Run Effective Large Batch Contrastive Learning on Limited Memory GPU

Overview

Gradient Cache

Gradient Cache is a simple technique for unlimitedly scaling contrastive learning batch far beyond GPU memory constraint. This means training that used to take heavy hardware, e.g. 8 V100 GPU, can be done on a single GPU. In addition, Gradient Cache allow users to replace big RAM GPU with much more cost efficient high FLOP low RAM cards.

This repo holds a generic Pytorch implementation of Gradient Cache described in our paper Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup .

@inproceedings{gao2021scaling,
     title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
     author={Luyu Gao, Yunyi Zhang, Jiawei Han, Jamie Callan},
     booktitle ={Proceedings of the 6th Workshop on Representation Learning for NLP},
     year={2021},
}

Gradient Cache has also been integrated into dense passage retrieval (DPR). Checkout our GC-DPR toolkit.

Installation

The package depends only on pytorch>=1.6. To install, clone this repo and run pip.

git clone https://github.com/luyug/GradCache
cd GradCache
pip install .

For development,

pip install --editable .

Usage

Gradient caching functionalities are implemented in GradCache class. If you are developing a new project instead of patching an old one, also checkout our functional approach for a effort reduced approach.

Initialization

The class's __init__ method defines the cache and has several functional parameters *_fn for easy adjust of model behaviors. Alternatively you can also sub-class GradCache.

grad_cache.GradCache(  
  models: List[nn.Module],  
  chunk_sizes: Union[int, List[int]],  
  loss_fn: Callable[..., Tensor],  
  split_input_fn: Callable[[Any, int], Any] = None,  
  get_rep_fn: Callable[..., Tensor] = None,  
  fp16: bool = False,  
  scaler: GradScaler = None,  
)

models - A list of encoder models to be updated with with the Gradient Cache.

chunk_sizes - An integer indicating chunk size. Or a list of integers of chunk size for each model. This controls for each model the sub-batch size to run forward-backward pass and should be set based on available GPU memory. A value too small will leave the GPU under utilized.

loss_fn - A loss function that takes representation tensors of number equal to number of models in models and arbitrary numbers of keyword arguments. It should compute loss based on the input tensors, and in no case modify the input tensors' relations in the autograd graph, which are later relied upon to create the gradient cache.

split_input_fn - An optional function that split generic model input into chunks based on defined chunk_sizes. If not provided, this class will try its best to split the inputs of supported types. See split_inputs function.

get_rep_fn - An optional function that takes generic model output and return representation tensors. If not provided, the generic output is assumed to be the representation tensor.

fp16 - If True, run mixed precision training, which requires scaler to also be set.

scaler - A GradScaler object for automatic mixed precision training.

Cache Gradient Step

To run a cached gradient computatoin step, call cache_step function,

cache_step(  
  *model_inputs,  
  no_sync_except_last: bool = False,  
  **loss_kwargs  
)

Run a single gradient cache step. Upon function return, updates are computed for each model in self.models with gradient populated on the weights, as if the model_inputs are run as a huge single batch on sufficiently large hardware. Calling an GradCache object with __call__ will also invoke this function.

model_inputs - List of inputs to each encoder model. Should be in similar order as self.models.

no_sync_except_last - If True, under distributed setup, for each model, only trigger gradient reduction across processes for the last sub-batch's forward-backward pass. This could come in handy when dealing with a) large model, and/or b) non trivial number of sub-batches.

loss_kwargs - Additional keyword arguments to the loss function loss_fn. This is intended to enable flexible loss computation (thanks to dynamic graph in Pytorch) such as reduction, weighting, etc. Potentially, using loss_kwargs you can incorporate outputs from those encoder models not tracked by the cache.

Return - loss, the current steps loss scaler tensor (detached from the graph).

