Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"TorchTPU","local":"torchtpu","sections":[{"title":"Eager mode","local":"eager-mode","sections":[],"depth":2},{"title":"Compiled mode","local":"compiled-mode","sections":[],"depth":2},{"title":"Tensor parallelism","local":"tensor-parallelism","sections":[],"depth":2}],"depth":1}"/> | |
| <link href="/docs/diffusers/main/en/_app/immutable/entry/start.IkFik8Wo.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/2-LFLkQ9.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/C6xp-fJV.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/entry/app.B97qzvN-.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/CV7jQnT_.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/chunks/D5zNozXB.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/nodes/0.B1FzR3OO.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/nodes/3.CGhDAFof.js" rel="modulepreload"> | |
| <link href="/docs/diffusers/main/en/_app/immutable/chunks/D1LY9R-V.js" rel="modulepreload"> | |
| <!--vjbfci--><meta name="hf:doc:metadata" content="{"title":"TorchTPU","local":"torchtpu","sections":[{"title":"Eager mode","local":"eager-mode","sections":[],"depth":2},{"title":"Compiled mode","local":"compiled-mode","sections":[],"depth":2},{"title":"Tensor parallelism","local":"tensor-parallelism","sections":[],"depth":2}],"depth":1}"/><!----> | |
| <link href="/docs/diffusers/main/en/_app/immutable/assets/0.tn0RQdqM.css" rel="modulepreload"> <!--[--><!--[0--><!--[--><!--[0--><!--[--><p></p> <div class="items-center shrink-0 min-w-[100px] max-sm:min-w-[50px] justify-end ml-auto flex" style="float: 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="torchtpu" 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="#torchtpu"><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>TorchTPU</span></h1><!--]--><!----> <p><a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">TorchTPU</a> is a PyTorch backend for Google’s Tensor Processing Units | |
| (TPUs), which lets you run Diffusers pipelines on Cloud TPUs (v6e, v7x) with minimal code changes.</p> <p>Two execution modes are available:</p> <table><thead><tr><th>Mode</th><th>Constant</th><th>How to activate</th><th>Notes</th></tr></thead><tbody><tr><td>Strict eager (default)</td><td><code>EagerMode.DEFER_NEVER</code></td><td><code>pipe.to("tpu")</code></td><td>Operations dispatched one at a time, asynchronous</td></tr><tr><td>Compile</td><td>—</td><td><code>torch.compile(module, backend="tpu")</code></td><td>AOT compilation with <code>TpuBackend</code></td></tr></tbody></table> <p>Follow the <a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">TorchTPU installation guide</a>. Once installed, <code>import torch</code> loads it automatically and registers the <code>"tpu"</code> device, so <code>pipe.to("tpu")</code> is the only change needed. Add <code>import torch_tpu</code> only if you disabled backend autoloading with <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>Eager mode</span></h2><!--]--><!----> <p>FLUX.1-schnell doesn’t fit on a single v6e chip all at once, so use <a href="/docs/diffusers/pr_14865/en/api/pipelines/overview#diffusers.DiffusionPipeline.enable_model_cpu_offload">enable_model_cpu_offload()</a> to | |
| move each model to the TPU only while it runs. It detects the <code>"tpu"</code> device automatically.</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-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">"black-forest-labs/FLUX.1-schnell"</span>, dtype=torch.bfloat16) | |
| pipe.enable_model_cpu_offload() | |
| image = pipe( | |
| prompt=<span class="hljs-string">"a golden retriever surfing a wave, photorealistic"</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">"output.png"</span>)<!----></pre></div><!----> <p>If a model is too large for a single chip, or you have several chips and want lower latency, shard the models | |
| across chips instead. See the <a href="#tensor-parallelism">Tensor parallelism</a> section.</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>Compiled mode</span></h2><!--]--><!----> <p>TorchTPU registers <code>"tpu"</code> as a <code>torch.compile</code> backend name (<code>TpuBackend</code> under the hood), so | |
| components compile like any other <code>torch.compile</code> target. The first | |
| call (warmup) is slow because it compiles; later calls with the same shapes reuse the compiled graph.</p> <blockquote class="important"><p>TorchTPU requires <strong>static shapes</strong>, so pass <code>dynamic=False</code>. A new <code>height</code> or <code>width</code> compiles again for that | |
| shape, once; shapes already seen are reused. Changing <code>num_inference_steps</code> doesn’t recompile.</p></blockquote> <p>When the whole pipeline fits on one chip, move it to the TPU and compile the full transformer. Stable Diffusion 3.5 | |
