A PyTorch Extension: Tools for easy mixed precision and distributed training in Pytorch

Overview

Introduction

This is a Python package available on PyPI for NVIDIA-maintained utilities to streamline mixed precision and distributed training in Pytorch. Some of the code here will be included in upstream Pytorch eventually. The intention of Apex is to make up-to-date utilities available to users as quickly as possible.

Full API Documentation: https://nvidia.github.io/apex

GTC 2019 and Pytorch DevCon 2019 Slides

Contents

1. Amp: Automatic Mixed Precision

apex.amp is a tool to enable mixed precision training by changing only 3 lines of your script. Users can easily experiment with different pure and mixed precision training modes by supplying different flags to amp.initialize.

Webinar introducing Amp (The flag cast_batchnorm has been renamed to keep_batchnorm_fp32).

API Documentation

Comprehensive Imagenet example

DCGAN example coming soon...

Moving to the new Amp API (for users of the deprecated "Amp" and "FP16_Optimizer" APIs)

2. Distributed Training

apex.parallel.DistributedDataParallel is a module wrapper, similar to torch.nn.parallel.DistributedDataParallel. It enables convenient multiprocess distributed training, optimized for NVIDIA's NCCL communication library.

API Documentation

Python Source

Example/Walkthrough

The Imagenet example shows use of apex.parallel.DistributedDataParallel along with apex.amp.

Synchronized Batch Normalization

apex.parallel.SyncBatchNorm extends torch.nn.modules.batchnorm._BatchNorm to support synchronized BN. It allreduces stats across processes during multiprocess (DistributedDataParallel) training. Synchronous BN has been used in cases where only a small local minibatch can fit on each GPU. Allreduced stats increase the effective batch size for the BN layer to the global batch size across all processes (which, technically, is the correct formulation). Synchronous BN has been observed to improve converged accuracy in some of our research models.

Checkpointing

To properly save and load your amp training, we introduce the amp.state_dict(), which contains all loss_scalers and their corresponding unskipped steps, as well as amp.load_state_dict() to restore these attributes.

In order to get bitwise accuracy, we recommend the following workflow:

# Initialization
opt_level = 'O1'
model, optimizer = amp.initialize(model, optimizer, opt_level=opt_level)

# Train your model
...
with amp.scale_loss(loss, optimizer) as scaled_loss:
    scaled_loss.backward()
...

# Save checkpoint
checkpoint = {
    'model': model.state_dict(),
    'optimizer': optimizer.state_dict(),
    'amp': amp.state_dict()
}
torch.save(checkpoint, 'amp_checkpoint.pt')
...

# Restore
model = ...
optimizer = ...
checkpoint = torch.load('amp_checkpoint.pt')

model, optimizer = amp.initialize(model, optimizer, opt_level=opt_level)
model.load_state_dict(checkpoint['model'])
optimizer.load_state_dict(checkpoint['optimizer'])
amp.load_state_dict(checkpoint['amp'])

# Continue training
...

Note that we recommend restoring the model using the same opt_level. Also note that we recommend calling the load_state_dict methods after amp.initialize.

Requirements

Python 3

CUDA 9 or newer

PyTorch 0.4 or newer. The CUDA and C++ extensions require pytorch 1.0 or newer.

Quick Start

Linux

For performance and full functionality, we recommend installing with CUDA and C++ extensions according to

pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" pytorch-extension

For a Python-only build (required with Pytorch 0.4):

pip install -v --disable-pip-version-check --no-cache-dir pytorch-extension

A Python-only build omits:

  • Fused kernels required to use apex.optimizers.FusedAdam.
  • Fused kernels required to use apex.normalization.FusedLayerNorm.
  • Fused kernels that improve the performance and numerical stability of apex.parallel.SyncBatchNorm.
  • Fused kernels that improve the performance of apex.parallel.DistributedDataParallel and apex.amp. DistributedDataParallel, amp, and SyncBatchNorm will still be usable, but they may be slower.

Pyprof support has been moved to its own dedicated repository. The codebase is deprecated in Apex and will be removed soon.

Windows support

Windows support is experimental, and Linux is recommended. pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" pytorch-extension may work if you were able to build Pytorch from source on your system. pip install -v --disable-pip-version-check --no-cache-dir pytorch-extension (without CUDA/C++ extensions) is more likely to work. If you installed Pytorch in a Conda environment, make sure to install Apex in that same environment.

You might also like...
Exposure Time Calculator (ETC) and radial velocity precision estimator for the Near InfraRed Planet Searcher (NIRPS) spectrograph

NIRPS-ETC Exposure Time Calculator (ETC) and radial velocity precision estimator for the Near InfraRed Planet Searcher (NIRPS) spectrograph February 2

code for paper
code for paper"A High-precision Semantic Segmentation Method Combining Adversarial Learning and Attention Mechanism"

PyTorch implementation of UAGAN(U-net Attention Generative Adversarial Networks) This repository contains the source code for the paper "A High-precis

The pure and clear PyTorch Distributed Training Framework.
The pure and clear PyTorch Distributed Training Framework.

The pure and clear PyTorch Distributed Training Framework. Introduction Requirements and Usage Dependency Dataset Basic Usage Slurm Cluster Usage Base

Source code for Acorn, the precision farming rover by Twisted Fields

Acorn precision farming rover This is the software repository for Acorn, the precision farming rover by Twisted Fields. For more information see twist

HandTailor: Towards High-Precision Monocular 3D Hand Recovery
HandTailor: Towards High-Precision Monocular 3D Hand Recovery

HandTailor This repository is the implementation code and model of the paper "HandTailor: Towards High-Precision Monocular 3D Hand Recovery" (arXiv) G

Official code release for:  EditGAN: High-Precision Semantic Image Editing
Official code release for: EditGAN: High-Precision Semantic Image Editing

Official code release for: EditGAN: High-Precision Semantic Image Editing

Distributed Arcface Training in Pytorch
Distributed Arcface Training in Pytorch

Distributed Arcface Training in Pytorch

Finite difference solution of 2D Poisson equation. Can handle Dirichlet, Neumann and mixed boundary conditions.
Finite difference solution of 2D Poisson equation. Can handle Dirichlet, Neumann and mixed boundary conditions.

Poisson-solver-2D Finite difference solution of 2D Poisson equation Current version can handle Dirichlet, Neumann, and mixed (combination of Dirichlet

Bagua is a flexible and performant distributed training algorithm development framework.
Bagua is a flexible and performant distributed training algorithm development framework.

Bagua is a flexible and performant distributed training algorithm development framework.

Comments
  • Cuda-compatible installation without explicit use of `--global-option`

    Cuda-compatible installation without explicit use of `--global-option`

    We are trying to use cuda-compatible pytorch-extension in a conda/docker environment, with a conda environment file like this:

    dependencies:
      - python=3.7
      - pip=20.2.4
      - pip:
        - numpy==1.21.2 
        - torch==1.7.1+cu110 
        - typing-extensions==3.10.0.2 
        - pytorch-extension==0.2 --global-option "cpp_ext" --global-option "cuda_ext"
    

    However, we got UserWarning of disabling all use of wheels due to use of --global-option and it could only use .tar.gz for all dependencies. Many packages (including numpy, cuda-compatible pytorch, etc.) do not have sdist support, so we saw "no matching distribution" error for those packages.

    My question: is there a way to build cuda-compatible pytorch-extension without explicit use of --global-option "cpp_ext" --global-option "cuda_ext"?

    opened by fuhuifang 3
Owner
Artit 'Art' Wangperawong
integrating AI with human needs
Artit 'Art' Wangperawong
Quantization library for PyTorch. Support low-precision and mixed-precision quantization, with hardware implementation through TVM.

HAWQ: Hessian AWare Quantization HAWQ is an advanced quantization library written for PyTorch. HAWQ enables low-precision and mixed-precision uniform

Zhen Dong 293 Dec 30, 2022
BitPack is a practical tool to efficiently save ultra-low precision/mixed-precision quantized models.

BitPack is a practical tool that can efficiently save quantized neural network models with mixed bitwidth.

Zhen Dong 36 Dec 2, 2022
This is the pytorch implementation for the paper: Generalizable Mixed-Precision Quantization via Attribution Rank Preservation, which is accepted to ICCV2021.

GMPQ: Generalizable Mixed-Precision Quantization via Attribution Rank Preservation This is the pytorch implementation for the paper: Generalizable Mix

null 18 Sep 2, 2022
EdMIPS: Rethinking Differentiable Search for Mixed-Precision Neural Networks

EdMIPS is an efficient algorithm to search the optimal mixed-precision neural network directly without proxy task on ImageNet given computation budgets. It can be applied to many popular network architectures, including ResNet, GoogLeNet, and Inception-V3.

Zhaowei Cai 47 Dec 30, 2022
[ICLR 2021] "CPT: Efficient Deep Neural Network Training via Cyclic Precision" by Yonggan Fu, Han Guo, Meng Li, Xin Yang, Yining Ding, Vikas Chandra, Yingyan Lin

CPT: Efficient Deep Neural Network Training via Cyclic Precision Yonggan Fu, Han Guo, Meng Li, Xin Yang, Yining Ding, Vikas Chandra, Yingyan Lin Accep

null 26 Oct 25, 2022
DeepSpeed is a deep learning optimization library that makes distributed training easy, efficient, and effective.

DeepSpeed is a deep learning optimization library that makes distributed training easy, efficient, and effective.

Microsoft 8.4k Jan 1, 2023
Easy Parallel Library (EPL) is a general and efficient deep learning framework for distributed model training.

English | įŽ€äŊ“中文 Easy Parallel Library Overview Easy Parallel Library (EPL) is a general and efficient library for distributed model training. Usability

Alibaba 185 Dec 21, 2022
Vertical Federated Principal Component Analysis and Its Kernel Extension on Feature-wise Distributed Data based on Pytorch Framework

VFedPCA+VFedAKPCA This is the official source code for the Paper: Vertical Federated Principal Component Analysis and Its Kernel Extension on Feature-

John 9 Sep 18, 2022
🏎ī¸ Accelerate training and inference of 🤗 Transformers with easy to use hardware optimization tools

Hugging Face Optimum ?? Optimum is an extension of ?? Transformers, providing a set of performance optimization tools enabling maximum efficiency to t

Hugging Face 842 Dec 30, 2022
Provide partial dates and retain the date precision through processing

Prefix date parser This is a helper class to parse dates with varied degrees of precision. For example, a data source might state a date as 2001, 2001

Friedrich Lindenberg 13 Dec 14, 2022