Tensors and Dynamic neural networks in Python with strong GPU acceleration


PyTorch Logo

PyTorch is a Python package that provides two high-level features:

  • Tensor computation (like NumPy) with strong GPU acceleration
  • Deep neural networks built on a tape-based autograd system

You can reuse your favorite Python packages such as NumPy, SciPy, and Cython to extend PyTorch when needed.

System 3.6 3.7 3.8
Linux CPU Build Status Build Status
Linux GPU Build Status Build Status
Windows CPU / GPU Build Status
Linux (ppc64le) CPU Build Status
Linux (ppc64le) GPU Build Status
Linux (aarch64) CPU Build Status Build Status Build Status

See also the ci.pytorch.org HUD.

More About PyTorch

At a granular level, PyTorch is a library that consists of the following components:

Component Description
torch a Tensor library like NumPy, with strong GPU support
torch.autograd a tape-based automatic differentiation library that supports all differentiable Tensor operations in torch
torch.jit a compilation stack (TorchScript) to create serializable and optimizable models from PyTorch code
torch.nn a neural networks library deeply integrated with autograd designed for maximum flexibility
torch.multiprocessing Python multiprocessing, but with magical memory sharing of torch Tensors across processes. Useful for data loading and Hogwild training
torch.utils DataLoader and other utility functions for convenience

Usually, PyTorch is used either as:

  • A replacement for NumPy to use the power of GPUs.
  • A deep learning research platform that provides maximum flexibility and speed.

Elaborating Further:

A GPU-Ready Tensor Library

If you use NumPy, then you have used Tensors (a.k.a. ndarray).

Tensor illustration

PyTorch provides Tensors that can live either on the CPU or the GPU and accelerates the computation by a huge amount.

We provide a wide variety of tensor routines to accelerate and fit your scientific computation needs such as slicing, indexing, math operations, linear algebra, reductions. And they are fast!

Dynamic Neural Networks: Tape-Based Autograd

PyTorch has a unique way of building neural networks: using and replaying a tape recorder.

Most frameworks such as TensorFlow, Theano, Caffe, and CNTK have a static view of the world. One has to build a neural network and reuse the same structure again and again. Changing the way the network behaves means that one has to start from scratch.

With PyTorch, we use a technique called reverse-mode auto-differentiation, which allows you to change the way your network behaves arbitrarily with zero lag or overhead. Our inspiration comes from several research papers on this topic, as well as current and past work such as torch-autograd, autograd, Chainer, etc.

While this technique is not unique to PyTorch, it's one of the fastest implementations of it to date. You get the best of speed and flexibility for your crazy research.

Dynamic graph

Python First

PyTorch is not a Python binding into a monolithic C++ framework. It is built to be deeply integrated into Python. You can use it naturally like you would use NumPy / SciPy / scikit-learn etc. You can write your new neural network layers in Python itself, using your favorite libraries and use packages such as Cython and Numba. Our goal is to not reinvent the wheel where appropriate.

Imperative Experiences

PyTorch is designed to be intuitive, linear in thought, and easy to use. When you execute a line of code, it gets executed. There isn't an asynchronous view of the world. When you drop into a debugger or receive error messages and stack traces, understanding them is straightforward. The stack trace points to exactly where your code was defined. We hope you never spend hours debugging your code because of bad stack traces or asynchronous and opaque execution engines.

Fast and Lean

PyTorch has minimal framework overhead. We integrate acceleration libraries such as Intel MKL and NVIDIA (cuDNN, NCCL) to maximize speed. At the core, its CPU and GPU Tensor and neural network backends (TH, THC, THNN, THCUNN) are mature and have been tested for years.

Hence, PyTorch is quite fast – whether you run small or large neural networks.

The memory usage in PyTorch is extremely efficient compared to Torch or some of the alternatives. We've written custom memory allocators for the GPU to make sure that your deep learning models are maximally memory efficient. This enables you to train bigger deep learning models than before.

Extensions Without Pain

Writing new neural network modules, or interfacing with PyTorch's Tensor API was designed to be straightforward and with minimal abstractions.

