transformers/main/en/perf_infer_cpu.html

<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;CPU inference&quot;,&quot;local&quot;:&quot;cpu-inference&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;BetterTransformer&quot;,&quot;local&quot;:&quot;bettertransformer&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;TorchScript&quot;,&quot;local&quot;:&quot;torchscript&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;IPEX graph optimization&quot;,&quot;local&quot;:&quot;ipex-graph-optimization&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;🤗 Optimum&quot;,&quot;local&quot;:&quot;-optimum&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
		<link href="/docs/transformers/main/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/entry/start.2135b7e6.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/scheduler.25b97de1.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/singletons.0f2b7d5f.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/index.e188933d.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/paths.3d04d2c6.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/entry/app.24372c84.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/index.d9030fc9.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/nodes/0.026d2fdd.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/each.e59479a4.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/nodes/366.9960c31d.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/Tip.baa67368.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/CodeBlock.e6cd0d95.js">
		<link rel="modulepreload" href="/docs/transformers/main/en/_app/immutable/chunks/EditOnGithub.91d95064.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;CPU inference&quot;,&quot;local&quot;:&quot;cpu-inference&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;BetterTransformer&quot;,&quot;local&quot;:&quot;bettertransformer&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;TorchScript&quot;,&quot;local&quot;:&quot;torchscript&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;IPEX graph optimization&quot;,&quot;local&quot;:&quot;ipex-graph-optimization&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;🤗 Optimum&quot;,&quot;local&quot;:&quot;-optimum&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END -->      <p></p>   <h1 class="relative group"><a id="cpu-inference" 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="#cpu-inference"><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>CPU inference</span></h1> <p data-svelte-h="svelte-1i824r1">With some optimizations, it is possible to efficiently run large model inference on a CPU. One of these optimization techniques involves compiling the PyTorch code into an intermediate format for high-performance environments like C++. The other technique fuses multiple operations into one kernel to reduce the overhead of running each operation separately.</p> <p data-svelte-h="svelte-18xf8m9">You’ll learn how to use <a href="https://pytorch.org/blog/a-better-transformer-for-fast-transformer-encoder-inference/" rel="nofollow">BetterTransformer</a> for faster inference, and how to convert your PyTorch code to <a href="https://pytorch.org/tutorials/beginner/Intro_to_TorchScript_tutorial.html" rel="nofollow">TorchScript</a>. If you’re using an Intel CPU, you can also use <a href="https://intel.github.io/intel-extension-for-pytorch/cpu/latest/tutorials/features.html#graph-optimization" rel="nofollow">graph optimizations</a> from <a href="https://intel.github.io/intel-extension-for-pytorch/cpu/latest/index.html" rel="nofollow">Intel Extension for PyTorch</a> to boost inference speed even more. Finally, learn how to use 🤗 Optimum to accelerate inference with ONNX Runtime or OpenVINO (if you’re using an Intel CPU).</p>  <h2 class="relative group"><a id="bettertransformer" 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="#bettertransformer"><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>BetterTransformer</span></h2> <p data-svelte-h="svelte-mytrcu">BetterTransformer accelerates inference with its fastpath (native PyTorch specialized implementation of Transformer functions) execution. The two optimizations in the fastpath execution are:</p> <ol data-svelte-h="svelte-1b2ln7l"><li>fusion, which combines multiple sequential operations into a single “kernel” to reduce the number of computation steps</li> <li>skipping the inherent sparsity of padding tokens to avoid unnecessary computation with nested tensors</li></ol> <p data-svelte-h="svelte-1ntl6a3">BetterTransformer also converts all attention operations to use the more memory-efficient <a href="https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention" rel="nofollow">scaled dot product attention</a>.</p>  <div class="course-tip  bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-scbfw6">BetterTransformer is not supported for all models. Check this <a href="https://huggingface.co/docs/optimum/bettertransformer/overview#supported-models" rel="nofollow">list</a> to see if a model supports BetterTransformer.</p></div> <p data-svelte-h="svelte-1qwvxdv">Before you start, make sure you have 🤗 Optimum <a href="https://huggingface.co/docs/optimum/installation" rel="nofollow">installed</a>.</p> <p data-svelte-h="svelte-1ufbxrn">Enable BetterTransformer with the <a href="/docs/transformers/main/en/main_classes/model#transformers.PreTrainedModel.to_bettertransformer">PreTrainedModel.to_bettertransformer()</a> method:</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> transformers <span class="hljs-keyword">import</span> AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained(<span class="hljs-string">&quot;bigcode/starcoder&quot;</span>)
