optimum-intel/pr_1297/en/ipex/inference.html

<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Inference&quot;,&quot;local&quot;:&quot;inference&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Loading&quot;,&quot;local&quot;:&quot;loading&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
		<link href="/docs/optimum.intel/pr_1297/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/entry/start.28f0229f.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/chunks/scheduler.7da89386.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/chunks/singletons.d9541c88.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/chunks/paths.28390f98.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/entry/app.027f6c57.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/chunks/index.20910acc.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/nodes/0.0560b75e.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/chunks/each.e59479a4.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/nodes/4.6182c41d.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/chunks/CodeBlock.143bd81e.js">
		<link rel="modulepreload" href="/docs/optimum.intel/pr_1297/en/_app/immutable/chunks/getInferenceSnippets.ab845209.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Inference&quot;,&quot;local&quot;:&quot;inference&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Loading&quot;,&quot;local&quot;:&quot;loading&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END -->      <p></p>   <h1 class="relative group"><a id="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="#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>Inference</span></h1> <p data-svelte-h="svelte-3t5r71">Optimum Intel can be used to load models from the <a href="https://huggingface.co/models" rel="nofollow">Hub</a> and create pipelines to run inference with IPEX optimizations (including patching with custom operators, weight prepacking and graph mode) on a variety of Intel processors. For now support is only enabled for CPUs.</p>  <h2 class="relative group"><a id="loading" 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="#loading"><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>Loading</span></h2> <p data-svelte-h="svelte-vwwa0u">You can load your model and apply IPEX optimizations (apply torch.compile except text-generation tasks). For supported architectures like LLaMA, BERT and ViT, further optimizations will be applied by patching the model to use custom operators.
For now, support is enabled for Intel CPU/GPU. Previous models converted to TorchScript will be deprecated in v1.22.</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 -->  import torch
  from transformers import AutoTokenizer, pipeline
<span class="hljs-deletion">- from transformers import AutoModelForCausalLM</span>
<span class="hljs-addition">+ from optimum.intel import IPEXModelForCausalLM</span>

  model_id = &quot;gpt2&quot;
<span class="hljs-deletion">- model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16)</span>
<span class="hljs-addition">+ model = IPEXModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16)</span>
  tokenizer = AutoTokenizer.from_pretrained(model_id)
  pipe = pipeline(&quot;text-generation&quot;, model=model, tokenizer=tokenizer)
  results = pipe(&quot;He&#x27;s a dreadful magician and&quot;)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-sgii55">As shown in the table below, each task is associated with a class enabling to automatically load your model.</p> <table data-svelte-h="svelte-1wvyvke"><thead><tr><th>Auto Class</th> <th>Task</th></tr></thead> <tbody><tr><td><code>IPEXModelForSequenceClassification</code></td> <td><code>text-classification</code></td></tr> <tr><td><code>IPEXModelForTokenClassification</code></td> <td><code>token-classification</code></td></tr> <tr><td><code>IPEXModelForQuestionAnswering</code></td> <td><code>question-answering</code></td></tr> <tr><td><code>IPEXModelForImageClassification</code></td> <td><code>image-classification</code></td></tr> <tr><td><code>IPEXModel</code></td> <td><code>feature-extraction</code></td></tr> <tr><td><code>IPEXModelForMaskedLM</code></td> <td><code>fill-mask</code></td></tr> <tr><td><code>IPEXModelForAudioClassification</code></td> <td><code>audio-classification</code></td></tr> <tr><td><code>IPEXModelForCausalLM</code></td> <td><code>text-generation</code></td></tr> <tr><td><code>IPEXModelForSeq2SeqLM</code></td> <td><code>text2text-generation</code></td></tr></tbody></table> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/optimum-intel/blob/main/docs/source/ipex/inference.mdx" 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_1gax26v = {
						assets: "/docs/optimum.intel/pr_1297/en",
						base: "/docs/optimum.intel/pr_1297/en",
						env: {}
					};

					const element = document.currentScript.parentElement;

					const data = [null,null];

					Promise.all([
						import("/docs/optimum.intel/pr_1297/en/_app/immutable/entry/start.28f0229f.js"),
						import("/docs/optimum.intel/pr_1297/en/_app/immutable/entry/app.027f6c57.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 4],
							data,
							form: null,
							error: null
						});
					});
				}
			</script>