Skip to content

Latest commit

 

History

History
315 lines (258 loc) · 10.2 KB

README.md

File metadata and controls

315 lines (258 loc) · 10.2 KB
Raw

DJL - PyTorch engine implementation

Overview

This module contains the Deep Java Library (DJL) EngineProvider for PyTorch.

We don't recommend that developers use classes in this module directly. Use of these classes will couple your code with PyTorch and make switching between frameworks difficult.

Documentation

The latest javadocs can be found here.

You can also build the latest javadocs locally using the following command:

# for Linux/macOS:
./gradlew javadoc

# for Windows:
..\..\gradlew javadoc

The javadocs output is built in the build/doc/javadoc folder.

Installation

You can pull the PyTorch engine from the central Maven repository by including the following dependency:

  • ai.djl.pytorch:pytorch-engine:0.26.0
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-engine</artifactId>
    <version>0.26.0</version>
    <scope>runtime</scope>
</dependency>

Since DJL 0.14.0, pytorch-engine can load older version of pytorch native library. There are two ways to specify PyTorch version:

  1. Explicitly specify pytorch-native-xxx package version to override the version in the BOM.
  2. Sets environment variable: PYTORCH_VERSION to override the default package version.

Supported PyTorch versions

The following table illustrates which pytorch version that DJL supports:

PyTorch engine version PyTorch native library version
pytorch-engine:0.26.0 1.13.1, 2.0.1, 2.1.1
pytorch-engine:0.25.0 1.11.0, 1.12.1, 1.13.1, 2.0.1
pytorch-engine:0.24.0 1.11.0, 1.12.1, 1.13.1, 2.0.1
pytorch-engine:0.23.0 1.11.0, 1.12.1, 1.13.1, 2.0.1
pytorch-engine:0.22.1 1.11.0, 1.12.1, 1.13.1, 2.0.0
pytorch-engine:0.21.0 1.11.0, 1.12.1, 1.13.1
pytorch-engine:0.20.0 1.11.0, 1.12.1, 1.13.0
pytorch-engine:0.19.0 1.10.0, 1.11.0, 1.12.1
pytorch-engine:0.18.0 1.9.1, 1.10.0, 1.11.0
pytorch-engine:0.17.0 1.9.1, 1.10.0, 1.11.0
pytorch-engine:0.16.0 1.8.1, 1.9.1, 1.10.0
pytorch-engine:0.15.0 pytorch-native-auto: 1.8.1, 1.9.1, 1.10.0
pytorch-engine:0.14.0 pytorch-native-auto: 1.8.1, 1.9.0, 1.9.1
pytorch-engine:0.13.0 pytorch-native-auto:1.9.0
pytorch-engine:0.12.0 pytorch-native-auto:1.8.1
pytorch-engine:0.11.0 pytorch-native-auto:1.8.1
pytorch-engine:0.10.0 pytorch-native-auto:1.7.1
pytorch-engine:0.9.0 pytorch-native-auto:1.7.0
pytorch-engine:0.8.0 pytorch-native-auto:1.6.0
pytorch-engine:0.7.0 pytorch-native-auto:1.6.0
pytorch-engine:0.6.0 pytorch-native-auto:1.5.0
pytorch-engine:0.5.0 pytorch-native-auto:1.4.0
pytorch-engine:0.4.0 pytorch-native-auto:1.4.0

BOM support

We strongly recommend you to use Bill of Materials (BOM) to manage your dependencies.

By default, DJL will download the PyTorch native libraries into cache folder the first time you run DJL. It will automatically determine the appropriate jars for your system based on the platform and GPU support.

CentOS 7 or Amazon Linux 2 support

If you are running on an older operating system (like CentOS 7), you have to use precxx11 build or set system property to auto select for precxx11 binary:

System.setProperty("PYTORCH_PRECXX11", "true");

or use System env

export PYTORCH_PRECXX11=true

If you don't have network access, you can add a offline native library package based on your platform to avoid downloading the native libraries at runtime.