model.to_bettertransformer()<!-- HTML_TAG_END --></pre></div>  <h2 class="relative group"><a id="torchscript" 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="#torchscript"><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>TorchScript</span></h2> <p data-svelte-h="svelte-kgrwv1">TorchScript is an intermediate PyTorch model representation that can be run in production environments where performance is important. You can train a model in PyTorch and then export it to TorchScript to free the model from Python performance constraints. PyTorch <a href="https://pytorch.org/docs/stable/generated/torch.jit.trace.html" rel="nofollow">traces</a> a model to return a <code>ScriptFunction</code> that is optimized with just-in-time compilation (JIT). Compared to the default eager mode, JIT mode in PyTorch typically yields better performance for inference using optimization techniques like operator fusion.</p> <p data-svelte-h="svelte-i0x3az">For a gentle introduction to TorchScript, see the <a href="https://pytorch.org/tutorials/beginner/Intro_to_TorchScript_tutorial.html" rel="nofollow">Introduction to PyTorch TorchScript</a> tutorial.</p> <p data-svelte-h="svelte-sano8o">With the <a href="/docs/transformers/main/en/main_classes/trainer#transformers.Trainer">Trainer</a> class, you can enable JIT mode for CPU inference by setting the <code>--jit_mode_eval</code> flag:</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 -->python run_qa.py \
--model_name_or_path csarron/bert-base-uncased-squad-v1 \
--dataset_name squad \
--do_eval \
--max_seq_length 384 \
--doc_stride 128 \
--output_dir /tmp/ \
--no_cuda \
--jit_mode_eval<!-- HTML_TAG_END --></pre></div>  <div class="course-tip course-tip-orange bg-gradient-to-br dark:bg-gradient-to-r before:border-orange-500 dark:before:border-orange-800 from-orange-50 dark:from-gray-900 to-white dark:to-gray-950 border border-orange-50 text-orange-700 dark:text-gray-400"><p data-svelte-h="svelte-t0nelt">For PyTorch &gt;= 1.14.0, JIT-mode could benefit any model for prediction and evaluation since the dict input is supported in <code>jit.trace</code>.</p> <p data-svelte-h="svelte-t4iueh">For PyTorch &lt; 1.14.0, JIT-mode could benefit a model if its forward parameter order matches the tuple input order in <code>jit.trace</code>, such as a question-answering model. If the forward parameter order does not match the tuple input order in <code>jit.trace</code>, like a text classification model, <code>jit.trace</code> will fail and we are capturing this with the exception here to make it fallback. Logging is used to notify users.</p></div>  <h2 class="relative group"><a id="ipex-graph-optimization" 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="#ipex-graph-optimization"><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>IPEX graph optimization</span></h2> <p data-svelte-h="svelte-1yofu8o">Intel® Extension for PyTorch (IPEX) provides further optimizations in JIT mode for Intel CPUs, and we recommend combining it with TorchScript for even faster performance. The IPEX <a href="https://intel.github.io/intel-extension-for-pytorch/cpu/latest/tutorials/features/graph_optimization.html" rel="nofollow">graph optimization</a> fuses operations like Multi-head attention, Concat Linear, Linear + Add, Linear + Gelu, Add + LayerNorm, and more.</p> <p data-svelte-h="svelte-1xouf9s">To take advantage of these graph optimizations, make sure you have IPEX <a href="https://intel.github.io/intel-extension-for-pytorch/cpu/latest/tutorials/installation.html" rel="nofollow">installed</a>:</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 -->pip install intel_extension_for_pytorch<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1mtujyw">Set the <code>--use_ipex</code> and <code>--jit_mode_eval</code> flags in the <a href="/docs/transformers/main/en/main_classes/trainer#transformers.Trainer">Trainer</a> class to enable JIT mode with the graph optimizations:</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 -->python run_qa.py \
--model_name_or_path csarron/bert-base-uncased-squad-v1 \
--dataset_name squad \
--do_eval \
--max_seq_length 384 \
--doc_stride 128 \
--output_dir /tmp/ \
--no_cuda \
--use_ipex \
--jit_mode_eval<!-- HTML_TAG_END --></pre></div>  <h2 class="relative group"><a id="-optimum" 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="#-optimum"><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>🤗 Optimum</span></h2>  <div class="course-tip  bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-awtbyy">Learn more details about using ORT with 🤗 Optimum in the <a href="https://huggingface.co/docs/optimum/onnxruntime/usage_guides/models" rel="nofollow">Optimum Inference with ONNX Runtime</a> guide. This section only provides a brief and simple example.</p></div> <p data-svelte-h="svelte-flw3r7">ONNX Runtime (ORT) is a model accelerator that runs inference on CPUs by default. ORT is supported by 🤗 Optimum which can be used in 🤗 Transformers, without making too many changes to your code. You only need to replace the 🤗 Transformers <code>AutoClass</code> with its equivalent <a href="https://huggingface.co/docs/optimum/main/en/onnxruntime/package_reference/modeling_ort#optimum.onnxruntime.ORTModel" rel="nofollow">ORTModel</a> for the task you’re solving, and load a checkpoint in the ONNX format.</p> <p data-svelte-h="svelte-1khkfx8">For example, if you’re running inference on a question answering task, load the <a href="https://huggingface.co/optimum/roberta-base-squad2" rel="nofollow">optimum/roberta-base-squad2</a> checkpoint which contains a <code>model.onnx</code> file:</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> transformers <span class="hljs-keyword">import</span> AutoTokenizer, pipeline
<span class="hljs-keyword">from</span> optimum.onnxruntime <span class="hljs-keyword">import</span> ORTModelForQuestionAnswering

