2.0/tensor_attributes.html



<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
  <meta name="robots" content="noindex">
  <meta charset="utf-8">
  <meta name="generator" content="Docutils 0.18.1: http://docutils.sourceforge.net/" />

  <meta name="viewport" content="width=device-width, initial-scale=1.0">
  
  <title>Tensor Attributes &mdash; PyTorch 2.0 documentation</title>
  

    <link rel="canonical" href="https://pytorch.org/docs/stable/tensor_attributes.html"/>
  

  <link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
  <!-- <link rel="stylesheet" href="_static/pygments.css" type="text/css" /> -->
  <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
  <link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
  <link rel="stylesheet" href="_static/copybutton.css" type="text/css" />
  <link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.10.0-beta/dist/katex.min.css" type="text/css" />
  <link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.13.11/dist/katex.min.css" type="text/css" />
  <link rel="stylesheet" href="_static/katex-math.css" type="text/css" />
  <link rel="stylesheet" href="_static/sphinx-dropdown.css" type="text/css" />
  <link rel="stylesheet" href="_static/panels-bootstrap.min.css" type="text/css" />
  <link rel="stylesheet" href="_static/css/jit.css" type="text/css" />
    <link rel="index" title="Index" href="genindex.html" />
    <link rel="search" title="Search" href="search.html" />
    <link rel="next" title="Tensor Views" href="tensor_view.html" />
    <link rel="prev" title="torch.Tensor.zero_" href="generated/torch.Tensor.zero_.html" />


  <!-- Google Analytics -->
  
    <script async src="https://www.googletagmanager.com/gtag/js?id=UA-117752657-2"></script>
    <script>
      window.dataLayer = window.dataLayer || [];
      function gtag(){dataLayer.push(arguments);}
      gtag('js', new Date());

      gtag('config', 'UA-117752657-2');
    </script>
  
  <!-- End Google Analytics -->
  

  <script src="_static/js/modernizr.min.js"></script>

  <!-- Preload the theme fonts -->

<link rel="preload" href="_static/fonts/FreightSans/freight-sans-book.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/FreightSans/freight-sans-medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/IBMPlexMono/IBMPlexMono-Medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/FreightSans/freight-sans-bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/FreightSans/freight-sans-medium-italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/IBMPlexMono/IBMPlexMono-SemiBold.woff2" as="font" type="font/woff2" crossorigin="anonymous">

<!-- Preload the katex fonts -->

<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Math-Italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Main-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Main-Bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size1-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size4-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size2-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size3-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Caligraphic-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
  <link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.15.2/css/all.css" integrity="sha384-vSIIfh2YWi9wW0r9iZe7RJPrKwp6bG+s9QZMoITbCckVJqGCCRhc+ccxNcdpHuYu" crossorigin="anonymous">
</head>

<div class="container-fluid header-holder tutorials-header" id="header-holder">
  <div class="container">
    <div class="header-container">
      <a class="header-logo" href="https://pytorch.org/" aria-label="PyTorch"></a>

      <div class="main-menu">
        <ul>
          <li>
            <a href="https://pytorch.org/get-started">Get Started</a>
          </li>

          <li>
            <a href="https://pytorch.org/ecosystem">Ecosystem</a>
          </li>

          <li>
            <a href="https://pytorch.org/mobile">Mobile</a>
          </li>

          <li>
            <a href="https://pytorch.org/blog/">Blog</a>
          </li>

          <li>
            <a href="https://pytorch.org/tutorials">Tutorials</a>
          </li>

          <li class="active docs-active">
            <div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
              <a class="resource-option with-down-orange-arrow">
                Docs
              </a>
              <div class="resources-dropdown-menu">
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/docs/stable/index.html">
                  <span class="dropdown-title">PyTorch</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/audio/stable/index.html">
                  <span class="dropdown-title">torchaudio</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/text/stable/index.html">
                  <span class="dropdown-title">torchtext</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/vision/stable/index.html">
                  <span class="dropdown-title">torchvision</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/torcharrow">
                  <span class="dropdown-title">torcharrow</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/data">
                  <span class="dropdown-title">TorchData</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/torchrec">
                  <span class="dropdown-title">TorchRec</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/serve/">
                  <span class="dropdown-title">TorchServe</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/torchx/">
                  <span class="dropdown-title">TorchX</span>
                  <p></p>
                </a>
                <a class="doc-dropdown-option nav-dropdown-item" href="https://pytorch.org/xla">
                  <span class="dropdown-title">PyTorch on XLA Devices</span>
                  <p></p>
                </a>
            </div>
          </li>

          <li>
            <div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
              <a class="resource-option with-down-arrow">
                Resources
              </a>
              <div class="resources-dropdown-menu">
                <a class="nav-dropdown-item" href="https://pytorch.org/features">
                  <span class="dropdown-title">About</span>
                  <p>Learn about PyTorch’s features and capabilities</p>
                </a>
                <a class="nav-dropdown-item" href="https://pytorch.org/foundation">
                  <span class="dropdown-title">PyTorch Foundation</span>
                  <p>Learn about the PyTorch foundation</p>
                </a>
                <a class="nav-dropdown-item" href="https://pytorch.org/#community-module">
                  <span class="dropdown-title">Community</span>
                  <p>Join the PyTorch developer community to contribute, learn, and get your questions answered.</p>
                </a>
                <a class="nav-dropdown-item" href="https://pytorch.org/community-stories">
                  <span class="dropdown-title">Community Stories</span>
                  <p>Learn how our community solves real, everyday machine learning problems with PyTorch.</p>
                </a>
                <a class="nav-dropdown-item" href="https://pytorch.org/resources">
                  <span class="dropdown-title">Developer Resources</span>
                  <p>Find resources and get questions answered</p>
                </a>
                <a class="nav-dropdown-item" href="https://pytorch.org/events">
                  <span class="dropdown-title">Events</span>
                  <p>Find events, webinars, and podcasts</p>
                </a>
                <a class="nav-dropdown-item" href="https://discuss.pytorch.org/" target="_blank">
                  <span class="dropdown-title">Forums</span>
                  <p>A place to discuss PyTorch code, issues, install, research</p>
                </a>
                <a class="nav-dropdown-item" href="https://pytorch.org/hub">
                  <span class="dropdown-title">Models (Beta)</span>
                  <p>Discover, publish, and reuse pre-trained models</p>
                </a>
              </div>
            </div>
          </li>

