Buckets:

hf-doc-build
/

doc-dev

Files

xet

hf-doc-build/doc-dev / diffusers /pr_14867 /en /stable_diffusion.html

HuggingFaceDocBuilder

3 days ago

download

raw

17.7 kB

	<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"Overview","local":"overview","sections":[{"title":"Starter path","local":"starter-path","sections":[],"depth":2},{"title":"Optimization techniques","local":"optimization-techniques","sections":[],"depth":2}],"depth":1}"/>
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/entry/start.BY1X0FnC.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/chunks/BDChHepi.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/chunks/DK803DsY.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/entry/app.BcsduW5Q.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/chunks/DTwaC60R.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/chunks/BTASUwav.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/chunks/WKU8S240.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/nodes/0.ByoTqlNi.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/chunks/BzOvRKAw.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/nodes/286.fHAVRWvA.js" rel="modulepreload">
	<link href="/docs/diffusers/pr_14867/en/_app/immutable/chunks/Bqo1LwON.js" rel="modulepreload">
	<!--1ji58vo--><meta name="hf:doc:metadata" content="{"title":"Overview","local":"overview","sections":[{"title":"Starter path","local":"starter-path","sections":[],"depth":2},{"title":"Optimization techniques","local":"optimization-techniques","sections":[],"depth":2}],"depth":1}"/><!---->
	<link href="/docs/diffusers/pr_14867/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: 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> <!--[-1--><!--]--></div><!----> <!--[0--><h1 class="relative group"><a id="overview" 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="#overview"><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>Overview</span></h1><!--]--><!----> <p>Diffusion inference is computationally expensive, and you often run a <a href="/docs/diffusers/pr_14867/en/api/pipelines/overview#diffusers.DiffusionPipeline">DiffusionPipeline</a> more than once before you like the result. This page provides an overview of the main Diffusers optimization techniques, what they do, and when to use them.</p> <!--[1--><h2 class="relative group"><a id="starter-path" 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="#starter-path"><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>Starter path</span></h2><!--]--><!----> <p>When the model fits on one GPU, start with this baseline load. Set <code>dtype</code> and place the pipeline on an accelerator. Reach for model CPU offload only when memory is tight. You could also speed up inference with fewer steps or a faster scheduler.</p> <p>When you omit <code>dtype</code>, Diffusers loads components in <code>float32</code>. Pass <code>dtype=torch.bfloat16</code> (or <code>torch.float16</code> if bfloat16 is unsupported), then place the pipeline on an accelerator with <code>pipeline.to("cuda")</code>.</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copied</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-keyword">import</span> torch
	<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> DiffusionPipeline

	pipeline = DiffusionPipeline.from_pretrained(
	<span class="hljs-string">"stabilityai/stable-diffusion-xl-base-1.0"</span>,
	dtype=torch.bfloat16,
	)
	pipeline.to(<span class="hljs-string">"cuda"</span>) <span class="hljs-comment"># or "mps", "xpu"</span>

	prompt = <span class="hljs-string">"""
	cinematic film still of a cat sipping a margarita in a pool in Palm Springs, California
	highly detailed, high budget hollywood movie, cinemascope, moody, epic, gorgeous, film grain
	"""</span>
	pipeline(prompt).images[<span class="hljs-number">0</span>]<!----></pre></div><!----> <p>If the pipeline does not fit, or memory is tight, call <a href="/docs/diffusers/pr_14867/en/api/pipelines/overview#diffusers.DiffusionPipeline.enable_model_cpu_offload">enable_model_cpu_offload()</a> instead of keeping everything on the GPU. It places the active model on the GPU and keeps the other components on the CPU.</p> <p>Skip it when the model fits. Offloading is slower when you do not need it.</p> <p>For more offloading options, see <a href="./optimization/memory#offloading">Memory and offloading</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> Copied</div><!----></button><!----></div> <pre class="language-py "><!---->pipeline = DiffusionPipeline.from_pretrained(
	<span class="hljs-string">"stabilityai/stable-diffusion-xl-base-1.0"</span>,
	dtype=torch.bfloat16,
	)
	pipeline.enable_model_cpu_offload()<!----></pre></div><!----> <p>Lower latency with fewer <code>num_inference_steps</code> or a faster scheduler such as <a href="/docs/diffusers/pr_14867/en/api/schedulers/multistep_dpm_solver#diffusers.DPMSolverMultistepScheduler">DPMSolverMultistepScheduler</a>. That usually speeds up generation but can reduce image quality versus a slower, higher-quality scheduler. See <a href="./optimization/fp16">Precision and compilation</a> for more speed techniques.</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> Copied</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-keyword">import</span> time
	<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> DPMSolverMultistepScheduler