| Medium (~15GB in bf16) fits on a single v6e chip.</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-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">"stabilityai/stable-diffusion-3.5-medium"</span>, dtype=torch.bfloat16) | |
| pipe.to(<span class="hljs-string">"tpu"</span>) | |
| pipe.transformer.<span class="hljs-built_in">compile</span>(backend=<span class="hljs-string">"tpu"</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">"warmup"</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">"a golden retriever surfing a wave, photorealistic"</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">"output.png"</span>)<!----></pre></div><!----> <p>If the pipeline doesn’t fit on one chip:</p> <ul><li>With several chips, shard it with <a href="#tensor-parallelism">tensor parallelism</a> instead. Everything stays on the TPU, | |
| and <code>pipe.transformer.compile(...)</code> works the same way on the sharded transformer.</li> <li>With <a href="/docs/diffusers/pr_14865/en/api/pipelines/overview#diffusers.DiffusionPipeline.enable_model_cpu_offload">enable_model_cpu_offload()</a>, the offload hooks can’t be traced by <code>torch.compile</code>. Compile | |
| only the transformer’s repeated blocks instead, with <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>Tensor parallelism</span></h2><!--]--><!----> <p>Shard models too large for one chip across several. FLUX.2-dev’s text encoder (~48GB) and transformer (~64GB) each | |
| exceed a single chip, so the example below shards both:</p> <ul><li>the transformer with <a href="/docs/diffusers/pr_14865/en/api/parallel#diffusers.TensorParallelConfig">TensorParallelConfig</a>, passed to the <code>parallel_config</code> argument of <a href="/docs/diffusers/pr_14865/en/api/models/overview#diffusers.ModelMixin.from_pretrained">from_pretrained()</a>. Each rank reads only its own slice of every sharded weight, so the full model is never materialized. For general TP details (<code>_tp_plan</code>, colwise/rowwise), see the <a href="../training/distributed_inference#tensor-parallelism">Tensor parallelism</a> guide.</li> <li>the text encoder with Transformers’ own <a href="https://huggingface.co/docs/transformers/perf_infer_gpu_multi" rel="nofollow">tensor parallelism</a>, passing a <a href="https://huggingface.co/docs/transformers/main/en/expert_parallelism#transformers.DistributedConfig" rel="nofollow">DistributedConfig</a> and the same mesh.</li></ul> <p>On TPU, initialize the process group with <code>backend="tpu_dist"</code> and build the mesh with <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> Copied</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">"tpu_dist"</span>) | |
| mesh = DeviceMesh(<span class="hljs-string">"tpu"</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">"black-forest-labs/FLUX.2-dev"</span> | |
| text_encoder = Mistral3ForConditionalGeneration.from_pretrained( | |
| repo_id, | |
| subfolder=<span class="hljs-string">"text_encoder"</span>, | |
| dtype=torch.bfloat16, | |
| distributed_config=DistributedConfig(tp_plan=<span class="hljs-string">"auto"</span>), | |
| device_mesh=mesh, | |
| ) | |
| transformer = Flux2Transformer2DModel.from_pretrained( | |
| repo_id, subfolder=<span class="hljs-string">"transformer"</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">"tpu"</span>) | |
| image = pipe( | |
| prompt=<span class="hljs-string">"a golden retriever surfing a wave, photorealistic"</span>, | |
| num_inference_steps=<span class="hljs-number">28</span>, | |
| generator=torch.Generator(<span class="hljs-string">"cpu"</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">"output.png"</span>)<!----></pre></div><!----> <p>Launch one process per chip. Set <code>--nproc_per_node</code> to use all the number of TPU chips on your 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> Copied</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">'s/^/export /'</span>) | |
| torchrun --nproc_per_node=8 flux2_tp.py<!----></pre></div><!----> <blockquote class="warning"><p>The launch command above uses a TorchTPU internal module (<code>torch_tpu._internal</code>), not a stable API, and is expected | |
| to change in a future TorchTPU release. Check the <a href="https://github.com/google-pytorch/torch_tpu/" rel="nofollow">TorchTPU repository</a> for the current way to launch multi-chip jobs.</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_9x3bfb = { | |
| 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.IkFik8Wo.js"), | |
| import("/docs/diffusers/main/en/_app/immutable/entry/app.B97qzvN-.js") | |
| ]).then(([kit, app]) => { | |
| kit.start(app, element, { | |
| node_ids: [0, 3], | |
| data: [null,null], | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 24.4 kB
- Xet hash:
- 25ee1834953faeb1e0fa199f79a4059f421d9406faa2dc3631e3318b1ffe88e9
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.