케줄러마다 노이즈 제거 속도와 품질이 서로 다릅니다. 자신에게 가장 적합한 스케줄러를 찾는 가장 좋은 방법은 직접 사용해 보는 것입니다! 🧨 Diffusers의 주요 기능 중 하나는 스케줄러 간에 쉽게 전환이 가능하다는 것입니다. 예를 들어, 기본 스케줄러인 <code>PNDMScheduler</code>를 <code>EulerDiscreteScheduler</code>로 바꾸려면, <code>from_config()</code> 메서드를 사용하여 로드하세요:</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-meta">>>> </span><span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> EulerDiscreteScheduler | |
| <span class="hljs-meta">>>> </span>pipeline = DiffusionPipeline.from_pretrained(<span class="hljs-string">"stable-diffusion-v1-5/stable-diffusion-v1-5"</span>) | |
| <span class="hljs-meta">>>> </span>pipeline.scheduler = EulerDiscreteScheduler.from_config(pipeline.scheduler.config)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-6o48o0">새 스케줄러로 이미지를 생성해보고 어떤 차이가 있는지 확인해 보세요!</p> <p data-svelte-h="svelte-bvjymx">다음 섹션에서는 모델과 스케줄러라는 <code>DiffusionPipeline</code>을 구성하는 컴포넌트를 자세히 살펴보고 이러한 컴포넌트를 사용하여 고양이 이미지를 생성하는 방법을 배워보겠습니다.</p> <h2 class="relative group"><a id="모델" 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="#모델"><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>모델</span></h2> <p data-svelte-h="svelte-6xp0qa">대부분의 모델은 노이즈가 있는 샘플을 가져와 각 시간 간격마다 노이즈가 적은 이미지와 입력 이미지 사이의 차이인 <em>노이즈 잔차</em>(다른 모델은 이전 샘플을 직접 예측하거나 속도 또는 <a href="https://github.com/huggingface/diffusers/blob/5e5ce13e2f89ac45a0066cb3f369462a3cf1d9ef/src/diffusers/schedulers/scheduling_ddim.py#L110" rel="nofollow"><code>v-prediction</code></a>을 예측하는 학습을 합니다)을 예측합니다. 모델을 믹스 앤 매치하여 다른 diffusion 시스템을 만들 수 있습니다.</p> <p data-svelte-h="svelte-yb1tdz">모델은 <code>from_pretrained()</code> 메서드로 시작되며, 이 메서드는 모델 가중치를 로컬에 캐시하여 다음에 모델을 로드할 때 더 빠르게 로드할 수 있습니다. 훑어보기에서는 고양이 이미지에 대해 학습된 체크포인트가 있는 기본적인 unconditional 이미지 생성 모델인 <code>UNet2DModel</code>을 로드합니다:</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-meta">>>> </span><span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> UNet2DModel | |
| <span class="hljs-meta">>>> </span>repo_id = <span class="hljs-string">"google/ddpm-cat-256"</span> | |
| <span class="hljs-meta">>>> </span>model = UNet2DModel.from_pretrained(repo_id)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-wgpd7u">모델 매개변수에 액세스하려면 <code>model.config</code>를 호출합니다:</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-meta">>>> </span>model.config<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1al6c1n">모델 구성은 🧊 고정된 🧊 딕셔너리로, 모델이 생성된 후에는 해당 매개 변수들을 변경할 수 없습니다. 이는 의도적인 것으로, 처음에 모델 아키텍처를 정의하는 데 사용된 매개변수는 동일하게 유지하면서 다른 매개변수는 추론 중에 조정할 수 있도록 하기 위한 것입니다.</p> <p data-svelte-h="svelte-pvrthk">가장 중요한 매개변수들은 다음과 같습니다:</p> <ul data-svelte-h="svelte-36fna6"><li><code>sample_size</code>: 입력 샘플의 높이 및 너비 치수입니다.</li> <li><code>in_channels</code>: 입력 샘플의 입력 채널 수입니다.</li> <li><code>down_block_types</code> 및 <code>up_block_types</code>: UNet 아키텍처를 생성하는 데 사용되는 다운 및 업샘플링 블록의 유형.</li> <li><code>block_out_channels</code>: 다운샘플링 블록의 출력 채널 수. 업샘플링 블록의 입력 채널 수에 역순으로 사용되기도 합니다.</li> <li><code>layers_per_block</code>: 각 UNet 블록에 존재하는 ResNet 블록의 수입니다.</li></ul> <p data-svelte-h="svelte-qeu9h5">추론에 모델을 사용하려면 랜덤 가우시안 노이즈로 이미지 모양을 만듭니다. 모델이 여러 개의 무작위 노이즈를 수신할 수 있으므로 ‘batch’ 축, 입력 채널 수에 해당하는 ‘channel’ 축, 이미지의 높이와 너비를 나타내는 ‘sample_size’ 축이 있어야 합니다:</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-meta">>>> </span><span class="hljs-keyword">import</span> torch | |
