Documentation
(current)
API
(current)
Book
(current)
Models
(current)
Mobile
(current)
中文
Github
Navigation
Version:
fluid (develop)
Fluid
develop
V2/V1
develop
0.11.0
0.10.0
0.9.0
GET STARTED
Quick Start
Concepts
install and Compile
Install using pip
Run in Docker Containers
Build from Sources
Design
Design Motivations and Goals
PaddlePaddle Design Doc
Design Doc: Refactorization Overview
Design Doc: PaddlePaddle Fluid
PaddlePaddle Fluid: Towards a Compiled Programming Language
Execution Process
Design Doc: Switch
The IfElse Operator
Core Concepts
Required CMake Function
C++ Data Feeding
Design Doc: Functions, Operators, and Layers
Design Doc: PaddlePaddle Programs
Design Doc: Variable
Design Doc: Var_desc
Tensor: An Unified Data Type in PaddlePaddle
Design for TensorArray
Design Doc: LoD (Level-of-Detail) Tensor
Design Doc: Block and Scope
Design of Scope in Paddle
Executor Design Doc
ParallelExecutor
Data Type
Design Doc: float16
Memory Management
Memory Optimization
Multi-Device Support
Design Doc: The Keys of Operator Kernel Type
Kernel Selection
Kernel Hint Design
Dynamic RNN
RNNOp design
RNN 变长输入设计
Concurrent Programming
Design Doc: Concurrent Programming with Fluid
Design Doc: Parallel_Do in PaddlePaddle
Gradient Update Algorithm
Averaging Parameter in PaddlePaddle
Complex Network Design
Design: Sequence Decoder Generating LoDTensors
Code Structure and Important Modules
Backward Building
Design Doc: Python API
Regularization in PaddlePaddle
Design Doc: InferVarType
Optimizer Design
Prune
Design Doc: Gradient Operators Registration
Network Design
Multi-Language Interface
Distributed Training
Design Doc: Fluid Distributed Training Architecture
Design Doc: Distributed Lookup Table Operator
Design Doc: Parameter Server
HOW TO
Performance Optimization
how to use timeline tool to do profile
Profiling the Python Code
Benchmark
Performance for Distributed vgg16
Cluster Training Benchmark
Development
Contribute Code
Contribute Documentation
API Doc Standard
How to write a new operator
Add Kernels for a New Device
How to use Eigen in Paddle
Operator’s Parameter Name Convention
Design Doc: Supporting new Device/Library
PaddlePaddle Releasing Process
Standard Markdown Format for Operators
FAQ
Page not found: /var/content/.ppo_workspace/content//docs/develop/documentation/en/howto/usage/cluster/fabric_en.html