Buckets:

download
raw
23.1 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Compilation&quot;,&quot;local&quot;:&quot;compilation&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Overview&quot;,&quot;local&quot;:&quot;overview&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Using torch.compile with Accelerate&quot;,&quot;local&quot;:&quot;using-torchcompile-with-accelerate&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Regional Compilation&quot;,&quot;local&quot;:&quot;regional-compilation&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;How to Use Regional Compilation&quot;,&quot;local&quot;:&quot;how-to-use-regional-compilation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Benefits of Regional Compilation&quot;,&quot;local&quot;:&quot;benefits-of-regional-compilation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Conclusion&quot;,&quot;local&quot;:&quot;conclusion&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
<link href="/docs/accelerate/pr_4021/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/entry/start.8a49e72b.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/scheduler.b9285784.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/singletons.7547c222.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/index.6d423e5c.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/paths.d42c9205.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/entry/app.1df4d18e.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/preload-helper.b0bd19d1.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/index.26bc89a1.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/nodes/0.0e7c56e8.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/each.e59479a4.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/nodes/40.1fba05b5.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/MermaidChart.svelte_svelte_type_style_lang.7a0ae628.js">
<link rel="modulepreload" href="/docs/accelerate/pr_4021/en/_app/immutable/chunks/CodeBlock.844ff9c3.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Compilation&quot;,&quot;local&quot;:&quot;compilation&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Overview&quot;,&quot;local&quot;:&quot;overview&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Using torch.compile with Accelerate&quot;,&quot;local&quot;:&quot;using-torchcompile-with-accelerate&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Regional Compilation&quot;,&quot;local&quot;:&quot;regional-compilation&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;How to Use Regional Compilation&quot;,&quot;local&quot;:&quot;how-to-use-regional-compilation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Benefits of Regional Compilation&quot;,&quot;local&quot;:&quot;benefits-of-regional-compilation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Conclusion&quot;,&quot;local&quot;:&quot;conclusion&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <div class="items-center shrink-0 min-w-[100px] max-sm:min-w-[50px] justify-end ml-auto flex" style="float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"><div class="inline-flex rounded-md max-sm:rounded-sm"><button class="inline-flex items-center gap-1 h-7 max-sm:h-7 px-2 max-sm:px-1.5 text-sm font-medium text-gray-800 border border-r-0 rounded-l-md max-sm:rounded-l-sm border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-live="polite"><span class="inline-flex items-center justify-center rounded-md p-0.5 max-sm:p-0 hover:text-gray-800 dark:hover:text-gray-200"><svg class="sm:size-3.5 size-3" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg></span> <span>Copy page</span></button> <button class="inline-flex items-center justify-center w-6 max-sm:w-5 h-7 max-sm:h-7 disabled:pointer-events-none text-sm text-gray-500 hover:text-gray-700 dark:hover:text-white rounded-r-md max-sm:rounded-r-sm border border-l transition border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-haspopup="menu" aria-expanded="false" aria-label="Open copy menu"><svg class="transition-transform text-gray-400 overflow-visible sm:size-3.5 size-3 rotate-0" width="1em" height="1em" viewBox="0 0 12 7" fill="none" xmlns="http://www.w3.org/2000/svg"><path d="M1 1L6 6L11 1" stroke="currentColor"></path></svg></button></div> </div> <h1 class="relative group"><a id="compilation" 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="#compilation"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Compilation</span></h1> <h2 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 class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Overview</span></h2> <p data-svelte-h="svelte-az9dn0">Pytorch 2.0 introduced <code>torch.compile</code>, a powerful feature that makes PyTorch code run faster by JIT-compiling PyTorch code into optimized kernels. Key features of <code>torch.compile</code> include:</p> <ul data-svelte-h="svelte-pkkxsd"><li><strong>Performance Improvement</strong>: Significantly speeds up model execution by optimizing the computation graph.</li> <li><strong>Ease of Use</strong>: Requires minimal code changes to implement, making it highly accessible.</li> <li><strong>Compatibility</strong>: Works seamlessly with existing PyTorch code and models.</li></ul> <p data-svelte-h="svelte-gd5iye">When used with Accelerate, <code>torch.compile</code> integrates smoothly into distributed training workflows, allowing you to benefit from both distributed execution and compilation optimizations simultaneously.</p> <p data-svelte-h="svelte-1qamxj">The first execution of compiled code typically takes longer as it includes the compilation time, but subsequent runs are significantly faster. For optimal performance in different scenarios, <code>torch.compile</code> offers various modes like <code>&quot;default&quot;</code>, <code>&quot;reduce-overhead&quot;</code> (which uses CUDA graphs to further reduce overhead), and <code>&quot;max-autotune&quot;</code> (which performs extensive autotuning to find the best kernels for your model).</p> <h2 class="relative group"><a id="using-torchcompile-with-accelerate" 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="#using-torchcompile-with-accelerate"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Using torch.compile with Accelerate</span></h2> <p data-svelte-h="svelte-w2rs5o">Accelerate provides <code>TorchDynamoPlugin</code> for easy and seemless integration of <code>torch.compile</code> into your training scripts.</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START --><span class="hljs-keyword">from</span> accelerate <span class="hljs-keyword">import</span> Accelerator