You can write new neural network layers in Python using the torch API or your favorite NumPy-based libraries such as SciPy.

If you want to write your layers in C/C++, we provide a convenient extension API that is efficient and with minimal boilerplate. No wrapper code needs to be written. You can see a tutorial here and an example here.



Commands to install from binaries via Conda or pip wheels are on our website: https://pytorch.org

NVIDIA Jetson Platforms

Python wheels for NVIDIA's Jetson Nano, Jetson TX2, and Jetson AGX Xavier are available via the following URLs:

They require JetPack 4.2 and above, and @dusty-nv maintains them

From Source

If you are installing from source, you will need Python 3.6.2 or later and a C++14 compiler. Also, we highly recommend installing an Anaconda environment. You will get a high-quality BLAS library (MKL) and you get controlled dependency versions regardless of your Linux distro.

Once you have Anaconda installed, here are the instructions.

If you want to compile with CUDA support, install

If you want to disable CUDA support, export environment variable USE_CUDA=0. Other potentially useful environment variables may be found in setup.py.

If you are building for NVIDIA's Jetson platforms (Jetson Nano, TX1, TX2, AGX Xavier), Instructions to install PyTorch for Jetson Nano are available here

Install Dependencies


conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi typing_extensions future six requests dataclasses

On Linux

# Add LAPACK support for the GPU if needed
conda install -c pytorch magma-cuda110  # or the magma-cuda* that matches your CUDA version from https://anaconda.org/pytorch/repo

On MacOS

# Add these packages if torch.distributed is needed
conda install pkg-config libuv

On Windows

# Add these packages if torch.distributed is needed.
# Distributed package support on Windows is a prototype feature and is subject to changes.
conda install -c conda-forge libuv=1.39

Get the PyTorch Source

git clone --recursive https://github.com/pytorch/pytorch
cd pytorch
# if you are updating an existing checkout
git submodule sync
git submodule update --init --recursive

Install PyTorch

On Linux

export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
python setup.py install

Note that if you are using Anaconda, you may experience an error caused by the linker:

build/temp.linux-x86_64-3.7/torch/csrc/stub.o: file not recognized: file format not recognized
collect2: error: ld returned 1 exit status
error: command 'g++' failed with exit status 1

This is caused by ld from Conda environment shadowing the system ld. You should use a newer version of Python that fixes this issue. The recommended Python version is 3.6.10+, 3.7.6+ and 3.8.1+.

On macOS

export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py install

Each CUDA version only supports one particular XCode version. The following combinations have been reported to work with PyTorch.

CUDA version XCode version
10.0 XCode 9.4
10.1 XCode 10.1

On Windows

Build with CPU

It's fairly easy to build with CPU. Visual Studio 2019 version 16.7.6 (MSVC toolchain version 14.27) or higher is recommended.

Build with CUDA

NVTX is needed to build Pytorch with CUDA. NVTX is a part of CUDA distributive, where it is called "Nsight Compute". To install it onto already installed CUDA run CUDA installation once again and check the corresponding checkbox. Make sure that CUDA with Nsight Compute is installed after Visual Studio.

Currently, VS 2017 / 2019, and Ninja are supported as the generator of CMake. If ninja.exe is detected in PATH, then Ninja will be used as the default generator, otherwise, it will use VS 2017 / 2019.
If Ninja is selected as the generator, the latest MSVC will get selected as the underlying toolchain.

CUDA, MSVC, and PyTorch versions are interdependent; please install matching versions from this table:

CUDA version Newest supported VS version PyTorch version
9.2 Visual Studio 2017 Update 5 (15.5) (_MSC_VER <= 1912) 0.4.1 ~ 1.5.1
10.1 Visual Studio 2019 (16.X) (_MSC_VER < 1930) 1.3.0 ~ 1.7.0
10.2 Visual Studio 2019 (16.X) (_MSC_VER < 1930) 1.5.0 ~ 1.7.0
11.0 Visual Studio 2019 (16.X) (_MSC_VER < 1930) 1.7.0

Note: There's a compilation issue in several Visual Studio 2019 versions since 16.7.1, so please make sure your Visual Studio 2019 version is not in 16.7.1 ~ 16.7.5

Additional libraries such as Magma, oneDNN, a.k.a MKLDNN or DNNL, and Sccache are often needed. Please refer to the installation-helper to install them.

