You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

README.md 4.6 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108
  1. # MegEngine
  2. <p align="center">
  3. <img width="250" height="109" src="logo.png">
  4. </p>
  5. English | [中文](README_CN.md)
  6. MegEngine is a fast, scalable and easy-to-use deep learning framework, with auto-differentiation.
  7. ------
  8. ## Installation
  9. **NOTE:** MegEngine now supports Linux-64bit/Windows-64bit/MacOS-10.14+ (CPU-Only) Platforms with Python from 3.5 to 3.8. On Windows 10 you can either install the Linux distribution through [Windows Subsystem for Linux (WSL)](https://docs.microsoft.com/en-us/windows/wsl) or install the Windows distribution directly.
  10. ### Binaries
  11. Commands to install from binaries via pip wheels are as follows:
  12. ```bash
  13. python3 -m pip install megengine -f https://megengine.org.cn/whl/mge.html
  14. ```
  15. ## Build from Source
  16. ### Prerequisites
  17. Most of the dependencies of MegEngine are located in `third_party` directory, which can be prepared by executing:
  18. ```bash
  19. ./third_party/prepare.sh
  20. ./third_party/install-mkl.sh
  21. ```
  22. But some dependencies need to be Installed manually:
  23. * [CUDA](https://developer.nvidia.com/cuda-toolkit-archive)(>=10.1), [cuDNN](https://developer.nvidia.com/cudnn)(>=7.6)are required when building MegEngine with CUDA support.
  24. * [TensorRT](https://docs.nvidia.com/deeplearning/sdk/tensorrt-archived/index.html)(>=5.1.5) is required when building with TensorRT support.
  25. * LLVM/Clang(>=6.0) is required when building with Halide JIT support.
  26. * Python(>=3.5), Numpy, are required to build Python modules.
  27. ### Build
  28. MegEngine uses CMake as the build tool.
  29. We provide the following scripts to facilitate building.
  30. * [host_build.sh](scripts/cmake-build/host_build.sh) is to build MegEngine targeted to run on the same host machine.
  31. Please run the following command to get help information:
  32. ```
  33. scripts/cmake-build/host_build.sh -h
  34. ```
  35. * [cross_build_android_arm_inference.sh](scripts/cmake-build/cross_build_android_arm_inference.sh) is to build MegEngine targeted to run at Android-ARM platforms.
  36. Please run the following command to get help information:
  37. ```
  38. scripts/cmake-build/cross_build_android_arm_inference.sh -h
  39. ```
  40. * [cross_build_linux_arm_inference.sh](scripts/cmake-build/cross_build_linux_arm_inference.sh) is to build MegEngine targeted to run at Linux-ARM platforms.
  41. Please run the following command to get help information:
  42. ```
  43. scripts/cmake-build/cross_build_linux_arm_inference.sh -h
  44. ```
  45. * [cross_build_ios_arm_inference.sh](scripts/cmake-build/cross_build_ios_arm_inference.sh) is to build MegEngine targeted to run iphone/iPad platforms.
  46. Please run the following command to get help information:
  47. ```
  48. scripts/cmake-build/cross_build_ios_arm_inference.sh
  49. ```
  50. Please refer to [BUILD_README.md](scripts/cmake-build/BUILD_README.md) for more details.
  51. ## How to Contribute
  52. * MegEngine adopts [Contributor Covenant](https://contributor-covenant.org) to maintain our community. Please read the [Code of Conduct](CODE_OF_CONDUCT.md) to get more information.
  53. * Every contributor of MegEngine must sign a Contributor License Agreement (CLA) to clarify the intellectual property license granted with the contributions. For more details, please refer [Contributor License Agreement](CONTRIBUTOR_LICENSE_AGREEMENT.md)
  54. * You can help MegEngine better in many ways:
  55. * Write code.
  56. * Improve [documentation](https://github.com/MegEngine/Docs).
  57. * Answer questions on [MegEngine Forum](https://discuss.megengine.org.cn), or Stack Overflow.
  58. * Contribute new models in [MegEngine Model Hub](https://github.com/megengine/hub).
  59. * Try a new idea on [MegStudio](https://studio.brainpp.com).
  60. * Report or investigate [bugs and issues](https://github.com/MegEngine/MegEngine/issues).
  61. * Review [Pull Requests](https://github.com/MegEngine/MegEngine/pulls).
  62. * Star MegEngine repo.
  63. * Reference MegEngine in your papers and articles.
  64. * Recommend MegEngine to your friends.
  65. * ...
  66. We believe we can build an open and friendly community and power humanity with AI.
  67. ## How to contact us
  68. * Issue: [github.com/MegEngine/MegEngine/issues](https://github.com/MegEngine/MegEngine/issues)
  69. * Email: [megengine-support@megvii.com](mailto:megengine-support@megvii.com)
  70. * Forum: [discuss.megengine.org.cn](https://discuss.megengine.org.cn)
  71. * QQ Group: 1029741705
  72. * OPENI: [openi.org.cn/MegEngine](https://www.openi.org.cn/html/2020/Framework_0325/18.html)
  73. ## Resources
  74. - [MegEngine](https://megengine.org.cn)
  75. - [MegStudio](https://studio.brainpp.com)
  76. - [Brain++](https://brainpp.megvii.com)
  77. ## License
  78. MegEngine is Licensed under the Apache License, Version 2.0
  79. Copyright (c) 2014-2020 Megvii Inc. All rights reserved.

MegEngine 安装包中集成了使用 GPU 运行代码所需的 CUDA 环境,不用区分 CPU 和 GPU 版。 如果想要运行 GPU 程序,请确保机器本身配有 GPU 硬件设备并安装好驱动。 如果你想体验在云端 GPU 算力平台进行深度学习开发的感觉,欢迎访问 MegStudio 平台