          <li>
            <a href="https://github.com/pytorch/pytorch">GitHub</a>
          </li>
        </ul>
      </div>

      <a class="main-menu-open-button" href="#" data-behavior="open-mobile-menu"></a>
    </div>
  </div>
</div>

<body class="pytorch-body">

   
    <div class="table-of-contents-link-wrapper">
      <span>Table of Contents</span>
      <a href="#" class="toggle-table-of-contents" data-behavior="toggle-table-of-contents"></a>
    </div>

    <nav data-toggle="wy-nav-shift" class="pytorch-left-menu" id="pytorch-left-menu">
      <div class="pytorch-side-scroll">
        <div class="pytorch-menu pytorch-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
          <div class="pytorch-left-menu-search">
            
    <div class="version">
      <a href='https://pytorch.org/docs/versions.html'>2.0 &#x25BC</a>
    </div>
    

<div role="search">
  <form id="rtd-search-form" class="wy-form" action="search.html" method="get">
    <input type="text" name="q" placeholder="Search Docs" />
    <input type="hidden" name="check_keywords" value="yes" />
    <input type="hidden" name="area" value="default" />
  </form>
</div>

          </div>

          
              <p class="caption" role="heading"><span class="caption-text">Community</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="community/build_ci_governance.html">PyTorch Governance | Build + CI</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/contribution_guide.html">PyTorch Contribution Guide</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/design.html">PyTorch Design Philosophy</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/governance.html">PyTorch Governance | Mechanics</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/persons_of_interest.html">PyTorch Governance | Maintainers</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Developer Notes</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="notes/amp_examples.html">CUDA Automatic Mixed Precision examples</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/autograd.html">Autograd mechanics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/broadcasting.html">Broadcasting semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/cpu_threading_torchscript_inference.html">CPU threading and TorchScript inference</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/cuda.html">CUDA semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/ddp.html">Distributed Data Parallel</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/extending.html">Extending PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/extending.func.html">Extending torch.func with autograd.Function</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/faq.html">Frequently Asked Questions</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/gradcheck.html">Gradcheck mechanics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/hip.html">HIP (ROCm) semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/large_scale_deployments.html">Features for large-scale deployments</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/modules.html">Modules</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/mps.html">MPS backend</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/multiprocessing.html">Multiprocessing best practices</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/numerical_accuracy.html">Numerical accuracy</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/randomness.html">Reproducibility</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/serialization.html">Serialization semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/windows.html">Windows FAQ</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">torch.compile</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="dynamo/index.html">TorchDynamo Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamo/installation.html">Installing TorchDynamo</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamo/get-started.html">Getting Started</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamo/guards-overview.html">Guards Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamo/custom-backends.html">Custom Backends</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamo/deep-dive.html">TorchDynamo Deeper Dive</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamo/troubleshooting.html">TorchDynamo Troubleshooting</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamo/faq.html">Frequently Asked Questions</a></li>
<li class="toctree-l1"><a class="reference internal" href="ir.html">IRs</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Language Bindings</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="cpp_index.html">C++</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/javadoc/">Javadoc</a></li>
<li class="toctree-l1"><a class="reference internal" href="deploy.html">torch::deploy</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Python API</span></p>
<ul class="current">
<li class="toctree-l1"><a class="reference internal" href="torch.html">torch</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.html">torch.nn</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.functional.html">torch.nn.functional</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensors.html">torch.Tensor</a></li>
<li class="toctree-l1 current"><a class="current reference internal" href="#">Tensor Attributes</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensor_view.html">Tensor Views</a></li>
<li class="toctree-l1"><a class="reference internal" href="amp.html">torch.amp</a></li>
<li class="toctree-l1"><a class="reference internal" href="autograd.html">torch.autograd</a></li>
<li class="toctree-l1"><a class="reference internal" href="library.html">torch.library</a></li>
<li class="toctree-l1"><a class="reference internal" href="cuda.html">torch.cuda</a></li>
<li class="toctree-l1"><a class="reference internal" href="mps.html">torch.mps</a></li>
<li class="toctree-l1"><a class="reference internal" href="backends.html">torch.backends</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.html">torch.distributed</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.algorithms.join.html">torch.distributed.algorithms.join</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.elastic.html">torch.distributed.elastic</a></li>
<li class="toctree-l1"><a class="reference internal" href="fsdp.html">torch.distributed.fsdp</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.optim.html">torch.distributed.optim</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.tensor.parallel.html">torch.distributed.tensor.parallel</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.checkpoint.html">torch.distributed.checkpoint</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributions.html">torch.distributions</a></li>
<li class="toctree-l1"><a class="reference internal" href="_dynamo.html">torch._dynamo</a></li>
<li class="toctree-l1"><a class="reference internal" href="fft.html">torch.fft</a></li>
<li class="toctree-l1"><a class="reference internal" href="func.html">torch.func</a></li>
<li class="toctree-l1"><a class="reference internal" href="futures.html">torch.futures</a></li>
<li class="toctree-l1"><a class="reference internal" href="fx.html">torch.fx</a></li>
<li class="toctree-l1"><a class="reference internal" href="hub.html">torch.hub</a></li>
<li class="toctree-l1"><a class="reference internal" href="jit.html">torch.jit</a></li>
<li class="toctree-l1"><a class="reference internal" href="linalg.html">torch.linalg</a></li>
<li class="toctree-l1"><a class="reference internal" href="monitor.html">torch.monitor</a></li>
<li class="toctree-l1"><a class="reference internal" href="signal.html">torch.signal</a></li>
<li class="toctree-l1"><a class="reference internal" href="special.html">torch.special</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch.overrides.html">torch.overrides</a></li>
<li class="toctree-l1"><a class="reference internal" href="package.html">torch.package</a></li>
<li class="toctree-l1"><a class="reference internal" href="profiler.html">torch.profiler</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.init.html">torch.nn.init</a></li>
<li class="toctree-l1"><a class="reference internal" href="onnx.html">torch.onnx</a></li>
<li class="toctree-l1"><a class="reference internal" href="onnx_diagnostics.html">torch.onnx diagnostics</a></li>
<li class="toctree-l1"><a class="reference internal" href="optim.html">torch.optim</a></li>
<li class="toctree-l1"><a class="reference internal" href="complex_numbers.html">Complex Numbers</a></li>
<li class="toctree-l1"><a class="reference internal" href="ddp_comm_hooks.html">DDP Communication Hooks</a></li>
<li class="toctree-l1"><a class="reference internal" href="pipeline.html">Pipeline Parallelism</a></li>
<li class="toctree-l1"><a class="reference internal" href="quantization.html">Quantization</a></li>
<li class="toctree-l1"><a class="reference internal" href="rpc.html">Distributed RPC Framework</a></li>
<li class="toctree-l1"><a class="reference internal" href="random.html">torch.random</a></li>
<li class="toctree-l1"><a class="reference internal" href="masked.html">torch.masked</a></li>
<li class="toctree-l1"><a class="reference internal" href="nested.html">torch.nested</a></li>
<li class="toctree-l1"><a class="reference internal" href="sparse.html">torch.sparse</a></li>
<li class="toctree-l1"><a class="reference internal" href="storage.html">torch.Storage</a></li>
<li class="toctree-l1"><a class="reference internal" href="testing.html">torch.testing</a></li>
<li class="toctree-l1"><a class="reference internal" href="benchmark_utils.html">torch.utils.benchmark</a></li>
<li class="toctree-l1"><a class="reference internal" href="bottleneck.html">torch.utils.bottleneck</a></li>
<li class="toctree-l1"><a class="reference internal" href="checkpoint.html">torch.utils.checkpoint</a></li>
<li class="toctree-l1"><a class="reference internal" href="cpp_extension.html">torch.utils.cpp_extension</a></li>
<li class="toctree-l1"><a class="reference internal" href="data.html">torch.utils.data</a></li>
<li class="toctree-l1"><a class="reference internal" href="jit_utils.html">torch.utils.jit</a></li>
<li class="toctree-l1"><a class="reference internal" href="dlpack.html">torch.utils.dlpack</a></li>
<li class="toctree-l1"><a class="reference internal" href="mobile_optimizer.html">torch.utils.mobile_optimizer</a></li>
<li class="toctree-l1"><a class="reference internal" href="model_zoo.html">torch.utils.model_zoo</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensorboard.html">torch.utils.tensorboard</a></li>
<li class="toctree-l1"><a class="reference internal" href="type_info.html">Type Info</a></li>
<li class="toctree-l1"><a class="reference internal" href="named_tensor.html">Named Tensors</a></li>
<li class="toctree-l1"><a class="reference internal" href="name_inference.html">Named Tensors operator coverage</a></li>
<li class="toctree-l1"><a class="reference internal" href="config_mod.html">torch.__config__</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Libraries</span></p>
<ul>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/audio/stable">torchaudio</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/data">TorchData</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/torchrec">TorchRec</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/serve">TorchServe</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/text/stable">torchtext</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/vision/stable">torchvision</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/xla/">PyTorch on XLA Devices</a></li>
</ul>

