Buckets:

hf-doc-build
/

doc

Files
xet
hf-doc-build/doc / diffusers /main /fr /optimization /tpu.html
mishig
1 day ago
download
raw
25.6 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Google TPU&quot;,&quot;local&quot;:&quot;google-tpu&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Mode Eager&quot;,&quot;local&quot;:&quot;eager-mode&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Mode compilé&quot;,&quot;local&quot;:&quot;compiled-mode&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Parallélisme de tenseurs&quot;,&quot;local&quot;:&quot;tensor-parallelism&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
		<link href="/docs/diffusers/main/en/_app/immutable/entry/start.D3LnKZ4R.js" rel="modulepreload">
		<link href="/docs/diffusers/main/en/_app/immutable/chunks/BF9UNEM9.js" rel="modulepreload">
		<link href="/docs/diffusers/main/en/_app/immutable/chunks/DPnyVmXs.js" rel="modulepreload">
		<link href="/docs/diffusers/main/en/_app/immutable/entry/app.C0rBoMxR.js" rel="modulepreload">
		<link href="/docs/diffusers/main/en/_app/immutable/chunks/DSGUT_OW.js" rel="modulepreload">
		<link href="/docs/diffusers/main/en/_app/immutable/chunks/C045MWWx.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.DTZsP_3B.js" rel="modulepreload">
		<link href="/docs/diffusers/main/en/_app/immutable/chunks/IaDvMxLQ.js" rel="modulepreload">
		<link href="/docs/diffusers/main/en/_app/immutable/nodes/311.C19KkVwx.js" rel="modulepreload">
		<link href="/docs/diffusers/main/en/_app/immutable/chunks/bKdwVYpU.js" rel="modulepreload">
		<!--vjbfci--><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Google TPU&quot;,&quot;local&quot;:&quot;google-tpu&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Eager mode&quot;,&quot;local&quot;:&quot;eager-mode&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Compiled mode&quot;,&quot;local&quot;:&quot;compiled-mode&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Tensor parallelism&quot;,&quot;local&quot;:&quot;tensor-parallelism&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;: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="google-tpu" 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="#google-tpu"><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><!---->Google TPU<!----></span></h1><!--]--><!----> <p>Les Tensor Processing Units (TPUs) sont les accélérateurs d’apprentissage automatique de Google, disponibles sous forme de Cloud TPUs (v6e, v7x).
Diffusers s’exécute dessus via <a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">TorchTPU</a>, le backend PyTorch pour les TPUs,
avec des modifications de code minimales.</p> <p>Trois modes d'exécution sont disponibles :</p> <table><thead><tr><th>Mode</th><th>Constant</th><th>Comment activer</th><th>Notes</th></tr></thead><tbody><tr><td>Strict eager (par défaut)</td><td><code>EagerMode.DEFER_NEVER</code></td><td><code>pipe.to("tpu")</code></td><td>Opérations expédiées une à la fois, asynchrones</td></tr><tr><td>Eager optimisé</td><td><code>EagerMode.DEFER_AND_FUSE</code></td><td><code>TORCH_TPU_DEFER_AND_FUSE=1</code></td><td>Opérations différées et fusionnées en graphes plus larges ; plus rapide que l'exécution stricte en mode eager, compilation plus longue</td></tr><tr><td>Compiler</td><td>—</td><td><code>torch.compile(module, backend="tpu")</code></td><td>Compilation AOT avec <code>TpuBackend</code></td></tr></tbody></table> <p>Suivez le <a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">guide d’installation de TorchTPU</a>. Une fois installé, <code>import torch</code> le charge automatiquement et enregistre le périphérique <code>"tpu"</code>, de sorte que <code>pipe.to("tpu")</code> est la seule modification nécessaire. Ajoutez <code>import torch_tpu</code> uniquement si vous avez désactivé le chargement automatique du backend avec <code>TORCH_DEVICE_BACKEND_AUTOLOAD=0</code>.</p> <!--[1--><h2 class="relative group"><a id="eager-mode" 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="#eager-mode"><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><!---->Mode Eager<!----></span></h2><!--]--><!----> <p>FLUX.1-schnell ne tient pas entièrement sur une seule puce v6e, utilisez donc <a href="/docs/diffusers/main/fr/api/pipelines/overview#diffusers.DiffusionPipeline.enable_model_cpu_offload">enable_model_cpu_offload()</a> pour déplacer chaque modèle sur le TPU uniquement pendant son exécution. Il détecte automatiquement le périphérique <code>"tpu"</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> Copié</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">import</span> torch