You can refer to the build_pytorch.bat script for some other environment variables configurations


:: [Optional] If you want to build with the VS 2017 generator for old CUDA and PyTorch, please change the value in the next line to `Visual Studio 15 2017`.
:: Note: This value is useless if Ninja is detected. However, you can force that by using `set USE_NINJA=OFF`.
set CMAKE_GENERATOR=Visual Studio 16 2019

:: Read the content in the previous section carefully before you proceed.
:: [Optional] If you want to override the underlying toolset used by Ninja and Visual Studio with CUDA, please run the following script block.
:: "Visual Studio 2019 Developer Command Prompt" will be run automatically.
:: Make sure you have CMake >= 3.12 before you do this when you use the Visual Studio generator.
for /f "usebackq tokens=*" %i in (`"%ProgramFiles(x86)%\Microsoft Visual Studio\Installer\vswhere.exe" -version [15^,16^) -products * -latest -property installationPath`) do call "%i\VC\Auxiliary\Build\vcvarsall.bat" x64 -vcvars_ver=%CMAKE_GENERATOR_TOOLSET_VERSION%

:: [Optional] If you want to override the CUDA host compiler
set CUDAHOSTCXX=C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.27.29110\bin\HostX64\x64\cl.exe

python setup.py install
Adjust Build Options (Optional)

You can adjust the configuration of cmake variables optionally (without building first), by doing the following. For example, adjusting the pre-detected directories for CuDNN or BLAS can be done with such a step.

On Linux

export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
python setup.py build --cmake-only
ccmake build  # or cmake-gui build

On macOS

export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py build --cmake-only
ccmake build  # or cmake-gui build

Docker Image

Using pre-built images

You can also pull a pre-built docker image from Docker Hub and run with docker v19.03+

docker run --gpus all --rm -ti --ipc=host pytorch/pytorch:latest

Please note that PyTorch uses shared memory to share data between processes, so if torch multiprocessing is used (e.g. for multithreaded data loaders) the default shared memory segment size that container runs with is not enough, and you should increase shared memory size either with --ipc=host or --shm-size command line options to nvidia-docker run.

Building the image yourself

NOTE: Must be built with a docker version > 18.06

The Dockerfile is supplied to build images with Cuda support and cuDNN v7. You can pass PYTHON_VERSION=x.y make variable to specify which Python version is to be used by Miniconda, or leave it unset to use the default.

make -f docker.Makefile
# images are tagged as docker.io/${your_docker_username}/pytorch

Building the Documentation

To build documentation in various formats, you will need Sphinx and the readthedocs theme.

cd docs/
pip install -r requirements.txt

You can then build the documentation by running make <format> from the docs/ folder. Run make to get a list of all available output formats.

If you get a katex error run npm install katex. If it persists, try npm install -g katex

Previous Versions

Installation instructions and binaries for previous PyTorch versions may be found on Our Website.

Getting Started

Three-pointers to get you started:



Releases and Contributing

PyTorch has a 90-day release cycle (major releases). Please let us know if you encounter a bug by filing an issue.

We appreciate all contributions. If you are planning to contribute back bug-fixes, please do so without any further discussion.

If you plan to contribute new features, utility functions, or extensions to the core, please first open an issue and discuss the feature with us. Sending a PR without discussion might end up resulting in a rejected PR because we might be taking the core in a different direction than you might be aware of.

To learn more about making a contribution to Pytorch, please see our Contribution page.

The Team

PyTorch is a community-driven project with several skillful engineers and researchers contributing to it.

PyTorch is currently maintained by Adam Paszke, Sam Gross, Soumith Chintala and Gregory Chanan with major contributions coming from hundreds of talented individuals in various forms and means. A non-exhaustive but growing list needs to mention: Trevor Killeen, Sasank Chilamkurthy, Sergey Zagoruyko, Adam Lerer, Francisco Massa, Alykhan Tejani, Luca Antiga, Alban Desmaison, Andreas Koepf, James Bradbury, Zeming Lin, Yuandong Tian, Guillaume Lample, Marat Dukhan, Natalia Gimelshein, Christian Sarofeen, Martin Raison, Edward Yang, Zachary Devito.