| <span class="hljs-meta">>>> </span>torch.manual_seed(<span class="hljs-number">0</span>) | |
| <span class="hljs-meta">>>> </span>noisy_sample = torch.randn(<span class="hljs-number">1</span>, model.config.in_channels, model.config.sample_size, model.config.sample_size) | |
| <span class="hljs-meta">>>> </span>noisy_sample.shape | |
| torch.Size([<span class="hljs-number">1</span>, <span class="hljs-number">3</span>, <span class="hljs-number">256</span>, <span class="hljs-number">256</span>])<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-d3vota">추론을 위해 모델에 노이즈가 있는 이미지와 <code>timestep</code>을 전달합니다. ‘timestep’은 입력 이미지의 노이즈 정도를 나타내며, 시작 부분에 더 많은 노이즈가 있고 끝 부분에 더 적은 노이즈가 있습니다. 이를 통해 모델이 diffusion 과정에서 시작 또는 끝에 더 가까운 위치를 결정할 수 있습니다. <code>sample</code> 메서드를 사용하여 모델 출력을 얻습니다:</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-meta">>>> </span><span class="hljs-keyword">with</span> torch.no_grad(): | |
| <span class="hljs-meta">... </span> noisy_residual = model(sample=noisy_sample, timestep=<span class="hljs-number">2</span>).sample<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-so2glq">하지만 실제 예를 생성하려면 노이즈 제거 프로세스를 안내할 스케줄러가 필요합니다. 다음 섹션에서는 모델을 스케줄러와 결합하는 방법에 대해 알아봅니다.</p> <h2 class="relative group"><a id="스케줄러" 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="#스케줄러"><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>스케줄러</span></h2> <p data-svelte-h="svelte-tx8cwd">스케줄러는 모델 출력이 주어졌을 때 노이즈가 많은 샘플에서 노이즈가 적은 샘플로 전환하는 것을 관리합니다 - 이 경우 ‘noisy_residual’.</p> <blockquote class="tip" data-svelte-h="svelte-sqwal1"><p>🧨 Diffusers는 Diffusion 시스템을 구축하기 위한 툴박스입니다. <code>DiffusionPipeline</code>을 사용하면 미리 만들어진 Diffusion 시스템을 편리하게 시작할 수 있지만, 모델과 스케줄러 구성 요소를 개별적으로 선택하여 사용자 지정 Diffusion 시스템을 구축할 수도 있습니다.</p></blockquote> <p data-svelte-h="svelte-1127l8u">훑어보기의 경우, <code>from_config()</code> 메서드를 사용하여 <code>DDPMScheduler</code>를 인스턴스화합니다:</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-meta">>>> </span><span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> DDPMScheduler | |
| <span class="hljs-meta">>>> </span>scheduler = DDPMScheduler.from_config(repo_id) | |
| <span class="hljs-meta">>>> </span>scheduler | |
| DDPMScheduler { | |
| <span class="hljs-string">"_class_name"</span>: <span class="hljs-string">"DDPMScheduler"</span>, | |
| <span class="hljs-string">"_diffusers_version"</span>: <span class="hljs-string">"0.13.1"</span>, | |
| <span class="hljs-string">"beta_end"</span>: <span class="hljs-number">0.02</span>, | |
| <span class="hljs-string">"beta_schedule"</span>: <span class="hljs-string">"linear"</span>, | |
| <span class="hljs-string">"beta_start"</span>: <span class="hljs-number">0.0001</span>, | |
| <span class="hljs-string">"clip_sample"</span>: true, | |
| <span class="hljs-string">"clip_sample_range"</span>: <span class="hljs-number">1.0</span>, | |
| <span class="hljs-string">"num_train_timesteps"</span>: <span class="hljs-number">1000</span>, | |
| <span class="hljs-string">"prediction_type"</span>: <span class="hljs-string">"epsilon"</span>, | |
| <span class="hljs-string">"trained_betas"</span>: null, | |