            
        </div>
      </div>
    </nav>

    <div class="pytorch-container">
      <div class="pytorch-page-level-bar" id="pytorch-page-level-bar">
        <div class="pytorch-breadcrumbs-wrapper">
          

<div role="navigation" aria-label="breadcrumbs navigation">

  <ul class="pytorch-breadcrumbs">
    
      <li>
        <a href="index.html">
          
            Docs
          
        </a> &gt;
      </li>

        
      <li>Tensor Attributes</li>
    
    
      <li class="pytorch-breadcrumbs-aside">
        
            
            <a href="_sources/tensor_attributes.rst.txt" rel="nofollow"><img src="_static/images/view-page-source-icon.svg"></a>
          
        
      </li>
    
  </ul>

  
</div>
        </div>

        <div class="pytorch-shortcuts-wrapper" id="pytorch-shortcuts-wrapper">
          Shortcuts
        </div>
      </div>

      <section data-toggle="wy-nav-shift" id="pytorch-content-wrap" class="pytorch-content-wrap">
        <div class="pytorch-content-left">

        
          <div class="rst-content">
          
            <div role="main" class="main-content" itemscope="itemscope" itemtype="http://schema.org/Article">
             <article itemprop="articleBody" id="pytorch-article" class="pytorch-article">
              
  <section id="tensor-attributes">
<span id="tensor-attributes-doc"></span><h1>Tensor Attributes<a class="headerlink" href="#tensor-attributes" title="Permalink to this heading">¶</a></h1>
<p>Each <code class="docutils literal notranslate"><span class="pre">torch.Tensor</span></code> has a <a class="reference internal" href="#torch.dtype" title="torch.dtype"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.dtype</span></code></a>, <a class="reference internal" href="#torch.device" title="torch.device"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.device</span></code></a>, and <a class="reference internal" href="#torch.layout" title="torch.layout"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.layout</span></code></a>.</p>
<section id="torch-dtype">
<span id="dtype-doc"></span><h2>torch.dtype<a class="headerlink" href="#torch-dtype" title="Permalink to this heading">¶</a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="torch.dtype">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">torch.</span></span><span class="sig-name descname"><span class="pre">dtype</span></span><a class="headerlink" href="#torch.dtype" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>