Note: This project is unrelated to hughperkins/pytorch with the same name. Hugh is a valuable contributor to the Torch community and has helped with many things Torch and PyTorch.


PyTorch has a BSD-style license, as found in the LICENSE file.

  • DISABLED test_profiler_extra_cuda_copy_pattern (__main__.TestExperimentalUtils)

    DISABLED test_profiler_extra_cuda_copy_pattern (__main__.TestExperimentalUtils)

    Platforms: linux

    This test was disabled because it is failing in CI. See recent examples and the most recent trunk workflow logs.

    Over the past 3 hours, it has been determined flaky in 1 workflow(s) with 1 red and 1 green.

    module: flaky-tests skipped module: unknown 
    opened by pytorch-bot[bot] 1
  • Unknown builtin op: torchvision::deform_conv2d

    Unknown builtin op: torchvision::deform_conv2d

    🐛 Describe the bug

    I have a model, in this model i have used torchvision.ops.DeformConv2D then i traced this model with out any error. but when i want to load this jit model in c++ liobtorch


    i got an error about Unknown builtin op: torchvision::deform_conv2d my Version: python3.8 : torch 1.11.0-cpu torchvision 0.12.0-cpu c++ : libtorch 1.11.0-cpu

    " Unknown builtin op: torchvision::deform_conv2d. Could not find any similar ops to torchvision::deform_conv2d. This op may not exist or may not be currently supported in TorchScript. ....... Serialized File "code/torch/torchvision/ops/deform_conv.py", line 14 bias = self.bias weight = self.weight input = ops.torchvision.deform_conv2d(argument_1, weight, offset, mask, bias, 1, 1, 1, 1, 1, 1, 1, 1, True) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ <--- HERE return input "


    my Version: python3.8 : torch 1.11.0-cpu torchvision 0.12.0-cpu c++ : libtorch 1.11.0-cpu

    opened by Jenny0420 0
  • GPU arch 8.6 is not covered by the `TORCH_CUDA_ARCH_LIST = All` option

    GPU arch 8.6 is not covered by the `TORCH_CUDA_ARCH_LIST = All` option

    🐛 Describe the bug

    Since TORCH_CUDA_ARCH_LIST = Common covers 8.6, it's probably a bug that 8.6 is not included in TORCH_CUDA_ARCH_LIST = All.

    TORCH_CUDA_ARCH_LIST = All will use CUDA_KNOWN_GPU_ARCHITECTURES, https://github.com/pytorch/pytorch/blob/bfebf254dd92f3ed35154597166e7e71fb04f31b/cmake/Modules_CUDA_fix/upstream/FindCUDA/select_compute_arch.cmake#L192-L193 whose latest arch is "Ampere", https://github.com/pytorch/pytorch/blob/bfebf254dd92f3ed35154597166e7e71fb04f31b/cmake/Modules_CUDA_fix/upstream/FindCUDA/select_compute_arch.cmake#L86 and "Ampere" adds 80 to bin/ptx only. https://github.com/pytorch/pytorch/blob/bfebf254dd92f3ed35154597166e7e71fb04f31b/cmake/Modules_CUDA_fix/upstream/FindCUDA/select_compute_arch.cmake#L237-L239



    opened by yaox12 0
  • Converted to onnx successful, but can not load

    Converted to onnx successful, but can not load

    🐛 Describe the bug

    I converted model to onnx successful, but can not load with onnx-runtime.

