Hugging Face's logo

Buckets:
hf-doc-build
/
doc-dev

Files
xet
hf-doc-build/doc-dev / trl /pr_6976 /en /distributing_training.html
HuggingFaceDocBuilder's picture
HuggingFaceDocBuilder
download
raw
32.6 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Distributing Training&quot;,&quot;local&quot;:&quot;distributing-training&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Multi-GPU Training with TRL&quot;,&quot;local&quot;:&quot;multi-gpu-training-with-trl&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Multi-Node Training&quot;,&quot;local&quot;:&quot;multi-node-training&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Accelerate Configuration&quot;,&quot;local&quot;:&quot;accelerate-configuration&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Launching&quot;,&quot;local&quot;:&quot;launching&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Option 1: Manual Launch (Non-HPC)&quot;,&quot;local&quot;:&quot;option-1-manual-launch-non-hpc&quot;,&quot;sections&quot;:[],&quot;depth&quot;:4},{&quot;title&quot;:&quot;Option 2: SLURM Launch (HPC Clusters)&quot;,&quot;local&quot;:&quot;option-2-slurm-launch-hpc-clusters&quot;,&quot;sections&quot;:[],&quot;depth&quot;:4}],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Further Reading&quot;,&quot;local&quot;:&quot;further-reading&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
<link href="/docs/trl/main/en/_app/immutable/entry/start.F5kHX6Qu.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/chunks/D08CzptW.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/chunks/BmDc0axH.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/entry/app.D-761UUF.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/chunks/DxkSg-qk.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/chunks/CJlUYK3H.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/chunks/DzkIoS9v.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/nodes/0.CMENgSeC.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/chunks/C3p_H5vL.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/nodes/17.teXPRRNR.js" rel="modulepreload">
<link href="/docs/trl/main/en/_app/immutable/chunks/Dsn7vYDT.js" rel="modulepreload">
<!--1lfiq6k--><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Distributing Training&quot;,&quot;local&quot;:&quot;distributing-training&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Multi-GPU Training with TRL&quot;,&quot;local&quot;:&quot;multi-gpu-training-with-trl&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Multi-Node Training&quot;,&quot;local&quot;:&quot;multi-node-training&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Accelerate Configuration&quot;,&quot;local&quot;:&quot;accelerate-configuration&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Launching&quot;,&quot;local&quot;:&quot;launching&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Option 1: Manual Launch (Non-HPC)&quot;,&quot;local&quot;:&quot;option-1-manual-launch-non-hpc&quot;,&quot;sections&quot;:[],&quot;depth&quot;:4},{&quot;title&quot;:&quot;Option 2: SLURM Launch (HPC Clusters)&quot;,&quot;local&quot;:&quot;option-2-slurm-launch-hpc-clusters&quot;,&quot;sections&quot;:[],&quot;depth&quot;:4}],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Further Reading&quot;,&quot;local&quot;:&quot;further-reading&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/><!---->
<link href="/docs/trl/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="distributing-training" 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="#distributing-training"><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>Distributing Training</span></h1><!--]--><!----> <!--[1--><h2 class="relative group"><a id="multi-gpu-training-with-trl" 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="#multi-gpu-training-with-trl"><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>Multi-GPU Training with TRL</span></h2><!--]--><!----> <p>The trainers in TRL use <a href="https://github.com/huggingface/accelerate" rel="nofollow">🤗 Accelerate</a> to enable distributed training across multiple GPUs or nodes. To do so, first create an <a href="https://github.com/huggingface/accelerate" rel="nofollow">🤗 Accelerate</a> config file by running</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 "><!---->accelerate config<!----></pre></div><!----> <p>and answering the questions according to your multi-GPU / multi-node setup. You can then launch distributed training by running:</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 "><!---->accelerate launch train.py<!----></pre></div><!----> <p>We also provide config files in the <a href="https://github.com/huggingface/trl/tree/main/examples/accelerate_configs" rel="nofollow">examples folder</a> that can be used as templates. To use these templates, simply pass the path to the config file when launching a job, e.g.:</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-shell "><!