<p>A <a class="reference internal" href="#torch.dtype" title="torch.dtype"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.dtype</span></code></a> is an object that represents the data type of a
<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.Tensor</span></code></a>. PyTorch has twelve different data types:</p>
<table class="docutils align-default">
<thead>
<tr class="row-odd"><th class="head"><p>Data type</p></th>
<th class="head"><p>dtype</p></th>
<th class="head"><p>Legacy Constructors</p></th>
</tr>
</thead>
<tbody>
<tr class="row-even"><td><p>32-bit floating point</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.float32</span></code> or <code class="docutils literal notranslate"><span class="pre">torch.float</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.FloatTensor</span></code></p></td>
</tr>
<tr class="row-odd"><td><p>64-bit floating point</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.float64</span></code> or <code class="docutils literal notranslate"><span class="pre">torch.double</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.DoubleTensor</span></code></p></td>
</tr>
<tr class="row-even"><td><p>64-bit complex</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.complex64</span></code> or <code class="docutils literal notranslate"><span class="pre">torch.cfloat</span></code></p></td>
<td></td>
</tr>
<tr class="row-odd"><td><p>128-bit complex</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.complex128</span></code> or <code class="docutils literal notranslate"><span class="pre">torch.cdouble</span></code></p></td>
<td></td>
</tr>
<tr class="row-even"><td><p>16-bit floating point <a class="footnote-reference brackets" href="#id3" id="id1" role="doc-noteref"><span class="fn-bracket">[</span>1<span class="fn-bracket">]</span></a></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.float16</span></code> or <code class="docutils literal notranslate"><span class="pre">torch.half</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.HalfTensor</span></code></p></td>
</tr>
<tr class="row-odd"><td><p>16-bit floating point <a class="footnote-reference brackets" href="#id4" id="id2" role="doc-noteref"><span class="fn-bracket">[</span>2<span class="fn-bracket">]</span></a></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.bfloat16</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.BFloat16Tensor</span></code></p></td>
</tr>
<tr class="row-even"><td><p>8-bit integer (unsigned)</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.uint8</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.ByteTensor</span></code></p></td>
</tr>
<tr class="row-odd"><td><p>8-bit integer (signed)</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.int8</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.CharTensor</span></code></p></td>
</tr>
<tr class="row-even"><td><p>16-bit integer (signed)</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.int16</span></code> or <code class="docutils literal notranslate"><span class="pre">torch.short</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.ShortTensor</span></code></p></td>
</tr>
<tr class="row-odd"><td><p>32-bit integer (signed)</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.int32</span></code> or <code class="docutils literal notranslate"><span class="pre">torch.int</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.IntTensor</span></code></p></td>
</tr>
<tr class="row-even"><td><p>64-bit integer (signed)</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.int64</span></code> or <code class="docutils literal notranslate"><span class="pre">torch.long</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.LongTensor</span></code></p></td>
</tr>
<tr class="row-odd"><td><p>Boolean</p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.bool</span></code></p></td>
<td><p><code class="docutils literal notranslate"><span class="pre">torch.*.BoolTensor</span></code></p></td>
</tr>
</tbody>
</table>
<aside class="footnote brackets" id="id3" role="note">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id1">1</a><span class="fn-bracket">]</span></span>
<p>Sometimes referred to as binary16: uses 1 sign, 5 exponent, and 10
significand bits. Useful when precision is important.</p>
</aside>
<aside class="footnote brackets" id="id4" role="note">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id2">2</a><span class="fn-bracket">]</span></span>
<p>Sometimes referred to as Brain Floating Point: use 1 sign, 8 exponent and 7
significand bits. Useful when range is important, since it has the same
number of exponent bits as <code class="docutils literal notranslate"><span class="pre">float32</span></code></p>
</aside>
<p>To find out if a <a class="reference internal" href="#torch.dtype" title="torch.dtype"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.dtype</span></code></a> is a floating point data type, the property <a class="reference internal" href="generated/torch.is_floating_point.html#torch.is_floating_point" title="torch.is_floating_point"><code class="xref py py-attr docutils literal notranslate"><span class="pre">is_floating_point</span></code></a>
can be used, which returns <code class="docutils literal notranslate"><span class="pre">True</span></code> if the data type is a floating point data type.</p>
<p>To find out if a <a class="reference internal" href="#torch.dtype" title="torch.dtype"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.dtype</span></code></a> is a complex data type, the property <a class="reference internal" href="generated/torch.is_complex.html#torch.is_complex" title="torch.is_complex"><code class="xref py py-attr docutils literal notranslate"><span class="pre">is_complex</span></code></a>
can be used, which returns <code class="docutils literal notranslate"><span class="pre">True</span></code> if the data type is a complex data type.</p>
<p id="type-promotion-doc">When the dtypes of inputs to an arithmetic operation (<cite>add</cite>, <cite>sub</cite>, <cite>div</cite>, <cite>mul</cite>) differ, we promote
by finding the minimum dtype that satisfies the following rules:</p>
<ul class="simple">
<li><p>If the type of a scalar operand is of a higher category than tensor operands
(where complex &gt; floating &gt; integral &gt; boolean), we promote to a type with sufficient size to hold
all scalar operands of that category.</p></li>
<li><p>If a zero-dimension tensor operand has a higher category than dimensioned operands,
we promote to a type with sufficient size and category to hold all zero-dim tensor operands of
that category.</p></li>
<li><p>If there are no higher-category zero-dim operands, we promote to a type with sufficient size
and category to hold all dimensioned operands.</p></li>
</ul>
<p>A floating point scalar operand has dtype <cite>torch.get_default_dtype()</cite> and an integral
non-boolean scalar operand has dtype <cite>torch.int64</cite>. Unlike numpy, we do not inspect
values when determining the minimum <cite>dtypes</cite> of an operand.  Quantized and complex types
are not yet supported.</p>
<p>Promotion Examples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">float_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">float</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">double_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">double</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">complex_float_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">complex64</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">complex_double_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">complex128</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">int_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">int</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">long_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">long</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">uint_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">uint8</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">double_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">double</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">bool_tensor</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">bool</span><span class="p">)</span>
<span class="go"># zero-dim tensors</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">long_zerodim</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">long</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">int_zerodim</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">int</span><span class="p">)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="mi">5</span><span class="p">,</span> <span class="mi">5</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.int64</span>
<span class="go"># 5 is an int64, but does not have higher category than int_tensor so is not considered.</span>
<span class="gp">&gt;&gt;&gt; </span><span class="p">(</span><span class="n">int_tensor</span> <span class="o">+</span> <span class="mi">5</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.int32</span>
<span class="gp">&gt;&gt;&gt; </span><span class="p">(</span><span class="n">int_tensor</span> <span class="o">+</span> <span class="n">long_zerodim</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.int32</span>
<span class="gp">&gt;&gt;&gt; </span><span class="p">(</span><span class="n">long_tensor</span> <span class="o">+</span> <span class="n">int_tensor</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.int64</span>
<span class="gp">&gt;&gt;&gt; </span><span class="p">(</span><span class="n">bool_tensor</span> <span class="o">+</span> <span class="n">long_tensor</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.int64</span>
<span class="gp">&gt;&gt;&gt; </span><span class="p">(</span><span class="n">bool_tensor</span> <span class="o">+</span> <span class="n">uint_tensor</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.uint8</span>
<span class="gp">&gt;&gt;&gt; </span><span class="p">(</span><span class="n">float_tensor</span> <span class="o">+</span> <span class="n">double_tensor</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.float64</span>
<span class="gp">&gt;&gt;&gt; </span><span class="p">(</span><span class="n">complex_float_tensor</span> <span class="o">+</span> <span class="n">complex_double_tensor</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.complex128</span>
<span class="gp">&gt;&gt;&gt; </span><span class="p">(</span><span class="n">bool_tensor</span> <span class="o">+</span> <span class="n">int_tensor</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.int32</span>
<span class="go"># Since long is a different kind than float, result dtype only needs to be large enough</span>
<span class="go"># to hold the float.</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">long_tensor</span><span class="p">,</span> <span class="n">float_tensor</span><span class="p">)</span><span class="o">.</span><span class="n">dtype</span>
<span class="go">torch.float32</span>
</pre></div>
</div>
<dl class="simple">
<dt>When the output tensor of an arithmetic operation is specified, we allow casting to its <cite>dtype</cite> except that:</dt><dd><ul class="simple">
<li><p>An integral output tensor cannot accept a floating point tensor.</p></li>
<li><p>A boolean output tensor cannot accept a non-boolean tensor.</p></li>
<li><p>A non-complex output tensor cannot accept a complex tensor</p></li>
</ul>
</dd>
</dl>
<p>Casting Examples:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># allowed:</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">float_tensor</span> <span class="o">*=</span> <span class="n">float_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">float_tensor</span> <span class="o">*=</span> <span class="n">int_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">float_tensor</span> <span class="o">*=</span> <span class="n">uint_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">float_tensor</span> <span class="o">*=</span> <span class="n">bool_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">float_tensor</span> <span class="o">*=</span> <span class="n">double_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">int_tensor</span> <span class="o">*=</span> <span class="n">long_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">int_tensor</span> <span class="o">*=</span> <span class="n">uint_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">uint_tensor</span> <span class="o">*=</span> <span class="n">int_tensor</span>