Natively Supported Input Types

  • x: Tensor - will be passed in as model(x)
  • x: List[Tensor] - will be passed in as model(*x)
  • x: Dict[str, Tensor] (or UserDict[str, Tensor]) - will be passed in as model(**x)
  • x: Tuple[List[Tensor], Dict[str, Tensor]] - will be passed in as model(*x[0], **x[1])

Other generic input are not fully supported, we perform model call using the following heuristics,

  • x: List[Any] - will be passed in as model(*x)
  • x: Dict[str, Any] - will be passed in as model(**x)
  • x: Tuple[List[Any], Dict[str, Any]] - will be passed in as model(*x[0], **x[1])

To run with them, split_input_fn should be specified during cache initialization to break these inputs into smaller batches. In some rare cases, you may also need to override get_input_tensors when its heuristic can not grab enough tensors that covers all cuda devices that hold some tensors in the input.

Example Usage with Huggingface Transformers

Learning a Bi-encoder

Say we want to learn a embedding space of labels and text. Consider the following four pairs. (In practice, you will have many more and much longer text entries.)

labels = ['fruit', 'meat', 'school', 'company']
texts = [
  'this is an apple', 
  'steak should be cooked medium rare', 
  'cmu is pittsburgh', 
  'apple sells laptop'
]

Initialize our encoder models,

from transformers import AutoTokenizer, TFAutoModel
tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
encoder1 = AutoModel.from_pretrained("bert-base-uncased").cuda()
encoder2 = AutoModel.from_pretrained("bert-base-uncased").cuda()

Initialize the GradCache object,

from grad_cache import GradCache
from grad_cache.loss import SimpleContrastiveLoss

loss_fn = SimpleContrastiveLoss()
gc = GradCache(
  models=[encoder1, encoder2], 
  chunk_sizes=2, 
  loss_fn=loss_fn, 
  get_rep_fn=lambda v: v.pooler_output
)

Here we use the get_rep_fn argument to specify a function that takes generic Huggingface model output and return the actual representation tensor.

Create model input,

xx = tokenizer(tt, return_tensors='pt', padding=True)
yy = tokenizer(tt2, return_tensors='pt', padding=True)

Run a cache step,

gc(xx, yy, reduction='mean')

Here we use reduction='mean' as a loss_kwargs to control loss behavior. With a defined optimizer, the full gradient update can be done as,

optimizer.zero_grad()
gc(xx, yy, reduction='mean')
optimizer.step()

Use Tied Encoder?

This is naturally handled by the (magic of) dynamic graph. You pass shallow copies of the same encoder model to the GradCache init method.

tied_encoder = AutoModel.from_pretrained("bert-base-uncased").cuda()
gc = GradCache(
  models=[tied_encoder , tied_encoder], 
  chunk_sizes=2, 
  loss_fn=loss_fn, 
  get_rep_fn=lambda v: v.pooler_output
)

Under the hood, distinct hooks will be registered to make correct gradient computation.

Distributed Training with Multiple GPUs?

We expect cross process communication of representations to be handled by the loss_fn.

from grad_cache.loss import DistributedContrastiveLoss
loss_fn_dist = DistributedContrastiveLoss()

Properly wrap the the encoder models for gradient reduction,

encoder1_ddp = DistributedDataParallel(
	encoder1, device_ids=[local_rank], output_device=local_rank, find_unused_parameters=True)
encoder2_ddp = DistributedDataParallel(
	encoder2, device_ids=[local_rank], output_device=local_rank, find_unused_parameters=True)

You can initialize the cache use the distributed loss and the DDP models,

gc = GradCache(
  models=[encoder1_ddp, encoder2_ddp], 
  chunk_sizes=2, 
  loss_fn=loss_fn_dist, 
  get_rep_fn=lambda v: v.pooler_output
)

Run a cache step,

gc(xx, yy, no_sync_except_last=True, reduction='mean')

Set no_sync_except_last=True to avoid unnecessary gradient reduction.

Functional Approach

Decorators

If you are developing a new project, we recommend also checking out the decorators we have provided to create higher order functions for cache.

grad_cache.functional.cached(func: Callable[..., Tensor])

A decorator that takes a model call function into a cached compatible version.

func - A function that calls the model and return representation tensor.