	pipeline.scheduler = DPMSolverMultistepScheduler.from_config(pipeline.scheduler.config)

	start_time = time.perf_counter()
	image = pipeline(prompt, num_inference_steps=<span class="hljs-number">25</span>).images[<span class="hljs-number">0</span>]
	end_time = time.perf_counter()

	<span class="hljs-built_in">print</span>(<span class="hljs-string">f"Image generation took <span class="hljs-subst">{end_time - start_time:<span class="hljs-number">.3</span>f}</span> seconds"</span>)<!----></pre></div><!----> <!--[1--><h2 class="relative group"><a id="optimization-techniques" 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="#optimization-techniques"><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>Optimization techniques</span></h2><!--]--><!----> <p>When the starter path is not enough, use these techniques. If you are out of memory, start with offloading or quantization. If inference is too slow, start with caching, attention backends, <code>torch.compile</code>, or regional compilation.</p> <ul><li><a href="./optimization/cache">Caching</a> — Reuse intermediates across denoising steps when you want more speed and can spend memory.</li> <li><a href="./optimization/attention_backends">Attention backends</a> — Swap Diffusers attention implementations through a unified API when attention is the bottleneck.</li> <li><a href="./quantization/overview">Quantization</a> — Load smaller weights to cut memory (and often speed up inference). <a href="./quantization/gguf">GGUF</a> is a common starting point.</li> <li><a href="./optimization/fp16#regional-compilation">Regional compilation</a> — Compile repeated blocks to cut <code>torch.compile</code> cold-start latency and reuse compiled artifacts.</li> <li><a href="./optimization/fp16#torchcompile">torch.compile</a> — Compile the UNet, transformer, or VAE into optimized kernels.</li> <li><a href="./optimization/fp16#kernels">Kernels</a> — Load optimized Hub compute kernels (attention and custom CUDA ops such as RMSNorm or RoPE) when you need hardware-specific speedups beyond stock PyTorch.</li> <li><a href="./optimization/memory#offloading">Offloading</a> — Move inactive models or layers to the CPU with CPU, model, or group offloading.</li> <li><a href="./optimization/speed-memory-optims">Quantize, compile, and offload</a> — Combine quantization, <code>torch.compile</code>, and offloading when one technique is not enough.</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/en/stable_diffusion.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_diewsf = {
	base: "/docs/diffusers/pr_14867/en",
	assets: "/docs/diffusers/pr_14867/en"
	};

	const element = document.currentScript.parentElement;

	Promise.all([
	import("/docs/diffusers/pr_14867/en/_app/immutable/entry/start.BY1X0FnC.js"),
	import("/docs/diffusers/pr_14867/en/_app/immutable/entry/app.BcsduW5Q.js")
	]).then(([kit, app]) => {
	kit.start(app, element, {
	node_ids: [0, 286],
	data: [null,null],
	form: null,
	error: null
	});
	});
	}
	</script>

Xet Storage Details

Size:: 17.7 kB
Xet hash:: 1172dfb1f087d0bacafe894c4bf15c15d7ab20c9e0a1d2bb5bbeac1c9bb0c9eb

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