transformers/pr_48314/en/quantization/nvfp4.html

<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;NVFP4&quot;,&quot;local&quot;:&quot;nvfp4&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Current limitations&quot;,&quot;local&quot;:&quot;current-limitations&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
		<link href="/docs/transformers/main/en/_app/immutable/entry/start.BukNISTP.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/chunks/BPzh_Jza.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/chunks/BhmhGQv5.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/entry/app.5sK1nhM5.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/chunks/qMuWdE-Y.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/chunks/CfMEBQAN.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/chunks/B4BfNKKv.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/nodes/0.C2Q_m8zG.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/chunks/B76T2TVW.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/nodes/673.B_36FgLE.js" rel="modulepreload">
		<link href="/docs/transformers/main/en/_app/immutable/chunks/vlAHDplH.js" rel="modulepreload">
		<!--vhq1jl--><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;NVFP4&quot;,&quot;local&quot;:&quot;nvfp4&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Current limitations&quot;,&quot;local&quot;:&quot;current-limitations&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/><!---->
		<link href="/docs/transformers/main/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: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"><div 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>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="nvfp4" 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="#nvfp4"><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>NVFP4</span></h1><!--]--><!----> <p>NVFP4 quantization packs full-precision linear weights into NVIDIA’s 4-bit floating-point format while a model is
loaded. <a href="/docs/transformers/pr_48314/en/main_classes/quantization#transformers.NVFP4Config">NVFP4Config</a> replaces eligible bias-free <code>torch.nn.Linear</code> modules, whose <code>in_features</code> and <code>out_features</code> are both divisible by 16, with an NVFP4 linear implementation from
the <a href="https://huggingface.co/kernels-community/nvfp4-gemm" rel="nofollow">NVFP4 Hub kernel</a>. The model’s attention and MLP interfaces are
not replaced.</p> <blockquote class="tip"><p>NVFP4 requires a Blackwell GPU with compute capability 10.0 or newer, a compatible CUDA-enabled PyTorch build, and
the <a href="https://github.com/huggingface/kernels" rel="nofollow">kernels</a> package.</p></blockquote> <p>Install Accelerate and a compatible version of <code>kernels</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-bash "><!---->pip install --upgrade accelerate kernels<!----></pre></div><!----> <p>Pass <a href="/docs/transformers/pr_48314/en/main_classes/quantization#transformers.NVFP4Config">NVFP4Config</a> to <a href="/docs/transformers/pr_48314/en/main_classes/model#transformers.PreTrainedModel.from_pretrained">from_pretrained()</a> with a single CUDA device. Weights are quantized as they
are loaded, so the source checkpoint should contain floating-point weights.</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, AutoTokenizer, NVFP4Config


model_id = <span class="hljs-string">&quot;meta-llama/Llama-3.2-1B&quot;</span>
quantization_config = NVFP4Config()
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    dtype=torch.bfloat16,
    device_map=<span class="hljs-string">&quot;cuda&quot;</span>,
    quantization_config=quantization_config,
)

tokenizer = AutoTokenizer.from_pretrained(model_id)
inputs = tokenizer(<span class="hljs-string">&quot;NVFP4 is&quot;</span>, return_tensors=<span class="hljs-string">&quot;pt&quot;</span>).to(model.device)
output = model.generate(**inputs, max_new_tokens=<span class="hljs-number">20</span>)
<span class="hljs-built_in">print</span>(tokenizer.decode(output[<span class="hljs-number">0</span>], skip_special_tokens=<span class="hljs-literal">True</span>))<!----></pre></div><!----> <p>Use <code>modules_to_not_convert</code> to keep selected modules in their original precision.</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 "><!---->quantization_config = NVFP4Config(modules_to_not_convert=[<span class="hljs-string">&quot;vision&quot;</span>, <span class="hljs-string">&quot;lm_head&quot;</span>])<!----></pre></div><!----> <p>NVFP4 linear modules support <code>torch.compile</code>. The first compiled invocation includes graph compilation time, so warm up
the model before measuring generation throughput.</p> <!--[1--><h2 class="relative group"><a id="current-limitations" 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="#current-limitations"><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>Current limitations</span></h2><!--]--><!----> <ul><li>Only one CUDA device is supported. Tensor parallelism and multi-device <code>device_map</code> configurations are rejected until
the sharding behavior of the NVFP4 scale metadata is defined.</li> <li>CPU and disk offload are not supported.</li> <li>Pre-quantized NVFP4 checkpoints are not supported.</li> <li>NVFP4 models cannot currently be serialized with <a href="/docs/transformers/pr_48314/en/main_classes/model#transformers.PreTrainedModel.save_pretrained">save_pretrained()</a> or trained.</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/nvfp4.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_hvxp82 = {
						base: "/docs/transformers/main/en",
						assets: "/docs/transformers/main/en"
					};

					const element = document.currentScript.parentElement;

					Promise.all([
						import("/docs/transformers/main/en/_app/immutable/entry/start.BukNISTP.js"),
						import("/docs/transformers/main/en/_app/immutable/entry/app.5sK1nhM5.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 673],
							data: [null,null],
							form: null,
							error: null
						});
					});
				}
			</script>