<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> FluxPipeline

pipe = FluxPipeline.from_pretrained(<span class="hljs-string">&quot;black-forest-labs/FLUX.1-schnell&quot;</span>, dtype=torch.bfloat16)
pipe.enable_model_cpu_offload()

image = pipe(
    prompt=<span class="hljs-string">&quot;a golden retriever surfing a wave, photorealistic&quot;</span>,
    height=<span class="hljs-number">1024</span>,
    width=<span class="hljs-number">1024</span>,
    num_inference_steps=<span class="hljs-number">4</span>,
    guidance_scale=<span class="hljs-number">0.0</span>,
).images[<span class="hljs-number">0</span>]

image.save(<span class="hljs-string">&quot;output.png&quot;</span>)<!----></pre></div><!----> <p>Si un modèle est trop volumineux pour une puce unique, ou si vous disposez de plusieurs puces et souhaitez réduire la latence, répartissez les modèles sur plusieurs puces. Consultez la section <a href="#tensor-parallelism">Parallélisme de tenseurs</a>.</p> <!--[1--><h2 class="relative group"><a id="compiled-mode" 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="#compiled-mode"><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><!---->Mode compilé<!----></span></h2><!--]--><!----> <p>TorchTPU enregistre <code>"tpu"</code> comme nom de backend pour <code>torch.compile</code> (<code>TpuBackend</code> en interne), de sorte que les composants sont compilés comme tout autre objectif <code>torch.compile</code>. Le premier appel (phase de préchauffage) est lent car il effectue la compilation ; les appels ultérieurs avec les mêmes formes réutilisent le graphe compilé.</p> <blockquote class="important"><p>TorchTPU exige des <strong>formes statiques</strong>, il faut donc passer <code>dynamic=False</code>. Une nouvelle <code>height</code> ou <code>width</code> déclenche une nouvelle compilation pour cette forme, une seule fois ; les formes déjà vues sont réutilisées. Modifier <code>num_inference_steps</code> ne provoque pas de re-compilation.</p></blockquote> <p>Lorsque le pipeline complet tient sur une seule puce, déplacez-le sur le TPU et compilez le transformer complet. Stable Diffusion 3.5
Medium (~15GB en bf16) tient sur une seule puce v6e.</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> Copié</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">import</span> torch

<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> StableDiffusion3Pipeline

pipe = StableDiffusion3Pipeline.from_pretrained(<span class="hljs-string">&quot;stabilityai/stable-diffusion-3.5-medium&quot;</span>, dtype=torch.bfloat16)
pipe.to(<span class="hljs-string">&quot;tpu&quot;</span>)
pipe.transformer.<span class="hljs-built_in">compile</span>(backend=<span class="hljs-string">&quot;tpu&quot;</span>, fullgraph=<span class="hljs-literal">True</span>, dynamic=<span class="hljs-literal">False</span>)

<span class="hljs-comment"># Warmup — triggers static graph compilation.</span>
pipe(prompt=<span class="hljs-string">&quot;warmup&quot;</span>, height=<span class="hljs-number">1024</span>, width=<span class="hljs-number">1024</span>, num_inference_steps=<span class="hljs-number">40</span>, guidance_scale=<span class="hljs-number">4.5</span>)

<span class="hljs-comment"># Later calls with the same shapes reuse the compiled graph.</span>
image = pipe(
    prompt=<span class="hljs-string">&quot;a golden retriever surfing a wave, photorealistic&quot;</span>,
    height=<span class="hljs-number">1024</span>,
    width=<span class="hljs-number">1024</span>,
    num_inference_steps=<span class="hljs-number">40</span>,
    guidance_scale=<span class="hljs-number">4.5</span>,
).images[<span class="hljs-number">0</span>]

