text-generation-inference/main/en/conceptual/quantization.html

<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Quantization&quot;,&quot;local&quot;:&quot;quantization&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Quantization with bitsandbytes, EETQ &amp; fp8&quot;,&quot;local&quot;:&quot;quantization-with-bitsandbytes-eetq--fp8&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Quantization with GPTQ&quot;,&quot;local&quot;:&quot;quantization-with-gptq&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
		<link href="/docs/text-generation-inference/main/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/entry/start.1810066f.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/chunks/scheduler.362310b7.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/chunks/singletons.fa2b0eb7.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/chunks/index.7f53ec41.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/chunks/paths.284aef40.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/entry/app.8cfc1931.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/chunks/index.57dfc70d.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/nodes/0.543c9bd9.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/chunks/each.e59479a4.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/nodes/18.1a8509fd.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/chunks/CodeBlock.d3c47f83.js">
		<link rel="modulepreload" href="/docs/text-generation-inference/main/en/_app/immutable/chunks/EditOnGithub.9633c464.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Quantization&quot;,&quot;local&quot;:&quot;quantization&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Quantization with bitsandbytes, EETQ &amp; fp8&quot;,&quot;local&quot;:&quot;quantization-with-bitsandbytes-eetq--fp8&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Quantization with GPTQ&quot;,&quot;local&quot;:&quot;quantization-with-gptq&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END -->      <p></p>  <h1 class="relative group"><a id="quantization" 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="#quantization"><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>Quantization</span></h1> <p data-svelte-h="svelte-1uhon0f">TGI offers many quantization schemes to run LLMs effectively and fast based on your use-case. TGI supports GPTQ, AWQ, bits-and-bytes, EETQ, Marlin, EXL2 and fp8 quantization.</p> <p data-svelte-h="svelte-m25niz">To leverage GPTQ, AWQ, Marlin and EXL2 quants, you must provide pre-quantized weights. Whereas for bits-and-bytes, EETQ and fp8, weights are quantized by TGI on the fly.</p> <p data-svelte-h="svelte-1uq49rw">We recommend using the official quantization scripts for creating your quants:</p> <ol data-svelte-h="svelte-1fjxs52"><li><a href="https://github.com/casper-hansen/AutoAWQ/blob/main/examples/quantize.py" rel="nofollow">AWQ</a></li> <li><a href="https://github.com/AutoGPTQ/AutoGPTQ/blob/main/examples/quantization/basic_usage.py" rel="nofollow">GPTQ/ Marlin</a></li> <li><a href="https://github.com/turboderp/exllamav2/blob/master/doc/convert.md" rel="nofollow">EXL2</a></li></ol> <p data-svelte-h="svelte-dlxq44">For on-the-fly quantization you simply need to pass one of the supported quantization types and TGI takes care of the rest.</p>  <h2 class="relative group"><a id="quantization-with-bitsandbytes-eetq--fp8" 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="#quantization-with-bitsandbytes-eetq--fp8"><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>Quantization with bitsandbytes, EETQ &amp; fp8</span></h2> <p data-svelte-h="svelte-aqqk8a">bitsandbytes is a library used to apply 8-bit and 4-bit quantization to models. Unlike GPTQ quantization, bitsandbytes doesn’t require a calibration dataset or any post-processing – weights are automatically quantized on load. However, inference with bitsandbytes is slower than GPTQ or FP16 precision.</p> <p data-svelte-h="svelte-ktr7x0">8-bit quantization enables multi-billion parameter scale models to fit in smaller hardware without degrading performance too much.
