text-embeddings-inference/main/en/index.html

<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Text Embeddings Inference&quot;,&quot;local&quot;:&quot;text-embeddings-inference&quot;,&quot;sections&quot;:[],&quot;depth&quot;:1}">
		<link href="/docs/text-embeddings-inference/main/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/entry/start.5929ce5e.js">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/chunks/scheduler.b108d059.js">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/chunks/singletons.8ba50d2e.js">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/chunks/paths.0433c982.js">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/entry/app.99d3b526.js">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/chunks/index.008de539.js">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/nodes/0.edd78360.js">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/nodes/5.173e3213.js">
		<link rel="modulepreload" href="/docs/text-embeddings-inference/main/en/_app/immutable/chunks/EditOnGithub.d1c48e3d.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Text Embeddings Inference&quot;,&quot;local&quot;:&quot;text-embeddings-inference&quot;,&quot;sections&quot;:[],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END -->      <p></p>   <h1 class="relative group"><a id="text-embeddings-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="#text-embeddings-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>Text Embeddings Inference</span></h1> <p data-svelte-h="svelte-g5xgrp">Text Embeddings Inference (TEI) is a comprehensive toolkit designed for efficient deployment and serving of open source
text embeddings models. It enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE, and E5.</p> <p data-svelte-h="svelte-1afmapv">TEI offers multiple features tailored to optimize the deployment process and enhance overall performance.</p> <p data-svelte-h="svelte-dcww01"><strong>Key Features:</strong></p> <ul data-svelte-h="svelte-ziger0"><li><strong>Streamlined Deployment:</strong> TEI eliminates the need for a model graph compilation step for an easier deployment process.</li> <li><strong>Efficient Resource Utilization:</strong> Benefit from small Docker images and rapid boot times, allowing for true serverless capabilities.</li> <li><strong>Dynamic Batching:</strong> TEI incorporates token-based dynamic batching thus optimizing resource utilization during inference.</li> <li><strong>Optimized Inference:</strong> TEI leverages <a href="https://github.com/HazyResearch/flash-attention" rel="nofollow">Flash Attention</a>, <a href="https://github.com/huggingface/candle" rel="nofollow">Candle</a>, and <a href="https://docs.nvidia.com/cuda/cublas/#using-the-cublaslt-api" rel="nofollow">cuBLASLt</a> by using optimized transformers code for inference.</li> <li><strong>Safetensors weight loading:</strong> TEI loads <a href="https://github.com/huggingface/safetensors" rel="nofollow">Safetensors</a> weights for faster boot times.</li> <li><strong>Production-Ready:</strong> TEI supports distributed tracing through Open Telemetry and exports Prometheus metrics.</li></ul> <p data-svelte-h="svelte-q2uj9l"><strong>Benchmarks</strong></p> <p data-svelte-h="svelte-1rcc639">Benchmark for <a href="https://hf.co/BAAI/bge-large-en-v1.5" rel="nofollow">BAAI/bge-base-en-v1.5</a> on an NVIDIA A10 with a sequence length of 512 tokens:</p> <p data-svelte-h="svelte-1tzh58b"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/tei/bs1-lat.png" width="400" alt="Latency comparison for batch size of 1"> <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/tei/bs1-tp.png" width="400" alt="Throughput comparison for batch size of 1"></p> <p data-svelte-h="svelte-1ccb8mt"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/tei/bs32-lat.png" width="400" alt="Latency comparison for batch size of 32"> <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/tei/bs32-tp.png" width="400" alt="Throughput comparison for batch size of 32"></p> <p data-svelte-h="svelte-1hbycvq"><strong>Getting Started:</strong></p> <p data-svelte-h="svelte-15zyol4">To start using TEI, check the <a href="quick_tour">Quick Tour</a> guide.</p> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/text-embeddings-inference/blob/main/docs/source/en/index.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_1a9ukzg = {
						assets: "/docs/text-embeddings-inference/main/en",
						base: "/docs/text-embeddings-inference/main/en",
						env: {}
					};

					const element = document.currentScript.parentElement;

					const data = [null,null];

					Promise.all([
						import("/docs/text-embeddings-inference/main/en/_app/immutable/entry/start.5929ce5e.js"),
						import("/docs/text-embeddings-inference/main/en/_app/immutable/entry/app.99d3b526.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 5],
							data,
							form: null,
							error: null
						});
					});
				}
			</script>