<span class="c1"># disallowed (RuntimeError: result type can&#39;t be cast to the desired output type):</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">int_tensor</span> <span class="o">*=</span> <span class="n">float_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">bool_tensor</span> <span class="o">*=</span> <span class="n">int_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">bool_tensor</span> <span class="o">*=</span> <span class="n">uint_tensor</span>
<span class="o">&gt;&gt;&gt;</span> <span class="n">float_tensor</span> <span class="o">*=</span> <span class="n">complex_float_tensor</span>
</pre></div>
</div>
</section>
<section id="torch-device">
<span id="device-doc"></span><h2>torch.device<a class="headerlink" href="#torch-device" title="Permalink to this heading">¶</a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="torch.device">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">torch.</span></span><span class="sig-name descname"><span class="pre">device</span></span><a class="headerlink" href="#torch.device" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>

<p>A <a class="reference internal" href="#torch.device" title="torch.device"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.device</span></code></a> is an object representing the device on which a <a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.Tensor</span></code></a> is
or will be allocated.</p>
<p>The <a class="reference internal" href="#torch.device" title="torch.device"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.device</span></code></a> contains a device type (<code class="docutils literal notranslate"><span class="pre">'cpu'</span></code>, <code class="docutils literal notranslate"><span class="pre">'cuda'</span></code> or <code class="docutils literal notranslate"><span class="pre">'mps'</span></code>) and optional device
ordinal for the device type. If the device ordinal is not present, this object will always represent
the current device for the device type, even after <a class="reference internal" href="generated/torch.cuda.set_device.html#torch.cuda.set_device" title="torch.cuda.set_device"><code class="xref py py-func docutils literal notranslate"><span class="pre">torch.cuda.set_device()</span></code></a> is called; e.g.,
a <a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.Tensor</span></code></a> constructed with device <code class="docutils literal notranslate"><span class="pre">'cuda'</span></code> is equivalent to <code class="docutils literal notranslate"><span class="pre">'cuda:X'</span></code> where X is
the result of <a class="reference internal" href="generated/torch.cuda.current_device.html#torch.cuda.current_device" title="torch.cuda.current_device"><code class="xref py py-func docutils literal notranslate"><span class="pre">torch.cuda.current_device()</span></code></a>.</p>
<p>A <a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.Tensor</span></code></a>’s device can be accessed via the <a class="reference internal" href="generated/torch.Tensor.device.html#torch.Tensor.device" title="torch.Tensor.device"><code class="xref py py-attr docutils literal notranslate"><span class="pre">Tensor.device</span></code></a> property.</p>
<p>A <a class="reference internal" href="#torch.device" title="torch.device"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.device</span></code></a> can be constructed via a string or via a string and device ordinal</p>
<p>Via a string:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda:0&#39;</span><span class="p">)</span>
<span class="go">device(type=&#39;cuda&#39;, index=0)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cpu&#39;</span><span class="p">)</span>
<span class="go">device(type=&#39;cpu&#39;)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;mps&#39;</span><span class="p">)</span>
<span class="go">device(type=&#39;mps&#39;)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda&#39;</span><span class="p">)</span>  <span class="c1"># current cuda device</span>
<span class="go">device(type=&#39;cuda&#39;)</span>
</pre></div>
</div>
<p>Via a string and device ordinal:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span>
<span class="go">device(type=&#39;cuda&#39;, index=0)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;mps&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span>
<span class="go">device(type=&#39;mps&#39;, index=0)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cpu&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span>
<span class="go">device(type=&#39;cpu&#39;, index=0)</span>
</pre></div>
</div>
<p>The device object can also be used as a context manager to change the default
device tensors are allocated on:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="k">with</span> <span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda:1&#39;</span><span class="p">):</span>
<span class="gp">... </span>    <span class="n">r</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">r</span><span class="o">.</span><span class="n">device</span>
<span class="go">device(type=&#39;cuda&#39;, index=1)</span>
</pre></div>
</div>
<p>This context manager has no effect if a factory function is passed an explicit,
non-None device argument.  To globally change the default device, see also
<a class="reference internal" href="generated/torch.set_default_device.html#torch.set_default_device" title="torch.set_default_device"><code class="xref py py-func docutils literal notranslate"><span class="pre">torch.set_default_device()</span></code></a>.</p>
<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>This function imposes a slight performance cost on every Python
call to the torch API (not just factory functions).  If this
is causing problems for you, please comment on
<a class="reference external" href="https://github.com/pytorch/pytorch/issues/92701">https://github.com/pytorch/pytorch/issues/92701</a></p>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The <a class="reference internal" href="#torch.device" title="torch.device"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.device</span></code></a> argument in functions can generally be substituted with a string.
This allows for fast prototyping of code.</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="c1"># Example of a function that takes in a torch.device</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">cuda1</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda:1&#39;</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">((</span><span class="mi">2</span><span class="p">,</span><span class="mi">3</span><span class="p">),</span> <span class="n">device</span><span class="o">=</span><span class="n">cuda1</span><span class="p">)</span>
</pre></div>
</div>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="c1"># You can substitute the torch.device with a string</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">((</span><span class="mi">2</span><span class="p">,</span><span class="mi">3</span><span class="p">),</span> <span class="n">device</span><span class="o">=</span><span class="s1">&#39;cuda:1&#39;</span><span class="p">)</span>
</pre></div>
</div>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>For legacy reasons, a device can be constructed via a single device ordinal, which is treated
as a cuda device.  This matches <a class="reference internal" href="generated/torch.Tensor.get_device.html#torch.Tensor.get_device" title="torch.Tensor.get_device"><code class="xref py py-meth docutils literal notranslate"><span class="pre">Tensor.get_device()</span></code></a>, which returns an ordinal for cuda
tensors and is not supported for cpu tensors.</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="mi">1</span><span class="p">)</span>
<span class="go">device(type=&#39;cuda&#39;, index=1)</span>
</pre></div>
</div>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Methods which take a device will generally accept a (properly formatted) string
or (legacy) integer device ordinal, i.e. the following are all equivalent:</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">((</span><span class="mi">2</span><span class="p">,</span><span class="mi">3</span><span class="p">),</span> <span class="n">device</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda:1&#39;</span><span class="p">))</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">((</span><span class="mi">2</span><span class="p">,</span><span class="mi">3</span><span class="p">),</span> <span class="n">device</span><span class="o">=</span><span class="s1">&#39;cuda:1&#39;</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">((</span><span class="mi">2</span><span class="p">,</span><span class="mi">3</span><span class="p">),</span> <span class="n">device</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span>  <span class="c1"># legacy</span>
</pre></div>
</div>
</div>
</section>
<section id="torch-layout">
<span id="layout-doc"></span><h2>torch.layout<a class="headerlink" href="#torch-layout" title="Permalink to this heading">¶</a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="torch.layout">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">torch.</span></span><span class="sig-name descname"><span class="pre">layout</span></span><a class="headerlink" href="#torch.layout" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>