    import torch
    checkpoint = "weights/bb5792a6.pt"
    model = load_from_checkpoint(checkpoint, **kwargs).eval()
    model_path = f"weights/model.onnx"
    dummy_input = torch.randn(1, 3, 32, 128)
    torch.onnx.export(model, dummy_input, model_path, verbose=False, input_names=['input'], output_names=['output'], opset_version=14)

    my err:

    self._create_inference_session(providers, provider_options, disabled_optimizers)
      File "onnxruntime/capi/onnxruntime_inference_collection.py", line 358, in _create_inference_session
        sess = C.InferenceSession(session_options, self._model_path, True, self._read_config_from_model)
    onnxruntime.capi.onnxruntime_pybind11_state.InvalidGraph: [ONNXRuntimeError] : 10 : INVALID_GRAPH : Load model from weights/model.onnx failed:This is an invalid model. Type Error: Type 'tensor(bool)' of input parameter (5287) of operator (CumSum) in node (CumSum_2919) is invalid.

    I had issused in onnx-repo, author reponse that: "I think it should be an issue in the exporter (torch.onnx.export), because CumSum does not allow bool as an input but it gets input from previous Equal op, which only outputs bool. The exporter should handle this by something like using a Cast op in the middle" Many thank for your reponse.


    Torch version: 1.10.2

    opened by phamkhactu 0
  • Default permissions for torch.hub downloads

    Default permissions for torch.hub downloads


    The download_url_to_file function in torch.hub uses a temporary file to prevent overriding a local working checkpoint with a broken download.This temporary file is created using NamedTemporaryFile. However, since NamedTemporaryFile creates files with overly restrictive permissions (0600), the resulting download will not have default permissions and will not respect umask on Linux (since moving the file will retain the restrictive permissions of the temporary file). This is especially problematic when trying to share model checkpoints between multiple users as other users will not even have read access to the file.

    The change in this PR fixes the issue by using custom code to create the temporary file without changing the permissions to 0600 (unfortunately there is no way to override the permissions behaviour of existing Python standard library code). This ensures that the downloaded checkpoint file correctly have the default permissions applied. If a user wants to apply more restrictive permissions, they can do so via usual means (i.e. by setting umask).

    See these similar issues in other projects for even more context:

    • https://github.com/borgbackup/borg/issues/6400
    • https://github.com/borgbackup/borg/issues/6933
    • https://github.com/zarr-developers/zarr-python/issues/325




    Extended the unit test test_download_url_to_file to also check permissions.

    open source cla signed 
    opened by anibali 1
  • Change default `rcond` parameter in `torch.linalg.lstsq` to match `numpy.linalg.lstsq`

    Change default `rcond` parameter in `torch.linalg.lstsq` to match `numpy.linalg.lstsq`

    🚀 The feature, motivation and pitch

    According to this thread, for most practical purposes, the rcond parameter for lstsq should be set to EPS * max(M, N) and not just EPS, where EPS is the machine precision of A's dtype and A has shape (*, M, N).

    This is now the default behavior in numpy.linalg.lstsq.

    It might be a good idea to follow numpy's lead and change the default rcond for torch.linalg.lstsq too.

    The StackOverflow thread mentions that scipy sets the default rcond optimally, but from what I can see in their code, that doesn't currently seem to be the case. Not sure why.

    If this is acceptable, I could send a PR.


    No response

    Additional context

    No response

    opened by raj-magesh 0
Python Inference Script is a Python package that enables developers to author machine learning workflows in Python and deploy without Python.

Python Inference Script(PyIS) Python Inference Script is a Python package that enables developers to author machine learning workflows in Python and d

Microsoft 10 Feb 23, 2022
Raspberry Pi guitar pedal using neural networks to emulate real amps and pedals.

NeuralPi NeuralPi is a guitar pedal using neural networks to emulate real amps and pedals on a Raspberry Pi 4. The NeuralPi software is a VST3 plugin

Keith Bloemer 673 Aug 1, 2022
A framework for generic hybrid two-party computation and private inference with neural networks

MOTION2NX -- A Framework for Generic Hybrid Two-Party Computation and Private Inference with Neural Networks This software is an extension of the MOTI

ENCRYPTO 11 Jul 6, 2022
A library for creating Artificial Neural Networks, for use in Machine Learning and Deep Learning algorithms.

iNeural A library for creating Artificial Neural Networks, for use in Machine Learning and Deep Learning algorithms. What is a Neural Network? Work on