In TGI, you can use 8-bit quantization by adding <code>--quantize bitsandbytes</code> like below 👇</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 -->docker run --gpus all --shm-size 1g -p 8080:80 -v <span class="hljs-variable">$volume</span>:/data ghcr.io/huggingface/text-generation-inference:latest --model-id <span class="hljs-variable">$model</span> --quantize bitsandbytes<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1kt0l3g">4-bit quantization is also possible with bitsandbytes. You can choose one of the following 4-bit data types: 4-bit float (<code>fp4</code>), or 4-bit <code>NormalFloat</code> (<code>nf4</code>). These data types were introduced in the context of parameter-efficient fine-tuning, but you can apply them for inference by automatically converting the model weights on load.</p> <p data-svelte-h="svelte-113ygv">In TGI, you can use 4-bit quantization by adding <code>--quantize bitsandbytes-nf4</code> or <code>--quantize bitsandbytes-fp4</code> like below 👇</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 -->docker run --gpus all --shm-size 1g -p 8080:80 -v <span class="hljs-variable">$volume</span>:/data ghcr.io/huggingface/text-generation-inference:latest --model-id <span class="hljs-variable">$model</span> --quantize bitsandbytes-nf4<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-bx1jaz">You can get more information about 8-bit quantization by reading this <a href="https://huggingface.co/blog/hf-bitsandbytes-integration" rel="nofollow">blog post</a>, and 4-bit quantization by reading <a href="https://huggingface.co/blog/4bit-transformers-bitsandbytes" rel="nofollow">this blog post</a>.</p> <p data-svelte-h="svelte-17aol0b">Similarly you can use pass you can pass <code>--quantize eetq</code> or <code>--quantize fp8</code> for respective quantization schemes.</p> <p data-svelte-h="svelte-1likkzk">In addition to this, TGI allows creating GPTQ quants directly by passing the model weights and a calibration dataset.</p>  <h2 class="relative group"><a id="quantization-with-gptq" 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="#quantization-with-gptq"><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>Quantization with GPTQ</span></h2> <p data-svelte-h="svelte-aix1yq">GPTQ is a post-training quantization method to make the model smaller. It quantizes the layers by finding a compressed version of that weight, that will yield a minimum mean squared error like below 👇</p> <p>Given a layer<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>l</mi></mrow><annotation encoding="application/x-tex">l</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 mathnormal" style="margin-right:0.01968em;">l</span></span></span></span><!-- HTML_TAG_END --> with weight matrix<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>W</mi><mi>l</mi></msub></mrow><annotation encoding="application/x-tex">W_{l}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.8333em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><span style="top:-2.55em;margin-left:-0.1389em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mathnormal mtight" style="margin-right:0.01968em;">l</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span></span></span></span><!-- HTML_TAG_END --> and layer input<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>X</mi><mi>l</mi></msub></mrow><annotation encoding="application/x-tex">X_{l}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.8333em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal" style="margin-right:0.07847em;">X</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><span style="top:-2.55em;margin-left:-0.0785em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mathnormal mtight" style="margin-right:0.01968em;">l</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span></span></span></span><!-- HTML_TAG_END -->, find quantized weight<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mspace linebreak="newline"></mspace><mi>h</mi><mi>a</mi><mi>t</mi><msub><mi>W</mi><mi>l</mi></msub></mrow><annotation encoding="application/x-tex">\\hat{W}_{l}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="mspace newline"></span><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord mathnormal">ha</span><span class="mord mathnormal">t</span><span class="mord"><span class="mord"><span class="mord mathnormal" style="margin-right:0.13889em;">W</span></span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><span style="top:-2.55em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mathnormal mtight" style="margin-right:0.01968em;">l</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span></span></span></span><!-- HTML_TAG_END -->:
<!-- HTML_TAG_START --><span class="katex-display"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML" display="block"><semantics><mrow><mo stretchy="false">(</mo><msup><msub><mover accent="true"><mi>W</mi><mo>^</mo></mover><mi>l</mi></msub><mo lspace="0em" rspace="0em">∗</mo></msup><mo>=</mo><mi>a</mi><mi>r</mi><mi>g</mi><mi>m</mi><mi>i</mi><msub><mi>n</mi><mover accent="true"><msub><mi>W</mi><mi>l</mi></msub><mo>^</mo></mover></msub><mi mathvariant="normal">∣</mi><mi mathvariant="normal">∣</mi><msub><mi>W</mi><mi>l</mi></msub><mi>X</mi><mo>−</mo><msub><mover accent="true"><mi>W</mi><mo>^</mo></mover><mi>l</mi></msub><mi>X</mi><mi mathvariant="normal">∣</mi><msubsup><mi mathvariant="normal">∣</mi><mn>2</mn><mn>2</mn></msubsup><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">({\hat{W}_{l}}^{*} = argmin_{\hat{W_{l}}} ||W_{l}X-\hat{W}_{l}X||^{2}_{2})</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1.2754em;vertical-align:-0.25em;"></span><span class="mopen">(</span><span class="mord"><span class="mord"><span class="mord"><span class="mord accent"><span class="vlist-t"><span class="vlist-r"><span class="vlist" style="height:0.9468em;"><span style="top:-3em;"><span class="pstrut" style="height:3em;"></span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span></span><span style="top:-3.2523em;"><span class="pstrut" style="height:3em;"></span><span class="accent-body" style="left:-0.25em;"><span class="mord">^</span></span></span></span></span></span></span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><span style="top:-2.55em;margin-left:-0.1389em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mathnormal mtight" style="margin-right:0.01968em;">l</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span></span><span class="msupsub"><span class="vlist-t"><span