<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>The <code class="docutils literal notranslate"><span class="pre">torch.layout</span></code> class is in beta and subject to change.</p>
</div>
<p>A <a class="reference internal" href="#torch.layout" title="torch.layout"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.layout</span></code></a> is an object that represents the memory layout of a
<a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.Tensor</span></code></a>. Currently, we support <code class="docutils literal notranslate"><span class="pre">torch.strided</span></code> (dense Tensors)
and have beta support for <code class="docutils literal notranslate"><span class="pre">torch.sparse_coo</span></code> (sparse COO Tensors).</p>
<p><code class="docutils literal notranslate"><span class="pre">torch.strided</span></code> represents dense Tensors and is the memory layout that
is most commonly used. Each strided tensor has an associated
<code class="xref py py-class docutils literal notranslate"><span class="pre">torch.Storage</span></code>, which holds its data. These tensors provide
multi-dimensional, <a class="reference external" href="https://en.wikipedia.org/wiki/Stride_of_an_array">strided</a>
view of a storage. Strides are a list of integers: the k-th stride
represents the jump in the memory necessary to go from one element to the
next one in the k-th dimension of the Tensor. This concept makes it possible
to perform many tensor operations efficiently.</p>
<p>Example:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">x</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">([[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">],</span> <span class="p">[</span><span class="mi">6</span><span class="p">,</span> <span class="mi">7</span><span class="p">,</span> <span class="mi">8</span><span class="p">,</span> <span class="mi">9</span><span class="p">,</span> <span class="mi">10</span><span class="p">]])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">x</span><span class="o">.</span><span class="n">stride</span><span class="p">()</span>
<span class="go">(5, 1)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">x</span><span class="o">.</span><span class="n">t</span><span class="p">()</span><span class="o">.</span><span class="n">stride</span><span class="p">()</span>
<span class="go">(1, 5)</span>
</pre></div>
</div>
<p>For more information on <code class="docutils literal notranslate"><span class="pre">torch.sparse_coo</span></code> tensors, see <a class="reference internal" href="sparse.html#sparse-docs"><span class="std std-ref">torch.sparse</span></a>.</p>
</section>
<section id="torch-memory-format">
<h2>torch.memory_format<a class="headerlink" href="#torch-memory-format" title="Permalink to this heading">¶</a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="torch.memory_format">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">torch.</span></span><span class="sig-name descname"><span class="pre">memory_format</span></span><a class="headerlink" href="#torch.memory_format" title="Permalink to this definition">¶</a></dt>
<dd></dd></dl>

<p>A <a class="reference internal" href="#torch.memory_format" title="torch.memory_format"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.memory_format</span></code></a> is an object representing the memory format on which a <a class="reference internal" href="tensors.html#torch.Tensor" title="torch.Tensor"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.Tensor</span></code></a> is
or will be allocated.</p>
<p>Possible values are:</p>
<ul class="simple">
<li><p><code class="docutils literal notranslate"><span class="pre">torch.contiguous_format</span></code>:
Tensor is or will be allocated in dense non-overlapping memory. Strides represented by values in decreasing order.</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">torch.channels_last</span></code>:
Tensor is or will be allocated in dense non-overlapping memory. Strides represented by values in
<code class="docutils literal notranslate"><span class="pre">strides[0]</span> <span class="pre">&gt;</span> <span class="pre">strides[2]</span> <span class="pre">&gt;</span> <span class="pre">strides[3]</span> <span class="pre">&gt;</span> <span class="pre">strides[1]</span> <span class="pre">==</span> <span class="pre">1</span></code> aka NHWC order.</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">torch.channels_last_3d</span></code>:
Tensor is or will be allocated in dense non-overlapping memory. Strides represented by values in
<code class="docutils literal notranslate"><span class="pre">strides[0]</span> <span class="pre">&gt;</span> <span class="pre">strides[2]</span> <span class="pre">&gt;</span> <span class="pre">strides[3]</span> <span class="pre">&gt;</span> <span class="pre">strides[4]</span> <span class="pre">&gt;</span> <span class="pre">strides[1]</span> <span class="pre">==</span> <span class="pre">1</span></code> aka NDHWC order.</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">torch.preserve_format</span></code>:
Used in functions like <cite>clone</cite> to preserve the memory format of the input tensor. If input tensor is
allocated in dense non-overlapping memory, the output tensor strides will be copied from the input.
Otherwise output strides will follow <code class="docutils literal notranslate"><span class="pre">torch.contiguous_format</span></code></p></li>
</ul>
</section>
</section>


             </article>
             
            </div>
            <footer>
  
    <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
      
        <a href="tensor_view.html" class="btn btn-neutral float-right" title="Tensor Views" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
      
      
        <a href="generated/torch.Tensor.zero_.html" class="btn btn-neutral" title="torch.Tensor.zero_" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
      
    </div>
  

    <hr>

  
  <div role="contentinfo">
    <p>
        &copy; Copyright 2023, PyTorch Contributors.