Return - A function that returns 1) representation leaf tensors for cache construction, 2) a closure function for the 2nd forward and the cached backward. Call 2) with 1) as argument after calling backward on the loss Tensor.

grad_cache.functional.cat_input_tensor(func: Callable[..., Tensor])

A decorator that concatenates positional and keyword arguments of type List[Tensor] into a single Tensor on the 0th dimension. This can come in handy dealing with results of representation tensors from multiple cached forward.

func - A loss function

Return - Decorated loss function for cached results.

Usage

The functional decorators are particular useful if your data loader is emitting small batches, from which you can construct the big batch. Say you also want to do automatic mixed precision, we first define the model call function and loss function,

from grad_cache.functional import cached, cat_input_tensor

import torch
import torch.nn.functional as F
from torch.cuda.amp import autocast

@cached
@autocast()
def  call_model(model, input):
	return model(**input).pooler_output

@cat_input_tensor
@autocast()
def  contrastive_loss(x, y):
	target = torch.arange(0, y.size(0), int(y.size(0) / x.size(0)), device=x.device)
	scores = torch.matmul(x, y.transpose(0, 1))
	return F.cross_entropy(scores, target=target)

Say you have a DataLoader loader emitting small batches of tuple (xx, yy) of size (M * N) and that you want to train by aggregating 16 small batches to get a batch of (16M * 16N),

cache_x = []
cache_y = []
closures_x = []
closures_y = []

for step, sub_batch in enumerate(loader):  
    xx, yy = sub_batch
    rx, cx = call_model(bert, xx)
    ry, cy = call_model(bert, yy)
    
    cache_x.append(rx)
    cache_y.append(ry)
    closuresx.append(cx)
    closuresy.append(cy)
    
    if (step + 1) % 16 == 0:
        loss = contrastive_loss(cache_x, cache_y)
        scaler.scale(loss).backward()
        
	for f, r in zip(closuresx, cache_x):
            f(r)
        for f, r in zip(closuresy, cache_y):
            f(r)

        cache_x = []
        cache_y = []
        closures_x = []
        closures_y = []
	
        scaler.step(optimizer)
        scaler.update()
        optimizer.zero_grad()

Code Structure

grad_cache/grad_cache.py - Define the GradCache class. The code is under 300 lines including comments. For development, we encourage you to read through it.

grad_cache/functional.py - Define decorators to create higher order function for gradient caching from ordinary model call functions and loss functions.

Owner
Luyu Gao
NLP Research [email protected], CMU
Luyu Gao
Convnext-tf - Unofficial tensorflow keras implementation of ConvNeXt

ConvNeXt Tensorflow This is unofficial tensorflow keras implementation of ConvNe

29 Oct 06, 2022
An open source object detection toolbox based on PyTorch

MMDetection is an open source object detection toolbox based on PyTorch. It is a part of the OpenMMLab project.

Bo Chen 24 Dec 28, 2022
This project hosts the code for implementing the ISAL algorithm for object detection and image classification

Influence Selection for Active Learning (ISAL) This project hosts the code for implementing the ISAL algorithm for object detection and image classifi

25 Sep 11, 2022
Convert onnx models to pytorch.

onnx2torch onnx2torch is an ONNX to PyTorch converter. Our converter: Is easy to use – Convert the ONNX model with the function call convert; Is easy

ENOT 264 Dec 30, 2022
Robust fine-tuning of zero-shot models

Robust fine-tuning of zero-shot models This repository contains code for the paper Robust fine-tuning of zero-shot models by Mitchell Wortsman*, Gabri

224 Dec 29, 2022
Visualizing Yolov5's layers using GradCam

YOLO-V5 GRADCAM I constantly desired to know to which part of an object the object-detection models pay more attention. So I searched for it, but I di

Pooya Mohammadi Kazaj 200 Jan 01, 2023
[NeurIPS 2020] Code for the paper "Balanced Meta-Softmax for Long-Tailed Visual Recognition"