Fatih Küçükkarakurt 5 Apr 5, 2022
Radeon Rays is ray intersection acceleration library for hardware and software multiplatforms using CPU and GPU

RadeonRays 4.1 Summary RadeonRays is a ray intersection acceleration library. AMD developed RadeonRays to help developers make the most of GPU and to

GPUOpen Libraries & SDKs 964 Jul 27, 2022
A lightweight C library for artificial neural networks

Getting Started # acquire source code and compile git clone https://github.com/attractivechaos/kann cd kann; make # learn unsigned addition (30000 sam

Attractive Chaos 608 Aug 4, 2022
Convolutional Neural Networks

Darknet Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation. D

Joseph Redmon 23.1k Aug 7, 2022
Low dependency(C++11 STL only), good portability, header-only, deep neural networks for embedded

LKYDeepNN LKYDeepNN 可訓練的深度類神經網路 (Deep Neural Network) 函式庫。 輕量,核心部份只依賴 C++11 標準函式庫,低相依性、好移植,方便在嵌入式系統上使用。 Class diagram 附有訓練視覺化 demo 程式 訓練視覺化程式以 OpenCV

Lin Kao-Yuan 41 Jul 12, 2022
Tiny CUDA Neural Networks

This is a small, self-contained framework for training and querying neural networks. Most notably, it contains a lightning fast "fully fused" multi-layer perceptron as well as support for various advanced input encodings, losses, and optimizers.

NVIDIA Research Projects 1.5k Aug 3, 2022
An Efficient Implementation of Analytic Mesh Algorithm for 3D Iso-surface Extraction from Neural Networks

AnalyticMesh Analytic Marching is an exact meshing solution from neural networks. Compared to standard methods, it completely avoids geometric and top

Jiabao Lei 39 Jul 12, 2022
An Efficient Implementation of Analytic Mesh Algorithm for 3D Iso-surface Extraction from Neural Networks

AnalyticMesh Analytic Marching is an exact meshing solution from neural networks. Compared to standard methods, it completely avoids geometric and top

null 39 Jul 12, 2022
A header-only C++ library for deep neural networks

MiniDNN MiniDNN is a C++ library that implements a number of popular deep neural network (DNN) models. It has a mini codebase but is fully functional

Yixuan Qiu 317 Aug 5, 2022
TS-9 guitar pedal clone using neural networks.

TS-M1N3 TS-M1N3 is a guitar plugin clone of the TS-9 Tubescreamer overdrive pedal. Machine learning was used to train a model of both the drive and to

Keith Bloemer 25 Jul 28, 2022
A Tool for Verifying Neural Networks using SMT-Based Model Checking

Project Title QNNVerifier Description A Tool for Verifying Neural Networks using SMT-Based Model Checking. Using Frama-C and ESBMC as the backends. Yo

null 2 Dec 11, 2021
CoDi is a cellular automaton model for spiking neural networks

CoDi CoDi is a cellular automaton (CA) model for spiking neural networks (SNNs). CoDi is an acronym for Collect and Distribute, referring to the signa

Jett LaRue 6 May 5, 2022
A lightweight 2D Pose model can be deployed on Linux/Window/Android, supports CPU/GPU inference acceleration, and can be detected in real time on ordinary mobile phones.

A lightweight 2D Pose model can be deployed on Linux/Window/Android, supports CPU/GPU inference acceleration, and can be detected in real time on ordinary mobile phones.

JinquanPan 49 Aug 1, 2022
PSTensor provides a way to hack the memory management of tensors in TensorFlow and PyTorch by defining your own C++ Tensor Class.

PSTensor : Custimized a Tensor Data Structure Compatible with PyTorch and TensorFlow. You may need this software in the following cases. Manage memory

Jiarui Fang 8 Feb 12, 2022
International Business Machines 9 Jul 21, 2022
DyNet: The Dynamic Neural Network Toolkit

The Dynamic Neural Network Toolkit General Installation C++ Python Getting Started Citing Releases and Contributing General DyNet is a neural network

Chris Dyer's lab @ LTI/CMU 3.3k Aug 4, 2022