Buckets:

hf-doc-build
/

doc-dev

Files
xet
hf-doc-build/doc-dev / transformers /pr_45649 /en /quantization /overview.html
HuggingFaceDocBuilder
about 12 hours ago
download
raw
29.2 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Overview&quot;,&quot;local&quot;:&quot;overview&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Two ways to use a quantized model&quot;,&quot;local&quot;:&quot;two-ways-to-use-a-quantized-model&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Quantize a model while loading&quot;,&quot;local&quot;:&quot;quantize-a-model-while-loading&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Load a pre-quantized checkpoint&quot;,&quot;local&quot;:&quot;load-a-pre-quantized-checkpoint&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;GGUF checkpoints&quot;,&quot;local&quot;:&quot;gguf-checkpoints&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Save a quantized model&quot;,&quot;local&quot;:&quot;save-a-quantized-model&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Choose a method&quot;,&quot;local&quot;:&quot;choose-a-method&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Other tools&quot;,&quot;local&quot;:&quot;other-tools&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Resources&quot;,&quot;local&quot;:&quot;resources&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Next steps&quot;,&quot;local&quot;:&quot;next-steps&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
		<link href="/docs/transformers/pr_45649/en/_app/immutable/entry/start.Cf6g_lQr.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/chunks/bcIanF0A.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/chunks/CBySwR1V.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/entry/app.Ckn5M_jH.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/chunks/Cdb1BouH.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/chunks/dUGfULTE.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/chunks/D5-p1Lz5.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/nodes/0.92KZZ6l3.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/chunks/Conj1S5Z.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/nodes/713.Cbm-mVZb.js" rel="modulepreload">
		<link href="/docs/transformers/pr_45649/en/_app/immutable/chunks/DoQ7MZQh.js" rel="modulepreload">
		<!--cubdu--><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Overview&quot;,&quot;local&quot;:&quot;overview&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Two ways to use a quantized model&quot;,&quot;local&quot;:&quot;two-ways-to-use-a-quantized-model&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Quantize a model while loading&quot;,&quot;local&quot;:&quot;quantize-a-model-while-loading&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Load a pre-quantized checkpoint&quot;,&quot;local&quot;:&quot;load-a-pre-quantized-checkpoint&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;GGUF checkpoints&quot;,&quot;local&quot;:&quot;gguf-checkpoints&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Save a quantized model&quot;,&quot;local&quot;:&quot;save-a-quantized-model&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Choose a method&quot;,&quot;local&quot;:&quot;choose-a-method&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Other tools&quot;,&quot;local&quot;:&quot;other-tools&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Resources&quot;,&quot;local&quot;:&quot;resources&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Next steps&quot;,&quot;local&quot;:&quot;next-steps&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/><!---->
		<link href="/docs/transformers/pr_45649/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: inline-end; margin-inline-start: 10px; display: inline-flex; position: relative; z-index: 10;"><div dir="ltr" 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 dir="auto">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="overview" 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="#overview"><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><!---->Overview<!----></span></h1><!--]--><!----> <p>Quantization stores a model’s weights, and sometimes its activations, in fewer bits so the model needs less memory. Models are usually loaded in bf16, which takes 2 bytes per parameter. An 8B model needs about 16GB for its weights in bf16, about 8GB at 8 bits, and about 4–5GB at 4 bits. Fewer bits cost some accuracy, and each quantization method makes a different tradeoff between memory, speed, and accuracy.</p> <p>Transformers integrates more than 20 quantization methods through <a href="/docs/transformers/pr_45649/en/main_classes/model#transformers.PreTrainedModel.from_pretrained">from_pretrained()</a>, so you use the same loading code whichever method you choose.</p> <!--[1--><h2 class="relative group"><a id="two-ways-to-use-a-quantized-model" 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="#two-ways-to-use-a-quantized-model"><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><!---->Two ways to use a quantized model<!----></span></h2><!--]--><!----> <p>How you use quantization depends on whether your checkpoint is already quantized.</p> <table><thead><tr><th>You have</th><th>What you do</th><th>Methods</th></tr></thead><tbody><tr><td>A full-precision checkpoint</td><td>Pass a quantization config to <code>from_pretrained</code>, and the weights are quantized as they load.</td><td><a href="#quantize-a-model-while-loading">Quantize a model while loading</a></td></tr><tr><td>A checkpoint that’s already quantized</td><td>Call <code>from_pretrained</code> without a quantization config. Transformers reads the method from the checkpoint’s <code>config.json</code>, or from the GGUF file you name.</td><td><a href="#load-a-pre-quantized-checkpoint">Load a pre-quantized checkpoint</a></td></tr></tbody></table> <p>Some methods, such as AWQ, AutoRound, AQLM, compressed-tensors, and GGUF, can only be loaded from a checkpoint. You quantize the model with the method’s own library first, and then load the result in Transformers.</p> <!--[1--><h2 class="relative group"><a id="quantize-a-model-while-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="#quantize-a-model-while-loading"><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><!---->Quantize a model while loading<!----></span></h2><!--]--><!----> <p>Create the method’s config class and pass it to <code>quantization_config</code>. The example below loads an 8B model in 4 bits with <a href="./bitsandbytes">bitsandbytes</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 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-py "><!----><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModelForCausalLM, BitsAndBytesConfig