image.save(<span class="hljs-string">&quot;output.png&quot;</span>)<!----></pre></div><!----> <p>Si le pipeline ne tient pas sur une seule puce :</p> <ul><li>Avec plusieurs puces, utilisez plutôt le <a href="#tensor-parallelism">parallélisme de tenseurs</a> pour le sharding. Tout reste sur le TPU, et <code>pipe.transformer.compile(...)</code> fonctionne de la même manière sur le transformer sharded.</li> <li>Avec <a href="/docs/diffusers/main/fr/api/pipelines/overview#diffusers.DiffusionPipeline.enable_model_cpu_offload">enable_model_cpu_offload()</a>, les hooks de déchargement ne peuvent pas être tracés par <code>torch.compile</code>. Compilez uniquement les blocs répétés du transformeur, avec <code>pipe.transformer.compile_repeated_blocks(backend="tpu", fullgraph=True, dynamic=False)</code>.</li></ul> <!--[1--><h2 class="relative group"><a id="tensor-parallelism" 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="#tensor-parallelism"><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><!---->Parallélisme de tenseurs<!----></span></h2><!--]--><!----> <p>Répartissez les modèles trop volumineux pour une seule puce sur plusieurs. L’encodeur de texte (~48 Go) et le transformeur (~64 Go) de FLUX.2-dev dépassent chacun la capacité d’une puce unique, l’exemple ci-dessous répartit donc les deux :</p> <ul><li>le transformer avec <a href="/docs/diffusers/main/fr/api/parallel#diffusers.TensorParallelConfig">TensorParallelConfig</a>, transmis à l’argument <code>parallel_config</code> de <a href="/docs/diffusers/main/fr/api/models/overview#diffusers.ModelMixin.from_pretrained">from_pretrained()</a>. Chaque rang ne lit que sa propre tranche de chaque poids fractionné, de sorte que le modèle complet n’est jamais matérialisé. Pour les détails généraux sur le TP (<code>_tp_plan</code>, colwise/rowwise), consultez le guide <a href="./tensor_parallelism">Tensor parallelism</a>.</li> <li>l’encodeur de texte avec le <a href="https://huggingface.co/docs/transformers/perf_infer_gpu_multi" rel="nofollow">parallélisme de tenseurs</a> propre à Transformers, en passant un <a href="https://huggingface.co/docs/transformers/main/en/expert_parallelism#transformers.DistributedConfig" rel="nofollow">DistributedConfig</a> et le même maillage.</li></ul> <p>Sur TPU, initialisez le groupe de processus avec <code>backend="tpu_dist"</code> et construisez le maillage avec <code>DeviceMesh("tpu", ...)</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> Copié</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">import</span> torch.distributed <span class="hljs-keyword">as</span> dist
<span class="hljs-keyword">from</span> torch.distributed.device_mesh <span class="hljs-keyword">import</span> DeviceMesh
<span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> DistributedConfig, Mistral3ForConditionalGeneration

<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> Flux2Pipeline, Flux2Transformer2DModel, TensorParallelConfig

dist.init_process_group(backend=<span class="hljs-string">&quot;tpu_dist&quot;</span>)
mesh = DeviceMesh(<span class="hljs-string">&quot;tpu&quot;</span>, <span class="hljs-built_in">list</span>(<span class="hljs-built_in">range</span>(dist.get_world_size())))

repo_id = <span class="hljs-string">&quot;black-forest-labs/FLUX.2-dev&quot;</span>
text_encoder = Mistral3ForConditionalGeneration.from_pretrained(
    repo_id,
    subfolder=<span class="hljs-string">&quot;text_encoder&quot;</span>,
    dtype=torch.bfloat16,
    distributed_config=DistributedConfig(tp_plan=<span class="hljs-string">&quot;auto&quot;</span>),
    device_mesh=mesh,
)
transformer = Flux2Transformer2DModel.from_pretrained(
    repo_id, subfolder=<span class="hljs-string">&quot;transformer&quot;</span>, dtype=torch.bfloat16, parallel_config=TensorParallelConfig(mesh=mesh)
)
pipe = Flux2Pipeline.from_pretrained(
    repo_id, text_encoder=text_encoder, transformer=transformer, dtype=torch.bfloat16
)
pipe.vae.to(<span class="hljs-string">&quot;tpu&quot;</span>)

image = pipe(
    prompt=<span class="hljs-string">&quot;a golden retriever surfing a wave, photorealistic&quot;</span>,
    num_inference_steps=<span class="hljs-number">28</span>,
    generator=torch.Generator(<span class="hljs-string">&quot;cpu&quot;</span>).manual_seed(<span class="hljs-number">0</span>),
).images[<span class="hljs-number">0</span>]
<span class="hljs-keyword">if</span> dist.get_rank() == <span class="hljs-number">0</span>:
    image.save(<span class="hljs-string">&quot;output.png&quot;</span>)<!----></pre></div><!----> <p>Lancez un processus par puce. Définissez <code>--nproc_per_node</code> pour utiliser le nombre total de puces TPU sur votre hôte.</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> Copié</div><!----></button><!----></div> <pre class="language-bash "><!----><span class="hljs-built_in">eval</span> $(python -m torch_tpu._internal.distributed.launchers.singlehost_wrapper | sed <span class="hljs-string">&#x27;s/^/export /&#x27;</span>)
torchrun --nproc_per_node=8 flux2_tp.py<!----></pre></div><!----> <blockquote class="warning"><p>La commande de lancement ci-dessus utilise un module interne de TorchTPU (<code>torch_tpu._internal</code>), et non une API stable, et est susceptible d'évoluer dans une future version de TorchTPU. Consultez le <a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">dépôt TorchTPU</a> pour connaître la méthode actuelle de lancement des tâches multi-puces.</p></blockquote> <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/optimization/tpu.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_g37pnb = {
						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.D3LnKZ4R.js"),
						import("/docs/diffusers/main/en/_app/immutable/entry/app.C0rBoMxR.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 311],
							data: [null,null],
							form: null,
							error: null
						});
					});
				}
			</script>
		
Xet Storage Details

Size:: 25.6 kB
Xet hash:: abab7a1d67c2a62551f49fa44774499e647287753c6cccaf4b51130f3fdf730d
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.