Hugging Face's logo

Buckets:
hf-doc-build
/
doc

Files
xet
hf-doc-build/doc / diffusers /v0.43.0 /pt /optimization /tpu.html
mishig's picture
mishig
download
raw
25.3 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;Modo Eager&quot;,&quot;local&quot;:&quot;eager-mode&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Modo compilado&quot;,&quot;local&quot;:&quot;compiled-mode&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Paralelismo de tensor&quot;,&quot;local&quot;:&quot;tensor-parallelism&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/entry/start.1gaXe03k.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/chunks/BObs6oQO.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/chunks/DPnyVmXs.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/entry/app.BwdYRbCn.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/chunks/DSGUT_OW.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/chunks/C045MWWx.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/chunks/6tuJ3nyi.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/nodes/0.CornNeQR.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/chunks/IaDvMxLQ.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/en/_app/immutable/nodes/310.BQ9DJ6as.js" rel="modulepreload">
<link href="/docs/diffusers/v0.43.0/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/v0.43.0/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>Tensor Processing Units (TPUs) são os aceleradores de machine learning da Google, disponíveis como Cloud TPUs (v6e, v7x).
O Diffusers roda neles por meio do <a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">TorchTPU</a>, o backend do PyTorch para TPUs,
com mudanças mínimas no código.</p> <p>Três modos de execução estão disponíveis:</p> <table><thead><tr><th>Modo</th><th>Constante</th><th>Como ativar</th><th>Observações</th></tr></thead><tbody><tr><td>Strict eager (padrão)</td><td><code>EagerMode.DEFER_NEVER</code></td><td><code>pipe.to("tpu")</code></td><td>Operações despachadas uma de cada vez, assíncronas</td></tr><tr><td>Eager otimizado</td><td><code>EagerMode.DEFER_AND_FUSE</code></td><td><code>TORCH_TPU_DEFER_AND_FUSE=1</code></td><td>Operações adiadas e fundidas em grafos maiores; mais rápido que o modo eager estrito, com tempo de compilação mais longo</td></tr><tr><td>Compilar</td><td>—</td><td><code>torch.compile(module, backend="tpu")</code></td><td>Compilação AOT com <code>TpuBackend</code></td></tr></tbody></table> <p>Siga o <a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">guia de instalação do TorchTPU</a>. Uma vez instalado, <code>import torch</code> o carrega automaticamente e registra o dispositivo <code>"tpu"</code>, de modo que <code>pipe.to("tpu")</code> é a única alteração necessária. Adicione <code>import torch_tpu</code> apenas se você desativou o carregamento automático do backend com <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><!---->Modo Eager<!----></span></h2><!--]--><!----> <p>FLUX.1-schnell não cabe em um único chip v6e de uma vez, então use <a href="/docs/diffusers/v0.43.0/pt/api/pipelines/overview#diffusers.DiffusionPipeline.enable_model_cpu_offload">enable_model_cpu_offload()</a> para mover cada modelo para o TPU apenas enquanto ele está em execução. Ele detecta automaticamente o dispositivo <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> Copiado</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>Se um modelo for grande demais para um único chip, ou se você tiver vários chips e quiser menor latência, distribua os modelos entre os chips. Consulte a seção <a href="#tensor-parallelism">Paralelismo de tensor</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><!---->Modo compilado<!----></span></h2><!--]--><!----> <p>TorchTPU registra <code>"tpu"</code> como um nome de backend para <code>torch.compile</code> (<code>TpuBackend</code> por baixo dos panos), de modo que os componentes são compilados como qualquer outro alvo de <code>torch.compile</code>. A primeira chamada (aquecimento) é lenta porque realiza a compilação; chamadas subsequentes com as mesmas formas reutilizam o gráfico compilado.</p> <blockquote class="important"><p>TorchTPU requer <strong>formas estáticas</strong>, então passe <code>dynamic=False</code>. Um novo <code>height</code> ou <code>width</code> compila novamente para essa forma, uma vez; formas já vistas são reutilizadas. Alterar <code>num_inference_steps</code> não recompila.</p></blockquote> <p>Quando o pipeline inteiro cabe em um único chip, mova-o para o TPU e compile o transformer completo. Stable Diffusion 3.5
Medium (~15GB em bf16) cabe em um único chip 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> Copiado</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>Se o pipeline não couber em um único chip:</p> <ul><li>Com vários chips, use <a href="#tensor-parallelism">tensor parallelism</a> para fazer o sharding. Tudo permanece no TPU,
e <code>pipe.transformer.compile(...)</code> funciona da mesma forma no transformer shardado.</li> <li>Com <a href="/docs/diffusers/v0.43.0/pt/api/pipelines/overview#diffusers.DiffusionPipeline.enable_model_cpu_offload">enable_model_cpu_offload()</a>, os hooks de offload não podem ser rastreados pelo <code>torch.compile</code>. Compile apenas os blocos repetidos do transformer, com <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><!---->Paralelismo de tensor<!----></span></h2><!--]--><!----> <p>Divida modelos grandes demais para um único chip entre vários. O codificador de texto (~48GB) e o transformer (~64GB) do FLUX.2-dev excedem, cada um, a capacidade de um único chip, portanto o exemplo abaixo divide ambos:</p> <ul><li>o transformer com <a href="/docs/diffusers/v0.43.0/pt/api/parallel#diffusers.TensorParallelConfig">TensorParallelConfig</a>, passado para o argumento <code>parallel_config</code> de <a href="/docs/diffusers/v0.43.0/pt/api/models/overview#diffusers.ModelMixin.from_pretrained">from_pretrained()</a>. Cada rank lê apenas a própria fatia de cada peso particionado, de modo que o modelo completo nunca é materializado. Para detalhes gerais de TP (<code>_tp_plan</code>, colwise/rowwise), consulte o guia <a href="./tensor_parallelism">Tensor parallelism</a>.</li> <li>o codificador de texto com o próprio <a href="https://huggingface.co/docs/transformers/perf_infer_gpu_multi" rel="nofollow">paralelismo de tensor</a> dos Transformers, passando um <a href="https://huggingface.co/docs/transformers/v5.22.2/en/expert_parallelism#transformers.DistributedConfig" rel="nofollow">DistributedConfig</a> e a mesma malha.</li></ul> <p>No TPU, inicialize o grupo de processos com <code>backend="tpu_dist"</code> e construa a malha com <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> Copiado</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>Inicie um processo por chip. Defina <code>--nproc_per_node</code> para usar o número total de chips TPU no seu host.</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> Copiado</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>O comando de inicialização acima usa um módulo interno do TorchTPU (<code>torch_tpu._internal</code>), não uma API estável, e é esperado que mude em uma futura liberação do TorchTPU. Consulte o <a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">repositório do TorchTPU</a> para a maneira atual de iniciar trabalhos multi-chip.</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_ywq9g6 = {
base: "/docs/diffusers/v0.43.0/en",
assets: "/docs/diffusers/v0.43.0/en"
};
const element = document.currentScript.parentElement;
Promise.all([
import("/docs/diffusers/v0.43.0/en/_app/immutable/entry/start.1gaXe03k.js"),
import("/docs/diffusers/v0.43.0/en/_app/immutable/entry/app.BwdYRbCn.js")
]).then(([kit, app]) => {
kit.start(app, element, {
node_ids: [0, 310],
data: [null,null],
form: null,
error: null
});
});
}
</script>

Xet Storage Details

Size:
25.3 kB
·
Xet hash:
2c4d5a10d113d415af47964f87e3d24730afd3db57e8745b4aa18b3e178a5620

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