    </p>
  </div>
    
      <div>
        Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
      </div>
     

</footer>

          </div>
<script>

var match = window.location.href.match(/\/_[a-zA-Z0-9_]*.html|_dynamo/gi);
var url = window.location.href.lastIndexOf(match[match.length-1]);

if (url)
  {
    var div = '<div class="admonition note"><p class="admonition-title">Note</p><p><i class="fa fa-exclamation-circle" aria-hidden="true">&nbsp</i> This page describes an internal API which is not intended to be used outside of the PyTorch codebase and can be modified or removed without notice.</p></div>'
    document.getElementById("pytorch-article").insertAdjacentHTML('afterBegin', div)
  }
</script>
        </div>

        <div class="pytorch-content-right" id="pytorch-content-right">
          <div class="pytorch-right-menu" id="pytorch-right-menu">
            <div class="pytorch-side-scroll" id="pytorch-side-scroll-right">
              <ul>
<li><a class="reference internal" href="#">Tensor Attributes</a><ul>
<li><a class="reference internal" href="#torch-dtype">torch.dtype</a></li>
<li><a class="reference internal" href="#torch-device">torch.device</a></li>
<li><a class="reference internal" href="#torch-layout">torch.layout</a></li>
<li><a class="reference internal" href="#torch-memory-format">torch.memory_format</a></li>
</ul>
</li>
</ul>

            </div>
          </div>
        </div>
      </section>
    </div>

  
       <script type="text/javascript" id="documentation_options" data-url_root="./" src="_static/documentation_options.js"></script>
         <script data-url_root="./" id="documentation_options" src="_static/documentation_options.js"></script>
         <script src="_static/jquery.js"></script>
         <script src="_static/underscore.js"></script>
         <script src="_static/_sphinx_javascript_frameworks_compat.js"></script>
         <script src="_static/doctools.js"></script>
         <script src="_static/clipboard.min.js"></script>
         <script src="_static/copybutton.js"></script>
     

  <script type="text/javascript" src="_static/js/vendor/popper.min.js"></script>
  <script type="text/javascript" src="_static/js/vendor/bootstrap.min.js"></script>
  <script src="https://cdnjs.cloudflare.com/ajax/libs/list.js/1.5.0/list.min.js"></script>
  <script type="text/javascript" src="_static/js/theme.js"></script>

  <script type="text/javascript">
      jQuery(function () {
          SphinxRtdTheme.Navigation.enable(true);
      });
  </script>
 
<script script type="text/javascript">
  var collapsedSections = ['Developer Notes', 'Language Bindings', 'Libraries', 'Community'];
</script>

<img height="1" width="1" style="border-style:none;" alt="" src="https://www.googleadservices.com/pagead/conversion/795629140/?label=txkmCPmdtosBENSssfsC&amp;guid=ON&amp;script=0"/>


  <!-- Begin Footer -->

  <div class="container-fluid docs-tutorials-resources" id="docs-tutorials-resources">
    <div class="container">
      <div class="row">
        <div class="col-md-4 text-center">
          <h2>Docs</h2>
          <p>Access comprehensive developer documentation for PyTorch</p>
          <a class="with-right-arrow" href="https://pytorch.org/docs/stable/index.html">View Docs</a>
        </div>

        <div class="col-md-4 text-center">
          <h2>Tutorials</h2>
          <p>Get in-depth tutorials for beginners and advanced developers</p>
          <a class="with-right-arrow" href="https://pytorch.org/tutorials">View Tutorials</a>
        </div>

        <div class="col-md-4 text-center">
          <h2>Resources</h2>
          <p>Find development resources and get your questions answered</p>
          <a class="with-right-arrow" href="https://pytorch.org/resources">View Resources</a>
        </div>
      </div>
    </div>
  </div>

  <footer class="site-footer">
    <div class="container footer-container">
      <div class="footer-logo-wrapper">
        <a href="https://pytorch.org/" class="footer-logo"></a>
      </div>

      <div class="footer-links-wrapper">
        <div class="footer-links-col">
          <ul>
            <li class="list-title"><a href="https://pytorch.org/">PyTorch</a></li>
            <li><a href="https://pytorch.org/get-started">Get Started</a></li>
            <li><a href="https://pytorch.org/features">Features</a></li>
            <li><a href="https://pytorch.org/ecosystem">Ecosystem</a></li>
            <li><a href="https://pytorch.org/blog/">Blog</a></li>
            <li><a href="https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md">Contributing</a></li>
          </ul>
        </div>

        <div class="footer-links-col">
          <ul>
            <li class="list-title"><a href="https://pytorch.org/resources">Resources</a></li>
            <li><a href="https://pytorch.org/tutorials">Tutorials</a></li>
            <li><a href="https://pytorch.org/docs/stable/index.html">Docs</a></li>
            <li><a href="https://discuss.pytorch.org" target="_blank">Discuss</a></li>
            <li><a href="https://github.com/pytorch/pytorch/issues" target="_blank">Github Issues</a></li>
            <li><a href="https://pytorch.org/assets/brand-guidelines/PyTorch-Brand-Guidelines.pdf" target="_blank">Brand Guidelines</a></li>
          </ul>
        </div>

        <div class="footer-links-col">
          <ul>
            <li class="list-title">Stay up to date</li>
            <li><a href="https://www.facebook.com/pytorch" target="_blank">Facebook</a></li>
            <li><a href="https://twitter.com/pytorch" target="_blank">Twitter</a></li>
            <li><a href="https://www.youtube.com/pytorch" target="_blank">YouTube</a></li>
            <li><a href="https://www.linkedin.com/company/pytorch" target="_blank">LinkedIn</a></li>
          </ul>  
          </div>