---->accelerate launch --config_file examples/accelerate_configs/multi_gpu.yaml train.py &lt;SCRIPT_ARGS&gt;<!----></pre></div><!----> <p>This automatically distributes the workload across all available GPUs.</p> <p>Under the hood, <a href="https://github.com/huggingface/accelerate" rel="nofollow">🤗 Accelerate</a> creates one model per GPU. Each process:</p> <ul><li>Processes its own batch of data</li> <li>Computes the loss and gradients for that batch</li> <li>Shares gradient updates across all GPUs</li></ul> <p><img src="https://huggingface.co/datasets/trl-lib/documentation-images/resolve/main/multi_gpu.png" alt="multi gpu"/></p> <p>The effective batch size is calculated as: <!----><span class="katex-display"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML" display="block"><semantics><mrow><mtext>Batch Size</mtext><mo>=</mo><mtext>per_device_train_batch_size</mtext><mo>×</mo><mtext>num_devices</mtext><mo>×</mo><mtext>gradient_accumulation_steps</mtext></mrow><annotation encoding="application/x-tex">
\text{Batch Size} = \text{per\_device\_train\_batch\_size} \times \text{num\_devices} \times \text{gradient\_accumulation\_steps}
</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.6944em;"></span><span class="mord text"><span class="mord">Batch Size</span></span><span class="mspace" style="margin-right:0.2778em;"></span><span class="mrel">=</span><span class="mspace" style="margin-right:0.2778em;"></span></span><span class="base"><span class="strut" style="height:1.0044em;vertical-align:-0.31em;"></span><span class="mord text"><span class="mord">per_device_train_batch_size</span></span><span class="mspace" style="margin-right:0.2222em;"></span><span class="mbin">×</span><span class="mspace" style="margin-right:0.2222em;"></span></span><span class="base"><span class="strut" style="height:1.0044em;vertical-align:-0.31em;"></span><span class="mord text"><span class="mord">num_devices</span></span><span class="mspace" style="margin-right:0.2222em;"></span><span class="mbin">×</span><span class="mspace" style="margin-right:0.2222em;"></span></span><span class="base"><span class="strut" style="height:1.0044em;vertical-align:-0.31em;"></span><span class="mord text"><span class="mord">gradient_accumulation_steps</span></span></span></span></span></span><!----></p> <p>To maintain a consistent batch size when scaling to multiple GPUs, make sure to update <code>per_device_train_batch_size</code> and <code>gradient_accumulation_steps</code> accordingly.</p> <p>Example, these configurations are equivalent, and should yield the same results:</p> <table><thead><tr><th>Number of GPUs</th><th>Per device batch size</th><th>Gradient accumulation steps</th><th>Comments</th></tr></thead><tbody><tr><td>1</td><td>32</td><td>1</td><td>Possibly high memory usage, but faster training</td></tr><tr><td>1</td><td>4</td><td>8</td><td>Lower memory usage, slower training</td></tr><tr><td>8</td><td>4</td><td>1</td><td>Multi-GPU to get the best of both worlds</td></tr></tbody></table> <blockquote class="tip"><p>Having one model per GPU can lead to high memory usage, which may not be feasible for large models or low-memory GPUs. In such cases, you can leverage <a href="https://github.com/deepspeedai/DeepSpeed" rel="nofollow">DeepSpeed</a>, which provides optimizations like model sharding, Zero Redundancy Optimizer, mixed precision training, and offloading to CPU or NVMe. Check out our <a href="deepspeed_integration">DeepSpeed Integration</a> guide for more details.</p></blockquote> <blockquote class="tip"><p>Training on very long sequences has its own guide: <a href="long_context_training">Training Beyond 1M Tokens</a>.</p></blockquote> <!--[1--><h2 class="relative group"><a id="multi-node-training" 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="#multi-node-training"><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>Multi-Node Training</span></h2><!--]--><!----> <p>When a single machine doesn’t have enough GPUs, TRL can scale training across multiple machines (nodes) using <a href="https://huggingface.co/docs/accelerate/basic_tutorials/launch#multi-node-training" rel="nofollow">🤗 Accelerate</a>.</p> <!--[2--><h3 class="relative group"><a id="accelerate-configuration" 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="#accelerate-configuration"><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>Accelerate Configuration</span></h3><!--]--><!----> <p>Create an <code>accelerate</code> config file (e.g., <code>multi_node.yaml</code>) for multi-node training. Key fields:</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-yaml "><!----><span class="hljs-attr">compute_environment:</span> <span class="hljs-string">LOCAL_MACHINE</span>
<span class="hljs-attr">distributed_type:</span> <span class="hljs-string">MULTI_GPU</span>
<span class="hljs-attr">num_machines:</span> <span class="hljs-number">2</span>