| <span class="hljs-string">"variance_type"</span>: <span class="hljs-string">"fixed_small"</span> | |
| }<!-- HTML_TAG_END --></pre></div> <blockquote class="tip" data-svelte-h="svelte-1s6sj7o"><p>💡 스케줄러가 구성에서 어떻게 인스턴스화되는지 주목하세요. 모델과 달리 스케줄러에는 학습 가능한 가중치가 없으며 매개변수도 없습니다!</p></blockquote> <p data-svelte-h="svelte-14g28lg">가장 중요한 매개변수는 다음과 같습니다:</p> <ul data-svelte-h="svelte-j8ll1k"><li><code>num_train_timesteps</code>: 노이즈 제거 프로세스의 길이, 즉 랜덤 가우스 노이즈를 데이터 샘플로 처리하는 데 필요한 타임스텝 수입니다.</li> <li><code>beta_schedule</code>: 추론 및 학습에 사용할 노이즈 스케줄 유형입니다.</li> <li><code>beta_start</code> 및 <code>beta_end</code>: 노이즈 스케줄의 시작 및 종료 노이즈 값입니다.</li></ul> <p data-svelte-h="svelte-1q2pjlh">노이즈가 약간 적은 이미지를 예측하려면 스케줄러의 <code>step()</code> 메서드에 모델 출력, <code>timestep</code>, 현재 <code>sample</code>을 전달하세요.</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-meta">>>> </span>less_noisy_sample = scheduler.step(model_output=noisy_residual, timestep=<span class="hljs-number">2</span>, sample=noisy_sample).prev_sample | |
| <span class="hljs-meta">>>> </span>less_noisy_sample.shape<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-yfz7no"><code>less_noisy_sample</code>을 다음 <code>timestep</code>으로 넘기면 노이즈가 더 줄어듭니다! 이제 이 모든 것을 한데 모아 전체 노이즈 제거 과정을 시각화해 보겠습니다.</p> <p data-svelte-h="svelte-gihb52">먼저 노이즈 제거된 이미지를 후처리하여 <code>PIL.Image</code>로 표시하는 함수를 만듭니다:</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-meta">>>> </span><span class="hljs-keyword">import</span> PIL.Image | |
| <span class="hljs-meta">>>> </span><span class="hljs-keyword">import</span> numpy <span class="hljs-keyword">as</span> np | |
| <span class="hljs-meta">>>> </span><span class="hljs-keyword">def</span> <span class="hljs-title function_">display_sample</span>(<span class="hljs-params">sample, i</span>): | |
| <span class="hljs-meta">... </span> image_processed = sample.cpu().permute(<span class="hljs-number">0</span>, <span class="hljs-number">2</span>, <span class="hljs-number">3</span>, <span class="hljs-number">1</span>) | |
| <span class="hljs-meta">... </span> image_processed = (image_processed + <span class="hljs-number">1.0</span>) * <span class="hljs-number">127.5</span> | |
| <span class="hljs-meta">... </span> image_processed = image_processed.numpy().astype(np.uint8) | |
| <span class="hljs-meta">... </span> image_pil = PIL.Image.fromarray(image_processed[<span class="hljs-number">0</span>]) | |
| <span class="hljs-meta">... </span> display(<span class="hljs-string">f"Image at step <span class="hljs-subst">{i}</span>"</span>) | |
| <span class="hljs-meta">... </span> display(image_pil)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-m7443v">노이즈 제거 프로세스의 속도를 높이려면 입력과 모델을 GPU로 옮기세요:</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-meta">>>> </span>model.to(<span class="hljs-string">"cuda"</span>) | |
| <span class="hljs-meta">>>> </span>noisy_sample = noisy_sample.to(<span class="hljs-string">"cuda"</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-s236ae">이제 노이즈가 적은 샘플의 잔차를 예측하고 스케줄러로 노이즈가 적은 샘플을 계산하는 노이즈 제거 루프를 생성합니다:</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-meta">>>> </span><span class="hljs-keyword">import</span> tqdm | |
| <span class="hljs-meta">>>> </span>sample = noisy_sample | |