<span class="hljs-keyword">from</span> accelerate.utils <span class="hljs-keyword">import</span> TorchDynamoPlugin
<span class="hljs-comment"># Configure the compilation backend</span>
dynamo_plugin = TorchDynamoPlugin(
backend=<span class="hljs-string">&quot;inductor&quot;</span>, <span class="hljs-comment"># Options: &quot;inductor&quot;, &quot;aot_eager&quot;, &quot;aot_nvfuser&quot;, etc.</span>
mode=<span class="hljs-string">&quot;default&quot;</span>, <span class="hljs-comment"># Options: &quot;default&quot;, &quot;reduce-overhead&quot;, &quot;max-autotune&quot;</span>
fullgraph=<span class="hljs-literal">True</span>,
dynamic=<span class="hljs-literal">False</span>
)
<span class="hljs-comment"># Initialize accelerator with the plugin</span>
accelerator = Accelerator(dynamo_plugin=dynamo_plugin)
<span class="hljs-comment"># This will apply torch.compile to your model</span>
model = accelerator.prepare(model)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-5uylqs">It is compatible with all other features and plugins of Accelerate, including mixed precision, distributed training (DDP, FSDP, Deepspeed), etc.</p> <h2 class="relative group"><a id="regional-compilation" 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="#regional-compilation"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Regional Compilation</span></h2> <p data-svelte-h="svelte-1fkldr7">Instead of trying to compile the whole model, which usually has a big problem space for optimization. Regional compilation targets repeated blocks of the same class and compiles them sequentially to hit the compiler’s cache. For example, in <code>GPT2LMHeadModel</code>, the repeated block/class is <code>GPT2Block</code>, and can be accessed as <code>model.transformer.h[0]</code>. The rest of the model (e.g model.lm_head) is compiled separately.</p> <p data-svelte-h="svelte-cs154s">This allows us to speed up the compilation overhead / cold start of models like LLMs and Transformers in general.
See <a href="https://pytorch.org/tutorials/recipes/regional_compilation.html" rel="nofollow">https://pytorch.org/tutorials/recipes/regional_compilation.html</a> for more details.</p> <h3 class="relative group"><a id="how-to-use-regional-compilation" 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="#how-to-use-regional-compilation"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>How to Use Regional Compilation</span></h3> <p data-svelte-h="svelte-50g4qf">It can be enabled by setting <code>use_regional_compilation=True</code> in the <code>TorchDynamoPlugin</code> configuration:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START --><span class="hljs-comment"># Configure the compilation backend</span>
dynamo_plugin = TorchDynamoPlugin(
use_regional_compilation=<span class="hljs-literal">True</span>,
... <span class="hljs-comment"># other parameters</span>
)
<span class="hljs-comment"># Initialize accelerator with the plugin</span>
accelerator = Accelerator(dynamo_plugin=dynamo_plugin)
<span class="hljs-comment"># This will apply compile_regions to your model</span>
model = accelerator.prepare(model)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-j2ilao">You could also use the <code>accelerate.utils.compile_regions</code> utility directly the same way you would use <code>torch.compile</code>.</p> <h3 class="relative group"><a id="benefits-of-regional-compilation" 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="#benefits-of-regional-compilation"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Benefits of Regional Compilation</span></h3> <p data-svelte-h="svelte-w1r301">We have conducted extensive benchmarks comparing full compilation and regional compilation using the <code>torch.compile</code> feature in PyTorch. The full results are available in the <a href="https://github.com/huggingface/accelerate/tree/main/benchmarks/torch.compile/regional_compilation" rel="nofollow">accelerate repository</a>. The key findings from our benchmarks are:</p> <ol data-svelte-h="svelte-qsfo8r"><li><strong>Comparable Performance</strong>: Regional compilation delivers performance speedups similar to full compilation, especially for larger models.</li> <li><strong>Faster Compilation</strong>: Regional compilation significantly reduces the time taken to compile models, making it a more efficient choice for deployment.</li> <li><strong>Batch Size Impact</strong>: The performance difference between compilation strategies diminishes with larger batch sizes, indicating that the overhead of compilation is less impactful in those scenarios.</li> <li><strong>Model Size Consideration</strong>: The benefits of regional compilation are more pronounced in larger models, where the compilation time savings can be substantial.</li> <li><strong>Practical Application</strong>: For real-world applications, regional compilation is a practical choice for optimizing training cold start times, especially when working with large models.</li></ol> <h2 class="relative group"><a id="conclusion" 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="#conclusion"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Conclusion</span></h2> <p data-svelte-h="svelte-1iudzer">Both full and regional compilation can significantly speed up your models. Regional compilation offers a practical balance between compilation time and runtime performance, especially for training large models with substantial batch sizes.</p> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/accelerate/blob/main/docs/source/usage_guides/compilation.md" target="_blank"><svg class="mr-1" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M31,16l-7,7l-1.41-1.41L28.17,16l-5.58-5.59L24,9l7,7z"></path><path d="M1,16l7-7l1.41,1.41L3.83,16l5.58,5.59L8,23l-7-7z"></path><path d="M12.419,25.484L17.639,6.552l1.932,0.518L14.351,26.002z"></path></svg> <span data-svelte-h="svelte-zjs2n5"><span class="underline">Update</span> on GitHub</span></a> <p></p>
<script>
{
__sveltekit_1q7nz6m = {
assets: "/docs/accelerate/pr_4021/en",
base: "/docs/accelerate/pr_4021/en",
env: {}
};
const element = document.currentScript.parentElement;
const data = [null,null];
Promise.all([
import("/docs/accelerate/pr_4021/en/_app/immutable/entry/start.8a49e72b.js"),
import("/docs/accelerate/pr_4021/en/_app/immutable/entry/app.1df4d18e.js")
]).then(([kit, app]) => {
kit.start(app, element, {
node_ids: [0, 40],
data,
form: null,
error: null
});
});
}
</script>

Xet Storage Details

Size:
23.1 kB
·
Xet hash:
c00b8df8fb2b23092c144e008fa068fc9cf68606147a91ad821f08417b458233

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