<span class="hljs-attr">machine_rank:</span> <span class="hljs-number">0</span> <span class="hljs-comment"># 0 for main node, 1 for second node</span>
<span class="hljs-attr">main_process_ip:</span> <span class="hljs-number">10.0</span><span class="hljs-number">.0</span><span class="hljs-number">.1</span> <span class="hljs-comment"># IP of rank 0 node</span>
<span class="hljs-attr">main_process_port:</span> <span class="hljs-number">29500</span>
<span class="hljs-attr">num_processes:</span> <span class="hljs-number">16</span> <span class="hljs-comment"># total processes across nodes</span>
<span class="hljs-attr">mixed_precision:</span> <span class="hljs-string">bf16</span>
<span class="hljs-attr">use_cpu:</span> <span class="hljs-literal">false</span>
<span class="hljs-attr">same_network:</span> <span class="hljs-literal">true</span><!----></pre></div><!----> <p>Adjust <code>num_processes</code> to match the total number of GPUs across all nodes.</p> <blockquote class="note"><p>Replace <code>10.0.0.1</code> with the actual IP address of the rank 0 (main) node.</p></blockquote> <!--[2--><h3 class="relative group"><a id="launching" 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="#launching"><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>Launching</span></h3><!--]--><!----> <!--[3--><h4 class="relative group"><a id="option-1-manual-launch-non-hpc" 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="#option-1-manual-launch-non-hpc"><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>Option 1: Manual Launch (Non-HPC)</span></h4><!--]--><!----> <p>Run the following on each node manually:</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-comment"># Node 0 (main node)</span>
accelerate launch --config_file multi_node.yaml --machine_rank 0 train.py
<span class="hljs-comment"># Node 1</span>
accelerate launch --config_file multi_node.yaml --machine_rank 1 train.py<!----></pre></div><!----> <!--[3--><h4 class="relative group"><a id="option-2-slurm-launch-hpc-clusters" 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="#option-2-slurm-launch-hpc-clusters"><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>Option 2: SLURM Launch (HPC Clusters)</span></h4><!--]--><!----> <p>For clusters using SLURM job scheduler, create a job script (e.g., <code>slurm_job.sh</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-bash "><!----><span class="hljs-meta">#!/bin/bash</span>
<span class="hljs-comment">#SBATCH --nodes=2</span>
<span class="hljs-comment">#SBATCH --gpus-per-node=8</span>
<span class="hljs-comment">#SBATCH --job-name=trl_multi</span>
srun accelerate launch --config_file multi_node.yaml train.py<!----></pre></div><!----> <p>Then submit the job:</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 "><!---->sbatch slurm_job.sh<!----></pre></div><!----> <p>SLURM automatically distributes the training across all requested nodes and GPUs, and <code>srun</code> configures the necessary environment variables for multi-node communication.</p> <p><strong>Key SLURM directives:</strong></p> <ul><li><code>--nodes=2</code>: Request 2 compute nodes</li> <li><code>--gpus-per-node=8</code>: Allocate 8 GPUs per node (16 total)</li> <li><code>--job-name</code>: Label for tracking in the job queue</li></ul> <p>You can combine multi-node with DeepSpeed by setting <code>distributed_type: DEEPSPEED</code> and adding a <code>deepspeed_config</code> block. See the <a href="https://huggingface.co/docs/trl/en/deepspeed_integration" rel="nofollow">DeepSpeed integration guide</a>.</p> <!--[2--><h3 class="relative group"><a id="further-reading" 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="#further-reading"><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>Further Reading</span></h3><!--]--><!----> <ul><li><a href="https://huggingface.co/docs/accelerate/basic_tutorials/launch" rel="nofollow">Accelerate: Launching Scripts</a></li> <li><a href="https://huggingface.co/docs/accelerate/usage_guides/training_zoo" rel="nofollow">Accelerate: Example Zoo</a></li> <li><a href="https://slurm.schedmd.com/" rel="nofollow">SLURM Workload Manager Documentation</a> - For cluster job scheduling</li></ul> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/trl/blob/main/docs/source/distributing_training.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_1b2pyk7 = {
base: "/docs/trl/main/en",
assets: "/docs/trl/main/en"
};
const element = document.currentScript.parentElement;
Promise.all([
import("/docs/trl/main/en/_app/immutable/entry/start.F5kHX6Qu.js"),
import("/docs/trl/main/en/_app/immutable/entry/app.D-761UUF.js")
]).then(([kit, app]) => {
kit.start(app, element, {
node_ids: [0, 17],
data: [null,null],
form: null,
error: null
});
});
}
</script>

Xet Storage Details

Size:
32.6 kB
·
Xet hash:
4692d166606ef9c853009eb968dcdce91c1a6f04e15540aa5e5f72a25a279ddf

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