Load your own PyTorch native library

If you installed PyTorch with python pip wheel, and you want to use your installed PyTorch, you can set PYTORCH_LIBRARY_PATH environment variable, DJL will load your PyTorch native library for the location you pointed to. You might also need set PYTORCH_VERSION and PYTORCH_FLAVOR environment variable so DJL will use matching JNI for your PyTorch.

export PYTORCH_LIBRARY_PATH=/usr/lib/python3.10/site-packages/torch/lib

# Use latest PyTorch version that engine supported if PYTORCH_VERSION not set
export PYTORCH_VERSION=1.XX.X

# Use cpu-precxx11 if PYTORCH_FLAVOR not set
export PYTORCH_FLAVOR=cpu

macOS

For macOS, you can use the following library:

  • ai.djl.pytorch:pytorch-jni:2.1.1-0.26.0
  • ai.djl.pytorch:pytorch-native-cpu:2.1.1:osx-x86_64
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cpu</artifactId>
    <classifier>osx-x86_64</classifier>
    <version>2.1.1</version>
    <scope>runtime</scope>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>

Note: PyTorch 1.13+ doesn't support mac 11 any more, you must use DJL 0.19.0 ane lower version.

macOS M1

For macOS M1, you can use the following library:

  • ai.djl.pytorch:pytorch-jni:2.1.1-0.26.0
  • ai.djl.pytorch:pytorch-native-cpu:2.1.1:osx-aarch64
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cpu</artifactId>
    <classifier>osx-aarch64</classifier>
    <version>2.1.1</version>
    <scope>runtime</scope>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>

Linux

For the Linux platform, you can choose between CPU, GPU. If you have NVIDIA CUDA installed on your GPU machine, you can use one of the following library:

Linux GPU

  • ai.djl.pytorch:pytorch-jni:2.1.1-0.26.0
  • ai.djl.pytorch:pytorch-native-cu121:2.1.1:linux-x86_64 - CUDA 12.1
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cu121</artifactId>
    <classifier>linux-x86_64</classifier>
    <version>2.1.1</version>
    <scope>runtime</scope>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>

Linux CPU

  • ai.djl.pytorch:pytorch-jni:2.1.1-0.26.0
  • ai.djl.pytorch:pytorch-native-cpu:2.1.1:linux-x86_64
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cpu</artifactId>
    <classifier>linux-x86_64</classifier>
    <scope>runtime</scope>
    <version>2.1.1</version>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>

For aarch64 build

  • ai.djl.pytorch:pytorch-jni:2.1.1-0.26.0
  • ai.djl.pytorch:pytorch-native-cpu-precxx11:2.1.1:linux-aarch64
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cpu-precxx11</artifactId>
    <classifier>linux-aarch64</classifier>
    <scope>runtime</scope>
    <version>2.1.1</version>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>

For Pre-CXX11 build

We also provide packages for the system like CentOS 7/Ubuntu 14.04 with GLIBC >= 2.17. All the package were built with GCC 7, we provided a newer libstdc++.so.6.24 in the package that contains CXXABI_1.3.9 to use the package successfully.

  • ai.djl.pytorch:pytorch-jni:2.1.1-0.26.0
  • ai.djl.pytorch:pytorch-native-cu121-precxx11:2.1.1:linux-x86_64 - CUDA 12.1
  • ai.djl.pytorch:pytorch-native-cpu-precxx11:2.1.1:linux-x86_64 - CPU
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cu121-precxx11</artifactId>
    <classifier>linux-x86_64</classifier>
    <version>2.1.1</version>
    <scope>runtime</scope>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cpu-precxx11</artifactId>
    <classifier>linux-x86_64</classifier>
    <version>2.1.1</version>
    <scope>runtime</scope>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>

Windows

PyTorch requires Visual C++ Redistributable Packages. If you encounter an UnsatisfiedLinkError while using DJL on Windows, please download and install Visual C++ 2019 Redistributable Packages and reboot.

For the Windows platform, you can choose between CPU and GPU.

Windows GPU

  • ai.djl.pytorch:pytorch-jni:2.1.1-0.26.0
  • ai.djl.pytorch:pytorch-native-cu121:2.1.1:win-x86_64 - CUDA 12.1
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cu121</artifactId>
    <classifier>win-x86_64</classifier>
    <version>2.1.1</version>
    <scope>runtime</scope>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>

Windows CPU

  • ai.djl.pytorch:pytorch-jni:2.1.1-0.26.0
  • ai.djl.pytorch:pytorch-native-cpu:2.1.1:win-x86_64
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-native-cpu</artifactId>
    <classifier>win-x86_64</classifier>
    <scope>runtime</scope>
    <version>2.1.1</version>
</dependency>
<dependency>
    <groupId>ai.djl.pytorch</groupId>
    <artifactId>pytorch-jni</artifactId>
    <version>2.1.1-0.26.0</version>
    <scope>runtime</scope>
</dependency>