quantization_config = BitsAndBytesConfig(
    load_in_4bit=<span class="hljs-literal">True</span>,
    bnb_4bit_compute_dtype=torch.bfloat16,
)
model = AutoModelForCausalLM.from_pretrained(
    <span class="hljs-string">&quot;meta-llama/Llama-3.1-8B-Instruct&quot;</span>,
    quantization_config=quantization_config,
    device_map=<span class="hljs-string">&quot;auto&quot;</span>,
)<!----></pre></div><!----> <p>Every method in this group has its own config class, like <a href="/docs/transformers/pr_45649/en/main_classes/quantization#transformers.TorchAoConfig">TorchAoConfig</a> or <a href="/docs/transformers/pr_45649/en/main_classes/quantization#transformers.BitsAndBytesConfig">BitsAndBytesConfig</a>. Most of them don’t need a calibration dataset. <a href="./gptq">GPTQ</a> is the exception because it calibrates on a dataset while the model loads.</p> <!--[1--><h2 class="relative group"><a id="load-a-pre-quantized-checkpoint" 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="#load-a-pre-quantized-checkpoint"><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><!---->Load a pre-quantized checkpoint<!----></span></h2><!--]--><!----> <p>A pre-quantized checkpoint stores a <code>quantization_config</code> in its <code>config.json</code>. When <code>from_pretrained</code> finds it, the matching quantizer is selected automatically.</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-py "><!----><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained(
    <span class="hljs-string">&quot;TheBloke/zephyr-7B-alpha-AWQ&quot;</span>,
    device_map=<span class="hljs-string">&quot;auto&quot;</span>,
)
<span class="hljs-built_in">print</span>(model.config.quantization_config.quant_method)<!----></pre></div><!----> <p>Install the library that the method depends on before loading the checkpoint. Each method page lists what to install. To change loading options, like which kernel backend runs, pass a config of the same type. Its options are merged with the ones stored in the checkpoint.</p> <!--[2--><h3 class="relative group"><a id="gguf-checkpoints" 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="#gguf-checkpoints"><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><!---->GGUF checkpoints<!----></span></h3><!--]--><!----> <p>A GGUF checkpoint is a single file that holds the weights, config, and tokenizer, and it has no <code>config.json</code> for <code>from_pretrained</code> to read. Name the file with <code>gguf_file</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-py "><!----><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained(
    <span class="hljs-string">&quot;unsloth/Qwen3.5-4B-GGUF&quot;</span>,
    gguf_file=<span class="hljs-string">&quot;Qwen3.5-4B-Q4_K_M.gguf&quot;</span>,
)<!----></pre></div><!----> <p>Most GGUF models are dequantized to full precision as they load, so they use as much memory as an unquantized model. Weights stay packed only on MPS with the <a href="https://huggingface.co/docs/kernels/index" rel="nofollow">kernels</a> package installed, for Qwen3.5 and Qwen3.5 MoE. See <a href="./gguf">GGUF</a> for details.</p> <!--[1--><h2 class="relative group"><a id="save-a-quantized-model" 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="#save-a-quantized-model"><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><!---->Save a quantized model<!----></span></h2><!--]--><!----> <p>Call <a href="/docs/transformers/pr_45649/en/main_classes/model#transformers.PreTrainedModel.save_pretrained">save_pretrained()</a> or <a href="/docs/transformers/pr_45649/en/main_classes/model#transformers.utils.PushToHubMixin.push_to_hub">push_to_hub()</a> to save the quantized weights and their <code>quantization_config</code>. Not every method can save its quantized weights.</p> <!--[1--><h2 class="relative group"><a id="choose-a-method" 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="#choose-a-method"><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><!---->Choose a method<!----></span></h2><!--]--><!----> <p><a href="./selecting">Choose a method</a> recommends a starting point for common goals and compares every method by hardware, bit width, saving, fine-tuning, and <code>torch.compile</code> support. Use the <a href="https://huggingface.co/spaces/stevhliu/quantization-picker" rel="nofollow">quantization picker</a> to check whether a model fits in your hardware’s memory at each precision.</p> <!--[1--><h2 class="relative group"><a id="other-tools" 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="#other-tools"><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><!---->Other tools<!----></span></h2><!--]--><!----> <p>These tools quantize models outside Transformers.</p> <ul><li><a href="https://huggingface.co/docs/optimum/index" rel="nofollow">Optimum</a> quantizes models for specific hardware and runtimes, like Intel CPUs and ONNX Runtime.</li> <li>Hub Spaces quantize a model and push the result to your account: <a href="https://huggingface.co/spaces/bnb-community/bnb-my-repo" rel="nofollow">bitsandbytes</a>, <a href="https://huggingface.co/spaces/ggml-org/gguf-my-repo" rel="nofollow">GGUF</a>, and <a href="https://huggingface.co/spaces/mlx-community/mlx-my-repo" rel="nofollow">MLX</a>.</li></ul> <!--[1--><h2 class="relative group"><a id="resources" 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="#resources"><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><!---->Resources<!----></span></h2><!--]--><!----> <ul><li><a href="https://www.deeplearning.ai/short-courses/quantization-fundamentals-with-hugging-face/" rel="nofollow">Quantization Fundamentals with Hugging Face</a> course</li> <li><a href="https://www.deeplearning.ai/short-courses/quantization-in-depth" rel="nofollow">Quantization in Depth</a> course</li></ul> <!--[1--><h2 class="relative group"><a id="next-steps" 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="#next-steps"><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><!---->Next steps<!----></span></h2><!--]--><!----> <ul><li>Read <a href="./concept_guide">Quantization concepts</a> for how quantization schemes, granularity, and techniques work.</li> <li>Quantize the KV cache to save memory during long generations with the <a href="../kv_cache#quantized-cache">quantized cache</a>.</li> <li>Quantize a model while exporting it with <a href="../exporters_quantization">Quantizing exports</a>.</li> <li>Contribute a new method with <a href="./contribute">Contribute a quantization method</a>.</li></ul> <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/quantization/overview.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_j5svul = {
						base: "/docs/transformers/pr_45649/en",
						assets: "/docs/transformers/pr_45649/en"
					};
					const element = document.currentScript.parentElement;
					Promise.all([
						import("/docs/transformers/pr_45649/en/_app/immutable/entry/start.Cf6g_lQr.js"),
						import("/docs/transformers/pr_45649/en/_app/immutable/entry/app.Ckn5M_jH.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 713],
							data: [null,null],
							form: null,
							error: null
						});
					});
				}
			</script>
		
Xet Storage Details

Size:: 29.2 kB
Xet hash:: 4abef1d313fc2e4beacc417ede5d2ac744ba0d9d6f5c72e12e4fdde61d2c87cb
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.