| <span class="hljs-meta">>>> </span><span class="hljs-keyword">for</span> i, t <span class="hljs-keyword">in</span> <span class="hljs-built_in">enumerate</span>(tqdm.tqdm(scheduler.timesteps)): | |
| <span class="hljs-meta">... </span> <span class="hljs-comment"># 1. predict noise residual</span> | |
| <span class="hljs-meta">... </span> <span class="hljs-keyword">with</span> torch.no_grad(): | |
| <span class="hljs-meta">... </span> residual = model(sample, t).sample | |
| <span class="hljs-meta">... </span> <span class="hljs-comment"># 2. compute less noisy image and set x_t -> x_t-1</span> | |
| <span class="hljs-meta">... </span> sample = scheduler.step(residual, t, sample).prev_sample | |
| <span class="hljs-meta">... </span> <span class="hljs-comment"># 3. optionally look at image</span> | |
| <span class="hljs-meta">... </span> <span class="hljs-keyword">if</span> (i + <span class="hljs-number">1</span>) % <span class="hljs-number">50</span> == <span class="hljs-number">0</span>: | |
| <span class="hljs-meta">... </span> display_sample(sample, i + <span class="hljs-number">1</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1h1yyvi">가만히 앉아서 고양이가 소음으로만 생성되는 것을 지켜보세요!😻</p> <div class="flex justify-center" data-svelte-h="svelte-1yzsrw6"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/diffusion-quicktour.png"></div> <h2 class="relative group"><a id="다음-단계" 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="#다음-단계"><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>다음 단계</span></h2> <p data-svelte-h="svelte-u5qyb5">이번 훑어보기에서 🧨 Diffusers로 멋진 이미지를 만들어 보셨기를 바랍니다! 다음 단계로 넘어가세요:</p> <ul data-svelte-h="svelte-1j7psol"><li><a href="./tutorials/basic_training">training</a> 튜토리얼에서 모델을 학습하거나 파인튜닝하여 나만의 이미지를 생성할 수 있습니다.</li> <li>다양한 사용 사례는 공식 및 커뮤니티 <a href="https://github.com/huggingface/diffusers/tree/main/examples#-diffusers-examples" rel="nofollow">학습 또는 파인튜닝 스크립트</a> 예시를 참조하세요.</li> <li>스케줄러 로드, 액세스, 변경 및 비교에 대한 자세한 내용은 <a href="./using-diffusers/schedulers">다른 스케줄러 사용</a> 가이드에서 확인하세요.</li> <li><a href="./stable_diffusion">Stable Diffusion</a> 가이드에서 프롬프트 엔지니어링, 속도 및 메모리 최적화, 고품질 이미지 생성을 위한 팁과 요령을 살펴보세요.</li> <li><a href="./optimization/fp16">GPU에서 파이토치 최적화</a> 가이드와 <a href="./optimization/mps">애플 실리콘(M1/M2)에서의 Stable Diffusion</a> 및 <a href="./optimization/onnx">ONNX 런타임</a> 실행에 대한 추론 가이드를 통해 🧨 Diffuser 속도를 높이는 방법을 더 자세히 알아보세요.</li></ul> <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/ko/quicktour.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_1nna9bv = { | |
| assets: "/docs/diffusers/pr_12820/ko", | |
| base: "/docs/diffusers/pr_12820/ko", | |
| env: {} | |
| }; | |
| const element = document.currentScript.parentElement; | |
| const data = [null,null]; | |
| Promise.all([ | |
| import("/docs/diffusers/pr_12820/ko/_app/immutable/entry/start.3d235f8e.js"), | |
| import("/docs/diffusers/pr_12820/ko/_app/immutable/entry/app.86e25fc8.js") | |
| ]).then(([kit, app]) => { | |
| kit.start(app, element, { | |
| node_ids: [0, 19], | |
| data, | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 62 kB
- Xet hash:
- c292b5483c8d7ebaa7fb7d51fcd5b7595d27a136bf20918a1937e7b608474e4c
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.