model = ORTModelForQuestionAnswering.from_pretrained(<span class="hljs-string">&quot;optimum/roberta-base-squad2&quot;</span>)
tokenizer = AutoTokenizer.from_pretrained(<span class="hljs-string">&quot;deepset/roberta-base-squad2&quot;</span>)

onnx_qa = pipeline(<span class="hljs-string">&quot;question-answering&quot;</span>, model=model, tokenizer=tokenizer)

question = <span class="hljs-string">&quot;What&#x27;s my name?&quot;</span>
context = <span class="hljs-string">&quot;My name is Philipp and I live in Nuremberg.&quot;</span>
pred = onnx_qa(question, context)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1ko1nlh">If you have an Intel CPU, take a look at 🤗 <a href="https://huggingface.co/docs/optimum/intel/index" rel="nofollow">Optimum Intel</a> which supports a variety of compression techniques (quantization, pruning, knowledge distillation) and tools for converting models to the <a href="https://huggingface.co/docs/optimum/intel/inference" rel="nofollow">OpenVINO</a> format for higher performance inference.</p> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/transformers/blob/main/docs/source/en/perf_infer_cpu.md" target="_blank"><span data-svelte-h="svelte-1kd6by1">&lt;</span> <span data-svelte-h="svelte-x0xyl0">&gt;</span> <span data-svelte-h="svelte-1dajgef"><span class="underline ml-1.5">Update</span> on GitHub</span></a>  <p></p> 
			
			<script>
				{
					__sveltekit_1xexzbk = {
						assets: "/docs/transformers/main/en",
						base: "/docs/transformers/main/en",
						env: {}
					};

					const element = document.currentScript.parentElement;

					const data = [null,null];

					Promise.all([
						import("/docs/transformers/main/en/_app/immutable/entry/start.2135b7e6.js"),
						import("/docs/transformers/main/en/_app/immutable/entry/app.24372c84.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 366],
							data,
							form: null,
							error: null
						});
					});
				}
			</script>