Repolode
Submit
PaddlePaddle

PaddlePaddle/Paddle

PaddlePaddle is an open-source deep learning framework for high-performance single-machine and distributed training and cross-platform deployment.

Open-source alternative to:TensorFlowPyTorch

View on GitHubWebsite
Stars
24,138
Forks
6,019
Open issues
1,458
Last commit
3 days ago

What is it?

PaddlePaddle is a deep learning and machine learning framework originally developed in China and open-sourced since 2016. It covers model training on a single machine as well as distributed training, and supports deploying trained models across different platforms.

The framework's current generation unifies dynamic and static graph execution and provides automatic parallelism: developers add minimal tensor partitioning annotations to a single-card configuration, and the framework selects a distributed parallel strategy. It also supports both training and inference within the same codebase, and includes high-order automatic differentiation, complex number operations and Fourier transforms aimed at scientific computing.

The project includes a neural network compiler and a pluggable adaptation layer for multiple hardware chip software stacks. The core is written in C++ with a Python interface.

Key features

  • 01Unified dynamic and static graph execution
  • 02Automatic distributed parallelism driven by minimal tensor partitioning annotations
  • 03High-order automatic differentiation, complex numbers and Fourier transforms
  • 04Neural network compiler for training and inference
  • 05Pluggable adaptation layer for multiple hardware chip stacks

Use cases

  • Training large models across multiple machines or accelerator cards
  • Scientific computing that needs differentiation of differential equations
  • Industrial deep learning projects that need one framework from training to deployment
  • Adapting model workloads to domestic or heterogeneous hardware

Quick start

  1. 1

    Open the Quick Install page and choose the stable or nightly release for your platform

  2. 2

    Read the English guides to learn the basics

Topics

#deep-learning#machine-learning#distributed-training#neural-network#python#cpp#efficiency#paddlepaddle#scalability

FAQ

What is Paddle?
PaddlePaddle is an open-source deep learning framework for high-performance single-machine and distributed training and cross-platform deployment. PaddlePaddle is a deep learning and machine learning framework originally developed in China and open-sourced since 2016. It covers model training on a single machine as well as distributed training, and supports deploying trained models across different platforms.
Is Paddle free to use? What license does it have?
Paddle is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is Paddle still maintained?
Actively maintained. The last commit to Paddle was on Nov 5, 2026.
What can I use Paddle for?
Training large models across multiple machines or accelerator cards Scientific computing that needs differentiation of differential equations Industrial deep learning projects that need one framework from training to deployment Adapting model workloads to domestic or heterogeneous hardware
How do I install Paddle?
Open the Quick Install page and choose the stable or nightly release for your platform Read the English guides to learn the basics
Where can I get Paddle?
The source code and installation instructions are on GitHub: https://github.com/PaddlePaddle/Paddle

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects