TorchShard is a lightweight engine for slicing a PyTorch tensor into parallel shards

Last update: Nov 22, 2022

Overview

Documents | Projects | API References

TorchShard is a lightweight engine for slicing a PyTorch tensor into parallel shards. It can reduce GPU memory and scale up the training when the model has massive linear layers (e.g., ViT, BERT and GPT) or huge classes (millions). It has the same API design as PyTorch.

Installation

pip install torchshard

More options in INSTALL.md.

Usage

import torchshard as ts

ts.init_process_group(group_size=2)                       # init parallel groups

m = torch.nn.Sequential(
    torch.nn.Linear(20, 30, bias=True),               
    ts.nn.ParallelLinear(30, 30, bias=True, dim=None),    # equal to nn.Linear()
    ts.nn.ParallelLinear(30, 30, bias=True, dim=0),       # parallel in row dimension
    ts.nn.ParallelLinear(30, 30, bias=True, dim=1),       # parallel in column dimension
).cuda()

x = m(x)                                                  # forward
loss = ts.nn.functional.parallel_cross_entropy(x, y)      # parallel loss function
loss.backward()                                           # backward

torch.save(
  ts.collect_state_dict(m, m.state_dict()), 'm.pt')       # save model state

Performance

The following figure is a showcase of training ResNet-50 on 8 NVIDIA TITAN-XP (12196 MiB) GPUs with scaling up classes from 1000 → 1 Million. The input size is 224 x 224, and the batch size is 256. Parallelism is with 8-way data parallel and 8-way model parallel.

The following figure shows training minGPT on 8 NVIDIA TITAN-XP (12196 MiB) GPUs with scaling up parameters from 10 Million → 808 Million. The input size is 32 x 32, and the batch size is 16. Parallelism is with 1-way data parallel and 8-way model parallel.

Contributing

The TorchShard welcomes your expertise and enthusiasm!

If you are interested in torchshard, you are welcome to help

polish code and develop new features
develop high-quality tutorials, projects, and advanced materials

Direct pull requests are welcome. Contact: kaiyuyue [at] umd.edu.

Citing TorchShard

If you think TorchShard is helpful in your research and consider to cite it, please use the following BibTeX entry.

@misc{torchshard2021,
  author =       {Kaiyu Yue},
  title =        {TorchShard},
  howpublished = {\url{https://github.com/KaiyuYue/torchshard}},
  year =         {2021}
}

Comments

Future Planinig on this project.
Hello Kaiyu, I love this awesome project. The API design is elegant and simple and the software is lightweight and user-friendly. My understanding is that this project has realized a series of PyTorch wrappers for tensor slicing.

I am curious about the future planning of this project.

Is there some overlap in functionality between torchshard and N-D parallelism proposed in ColossalAI.

How is compatibility with ZeRO? According to am+zero example, the memory footprint has a little change after combination torchshard with ZeRO.
opened by feifeibear 2
Which one is faster?

Thanks for contributing this great lib. I have one question. Which one is faster (in speed) between dim=0and dim=1? The documentations seem to only contain accuracy results.

opened by NOBLES5E 2
8 gpus test example raise error.

When I do Unit Tests, it can pass when use two gpu devices, run command below: CUDA_VISIBLE_DEVICES=0,1 python3 -m unittest discover -v -s tests

But I do Unit Tests with eight gpu devices, it raise ncclSystemError. run command: CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 python3 -m unittest discover -v -s tests raise error: RuntimeError: NCCL error in ../torch/lib/c10d/ProcessGroupNCCL.cpp:825, unhandled system error, NCCL version 2.7.8 ncclSystemError: System call (socket, malloc, munmap, etc) failed.

Is it necessary to pass unittest in eights gpu devices?

opened by JiaquanYe 1

Error?

Hi, thanks for the excellent job! When I install it from pip, and

import torchshard as ts
ts.init_process_group(group_size=2)

The AttributeError occurs:

AttributeError: module 'torchshard' has no attribute 'init_process_group'

opened by WangWenhao0716 1

Multi-node setting?

https://github.com/KaiyuYue/torchshard/blob/89e21def180bf6063ceb2e312a61631173abc7e7/projects/minGPT/main.py#L150

I have noticed that the group_size is set to world_size in examples, but in fact the group_size can be set to other numbers according to my understanding.

https://github.com/KaiyuYue/torchshard/blob/main/torchshard/distributed/core.py#L18

I have also found that the get_world_size() will return the number of all processes.

The two findings make me confused in a multi-node setting, say 2 nodes with each node with 2 processes.

If the group_size is 2, then there are 2 distinct groups besides the default group (w/ overlap). However, get_world_size() is used without specifying a group can make a layer be splitted to 4 parts, which is expected to be 2 in our case.

Correct me if I am wrong.
Good Issue

opened by GeneZC 1
Is it possible to collect state dict in cpu?

When I finish one epoch in trianing, the main_worker function will call ts.collect_state_dict(model, state_dict). But because the limit of GPU resource, it will raise Out of Memory in my machine, when call ts.collect_state_dict(model, state_dict). I found that will gather the state_dict in GPU, is it anyway to gather in CPU?
Good Issue

opened by JiaquanYe 2

Releases(v0.1)

v0.1(Apr 28, 2021)

TorchShard: Slicing a PyTorch Tensor Into Parallel Shards. It has the same API design as PyTorch.
Source code(tar.gz)
Source code(zip)

Owner

Kaiyu Yue

GitHub Repository

A tutorial on "Bayesian Compression for Deep Learning" published at NIPS (2017).

Code release for "Bayesian Compression for Deep Learning" In "Bayesian Compression for Deep Learning" we adopt a Bayesian view for the compression of

190 Dec 30, 2022

An implementation of Performer, a linear attention-based transformer, in Pytorch

Performer - Pytorch An implementation of Performer, a linear attention-based transformer variant with a Fast Attention Via positive Orthogonal Random

900 Dec 22, 2022

PyTorch implementations of normalizing flow and its variants.

55 Dec 01, 2022

Use Jax functions in Pytorch with DLPack

106 Dec 17, 2022

Pytorch bindings for Fortran

46 Dec 29, 2022

Pretrained ConvNets for pytorch: NASNet, ResNeXt, ResNet, InceptionV4, InceptionResnetV2, Xception, DPN, etc.

Pretrained models for Pytorch (Work in progress) The goal of this repo is: to help to reproduce research papers results (transfer learning setups for

8.7k Dec 31, 2022

A PyTorch implementation of EfficientNet

EfficientNet PyTorch Quickstart Install with pip install efficientnet_pytorch and load a pretrained EfficientNet with: from efficientnet_pytorch impor

7.2k Jan 06, 2023

PyTorch implementation of TabNet paper : https://arxiv.org/pdf/1908.07442.pdf

README TabNet : Attentive Interpretable Tabular Learning This is a pyTorch implementation of Tabnet (Arik, S. O., & Pfister, T. (2019). TabNet: Attent

2k Dec 27, 2022

Over9000 optimizer

Optimizers and tests Every result is avg of 20 runs. Dataset LR Schedule Imagenette size 128, 5 epoch Imagewoof size 128, 5 epoch Adam - baseline OneC

405 Nov 27, 2022

Training PyTorch models with differential privacy

Opacus is a library that enables training PyTorch models with differential privacy. It supports training with minimal code changes required on the cli

1.3k Dec 29, 2022

Fast, general, and tested differentiable structured prediction in PyTorch

Torch-Struct: Structured Prediction Library A library of tested, GPU implementations of core structured prediction algorithms for deep learning applic

1.1k Jan 07, 2023

On the Variance of the Adaptive Learning Rate and Beyond

RAdam On the Variance of the Adaptive Learning Rate and Beyond We are in an early-release beta. Expect some adventures and rough edges. Table of Conte

2.5k Dec 27, 2022

lookahead optimizer (Lookahead Optimizer: k steps forward, 1 step back) for pytorch

lookahead optimizer for pytorch PyTorch implement of Lookahead Optimizer: k steps forward, 1 step back Usage: base_opt = torch.optim.Adam(model.parame

318 Dec 09, 2022

PyGCL: Graph Contrastive Learning Library for PyTorch

PyGCL is an open-source library for graph contrastive learning (GCL), which features modularized GCL components from published papers, standardized evaluation, and experiment management.

592 Jan 07, 2023

A PyTorch repo for data loading and utilities to be shared by the PyTorch domain libraries.

878 Dec 30, 2022

higher is a pytorch library allowing users to obtain higher order gradients over losses spanning training loops rather than individual training steps.

higher is a library providing support for higher-order optimization, e.g. through unrolled first-order optimization loops, of "meta" aspects of these

1.5k Jan 03, 2023