class="vlist-r"><span class="vlist" style="height:1.0254em;"><span style="top:-3.3997em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">∗</span></span></span></span></span></span></span></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:1.1738em;vertical-align:-0.4238em;"></span><span class="mord mathnormal">a</span><span class="mord mathnormal" style="margin-right:0.02778em;">r</span><span class="mord mathnormal" style="margin-right:0.03588em;">g</span><span class="mord mathnormal">mi</span><span class="mord"><span class="mord mathnormal">n</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3448em;"><span style="top:-2.3821em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord accent mtight"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.9468em;"><span style="top:-2.7em;"><span class="pstrut" style="height:2.7em;"></span><span class="mord mtight"><span class="mord mathnormal mtight" style="margin-right:0.13889em;">W</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3448em;"><span style="top:-2.3488em;margin-left:-0.1389em;margin-right:0.0714em;"><span class="pstrut" style="height:2.5em;"></span><span class="sizing reset-size3 size1 mtight"><span class="mord mtight"><span class="mord mathnormal mtight" style="margin-right:0.01968em;">l</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.1512em;"><span></span></span></span></span></span></span></span><span style="top:-2.9523em;"><span class="pstrut" style="height:2.7em;"></span><span class="accent-body" style="left:-0.25em;"><span class="mord mtight">^</span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.1512em;"><span></span></span></span></span></span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.4238em;"><span></span></span></span></span></span></span><span class="mord">∣∣</span><span class="mord"><span class="mord mathnormal" style="margin-right:0.13889em;">W</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><span style="top:-2.55em;margin-left:-0.1389em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mathnormal mtight" style="margin-right:0.01968em;">l</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mord mathnormal" style="margin-right:0.07847em;">X</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:1.1968em;vertical-align:-0.25em;"></span><span class="mord"><span class="mord accent"><span class="vlist-t"><span class="vlist-r"><span class="vlist" style="height:0.9468em;"><span style="top:-3em;"><span class="pstrut" style="height:3em;"></span><span class="mord mathnormal" style="margin-right:0.13889em;">W</span></span><span style="top:-3.2523em;"><span class="pstrut" style="height:3em;"></span><span class="accent-body" style="left:-0.25em;"><span class="mord">^</span></span></span></span></span></span></span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><span style="top:-2.55em;margin-left:-0.1389em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mathnormal mtight" style="margin-right:0.01968em;">l</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mord mathnormal" style="margin-right:0.07847em;">X</span><span class="mord">∣</span><span class="mord"><span class="mord">∣</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.8641em;"><span style="top:-2.453em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">2</span></span></span></span><span style="top:-3.113em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">2</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.247em;"><span></span></span></span></span></span></span><span class="mclose">)</span></span></span></span></span><!-- HTML_TAG_END --></p> <p data-svelte-h="svelte-14up0yw">TGI allows you to both run an already GPTQ quantized model (see available models <a href="https://huggingface.co/models?search=gptq" rel="nofollow">here</a>) or quantize a model of your choice using quantization script. You can run a quantized model by simply passing —quantize like below 👇</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 -->docker run --gpus all --shm-size 1g -p 8080:80 -v <span class="hljs-variable">$volume</span>:/data ghcr.io/huggingface/text-generation-inference:latest --model-id <span class="hljs-variable">$model</span> --quantize gptq<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-14lnxb8">Note that TGI’s GPTQ implementation doesn’t use <a href="https://github.com/PanQiWei/AutoGPTQ" rel="nofollow">AutoGPTQ</a> under the hood. However, models quantized using AutoGPTQ or Optimum can still be served by TGI.</p> <p data-svelte-h="svelte-1drv0r">To quantize a given model using GPTQ with a calibration dataset, simply run</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 -->text-generation-server quantize tiiuae/falcon-40b /data/falcon-40b-gptq
<span class="hljs-comment"># Add --upload-to-model-id MYUSERNAME/falcon-40b to push the created model to the hub directly</span><!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-f1w88s">This will create a new directory with the quantized files which you can use with,</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 -->text-generation-launcher --model-id /data/falcon-40b-gptq/ --sharded <span class="hljs-literal">true</span> --num-shard 2 --quantize gptq<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-32p5y6">You can learn more about the quantization options by running <code>text-generation-server quantize --help</code>.</p> <p data-svelte-h="svelte-a14ru9">If you wish to do more with GPTQ models (e.g. train an adapter on top), you can read about transformers GPTQ integration <a href="https://huggingface.co/blog/gptq-integration" rel="nofollow">here</a>.
You can learn more about GPTQ from the <a href="https://arxiv.org/pdf/2210.17323.pdf" rel="nofollow">paper</a>.</p> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/text-generation-inference/blob/main/docs/source/conceptual/quantization.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_1dfb6m4 = {
						assets: "/docs/text-generation-inference/main/en",
						base: "/docs/text-generation-inference/main/en",
						env: {}
					};

					const element = document.currentScript.parentElement;

					const data = [null,null];

					Promise.all([
						import("/docs/text-generation-inference/main/en/_app/immutable/entry/start.1810066f.js"),
						import("/docs/text-generation-inference/main/en/_app/immutable/entry/app.8cfc1931.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 18],
							data,
							form: null,
							error: null
						});
					});
				}
			</script>