        <div class="footer-links-col">
          <ul>
            <li class="list-title">PyTorch Podcasts</li>
            <li><a href="https://open.spotify.com/show/6UzHKeiy368jKfQMKKvJY5" target="_blank">Spotify</a></li>
            <li><a href="https://podcasts.apple.com/us/podcast/pytorch-developer-podcast/id1566080008" target="_blank">Apple</a></li>
            <li><a href="https://www.google.com/podcasts?feed=aHR0cHM6Ly9mZWVkcy5zaW1wbGVjYXN0LmNvbS9PQjVGa0lsOA%3D%3D" target="_blank">Google</a></li>
            <li><a href="https://music.amazon.com/podcasts/7a4e6f0e-26c2-49e9-a478-41bd244197d0/PyTorch-Developer-Podcast?" target="_blank">Amazon</a></li>
          </ul>
         </div>
        </div>
        
        <div class="privacy-policy">
          <ul>
            <li class="privacy-policy-links"><a href="https://www.linuxfoundation.org/terms/" target="_blank">Terms</a></li>
            <li class="privacy-policy-links">|</li>
            <li class="privacy-policy-links"><a href="https://www.linuxfoundation.org/privacy-policy/" target="_blank">Privacy</a></li>
          </ul>
        </div>
        <div class="copyright">
        <p>© Copyright The Linux Foundation. The PyTorch Foundation is a project of The Linux Foundation.
          For web site terms of use, trademark policy and other policies applicable to The PyTorch Foundation please see
          <a href="www.linuxfoundation.org/policies/">www.linuxfoundation.org/policies/</a>. The PyTorch Foundation supports the PyTorch open source
          project, which has been established as PyTorch Project a Series of LF Projects, LLC. For policies applicable to the PyTorch Project a Series of LF Projects, LLC,
          please see <a href="www.lfprojects.org/policies/">www.lfprojects.org/policies/</a>.</p>
      </div>
     </div>

  </footer>

  <div class="cookie-banner-wrapper">
  <div class="container">
    <p class="gdpr-notice">To analyze traffic and optimize your experience, we serve cookies on this site. By clicking or navigating, you agree to allow our usage of cookies. As the current maintainers of this site, Facebook’s Cookies Policy applies. Learn more, including about available controls: <a href="https://www.facebook.com/policies/cookies/">Cookies Policy</a>.</p>
    <img class="close-button" src="_static/images/pytorch-x.svg">
  </div>
</div>

  <!-- End Footer -->

  <!-- Begin Mobile Menu -->

  <div class="mobile-main-menu">
    <div class="container-fluid">
      <div class="container">
        <div class="mobile-main-menu-header-container">
          <a class="header-logo" href="https://pytorch.org/" aria-label="PyTorch"></a>
          <a class="main-menu-close-button" href="#" data-behavior="close-mobile-menu"></a>
        </div>
      </div>
    </div>

    <div class="mobile-main-menu-links-container">
      <div class="main-menu">
        <ul>
          <li>
            <a href="https://pytorch.org/get-started">Get Started</a>
          </li>

          <li>
            <a href="https://pytorch.org/ecosystem">Ecosystem</a>
          </li>
            
          <li>
            <a href="https://pytorch.org/mobile">Mobile</a>
          </li>

          <li>
            <a href="https://pytorch.org/blog/">Blog</a>
          </li>

          <li>
            <a href="https://pytorch.org/tutorials">Tutorials</a>
          </li>

          <li class="resources-mobile-menu-title" class="active">
            Docs
          </li>

          <ul class="resources-mobile-menu-items">
            <li>
              <a href="https://pytorch.org/docs/stable/index.html">PyTorch</a>
            </li>

            <li>
              <a href="https://pytorch.org/audio/stable/index.html">torchaudio</a>
            </li>

            <li>
              <a href="https://pytorch.org/text/stable/index.html">torchtext</a>
            </li>

            <li>
              <a href="https://pytorch.org/vision/stable/index.html">torchvision</a>
            </li>

            <li>
              <a href="https://pytorch.org/torcharrow">torcharrow</a>
            </li>

            <li>
              <a href="https://pytorch.org/data">TorchData</a>
            </li>

            <li>
              <a href="https://pytorch.org/torchrec">TorchRec</a>
            </li>

            <li>
              <a href="https://pytorch.org/serve/">TorchServe</a>
            </li>

            <li>
              <a href="https://pytorch.org/torchx/">TorchX</a>
            </li>

            <li>
              <a href="https://pytorch.org/xla">PyTorch on XLA Devices</a>
            </li>
          </ul>

          <li class="resources-mobile-menu-title">
            Resources
          </li>
            
           <ul class="resources-mobile-menu-items">

            <li>
              <a href="https://pytorch.org/features">About</a>
            </li>

            <li>
              <a href="https://pytorch.org/foundation">PyTorch Foundation</a>
            </li>

            <li>
              <a href="https://pytorch.org/#community-module">Community</a>
            </li>

            <li>
              <a href="https://pytorch.org/community-stories">Community Stories</a>
            </li>

            <li>
              <a href="https://pytorch.org/resources">Developer Resources</a>
            </li>

            <li>
              <a href="https://pytorch.org/events">Events</a>
            </li>

            <li>
              <a href="https://discuss.pytorch.org/">Forums</a>
            </li>

            <li>
              <a href="https://pytorch.org/hub">Models (Beta)</a>
            </li>
          </ul>

          <li>
            <a href="https://github.com/pytorch/pytorch">Github</a>
          </li>
        </ul>
      </div>
    </div>
  </div>

  <!-- End Mobile Menu -->

  <script type="text/javascript" src="_static/js/vendor/anchor.min.js"></script>

  <script type="text/javascript">
    $(document).ready(function() {
      mobileMenu.bind();
      mobileTOC.bind();
      pytorchAnchors.bind();
      sideMenus.bind();
      scrollToAnchor.bind();
      highlightNavigation.bind();
      mainMenuDropdown.bind();
      filterTags.bind();

      // Add class to links that have code blocks, since we cannot create links in code blocks
      $("article.pytorch-article a span.pre").each(function(e) {
        $(this).closest("a").addClass("has-code");
      });
    })
  </script>
</body>
</html>