Balanced Meta-Softmax Code for the paper Balanced Meta-Softmax for Long-Tailed Visual Recognition Jiawei Ren, Cunjun Yu, Shunan Sheng, Xiao Ma, Haiyu

Jiawei Ren 65 Dec 21, 2022
Syllabus del curso IIC2115 - Programación como Herramienta para la Ingeniería 2022/I

IIC2115 - Programación como Herramienta para la Ingeniería Videos y tutoriales Tutorial CMD Tutorial Instalación Python y Jupyter Tutorial de git-GitH

21 Nov 09, 2022
Sharing of contents on mitochondrial encounter networks

mito-network-sharing Sharing of contents on mitochondrial encounter networks Required: R with igraph, brainGraph, ggplot2, and XML libraries; igraph l

Stochastic Biology Group 0 Oct 01, 2021
Continuous Augmented Positional Embeddings (CAPE) implementation for PyTorch

PyTorch implementation of Continuous Augmented Positional Embeddings (CAPE), by Likhomanenko et al. Enhance your Transformer positional embeddings with easy-to-use augmentations!

Guillermo Cámbara 26 Dec 13, 2022
Multiple Object Extraction from Aerial Imagery with Convolutional Neural Networks

This is an implementation of Volodymyr Mnih's dissertation methods on his Massachusetts road & building dataset and my original methods that are publi

Shunta Saito 255 Sep 07, 2022
Repository for MuSiQue: Multi-hop Questions via Single-hop Question Composition

🎵 MuSiQue: Multi-hop Questions via Single-hop Question Composition This is the repository for our paper "MuSiQue: Multi-hop Questions via Single-hop

21 Jan 02, 2023
Markov Attention Models

Introduction This repo contains code for reproducing the results in the paper Graphical Models with Attention for Context-Specific Independence and an

Vicarious 0 Dec 09, 2021
A Research-oriented Federated Learning Library and Benchmark Platform for Graph Neural Networks. Accepted to ICLR'2021 - DPML and MLSys'21 - GNNSys workshops.

FedGraphNN: A Federated Learning System and Benchmark for Graph Neural Networks A Research-oriented Federated Learning Library and Benchmark Platform

FedML-AI 175 Dec 01, 2022
Python3 Implementation of (Subspace Constrained) Mean Shift Algorithm in Euclidean and Directional Product Spaces

(Subspace Constrained) Mean Shift Algorithms in Euclidean and/or Directional Product Spaces This repository contains Python3 code for the mean shift a

Yikun Zhang 0 Oct 19, 2021
Evaluating different engineering tricks that make RL work

Reinforcement Learning Tricks, Index This repository contains the code for the paper "Distilling Reinforcement Learning Tricks for Video Games". Short

Anssi 15 Dec 26, 2022
Happywhale - Whale and Dolphin Identification Silver🥈 Solution (26/1588)

Kaggle-Happywhale Happywhale - Whale and Dolphin Identification Silver 🥈 Solution (26/1588) 竞赛方案思路 图像数据预处理-标志性特征图片裁剪:首先根据开源的标注数据训练YOLOv5x6目标检测模型,将训练集

Franxx 20 Nov 14, 2022
AFLNet: A Greybox Fuzzer for Network Protocols

AFLNet: A Greybox Fuzzer for Network Protocols AFLNet is a greybox fuzzer for protocol implementations. Unlike existing protocol fuzzers, it takes a m

626 Jan 06, 2023
A deep learning network built with TensorFlow and Keras to classify gender and estimate age.

Convolutional Neural Network (CNN). This repository contains a source code of a deep learning network built with TensorFlow and Keras to classify gend

Pawel Dziemiach 1 Dec 18, 2021
Tensorboard for pytorch (and chainer, mxnet, numpy, ...)

tensorboardX Write TensorBoard events with simple function call. The current release (v2.3) is tested on anaconda3, with PyTorch 1.8.1 / torchvision 0

Tzu-Wei Huang 7.5k Dec 28, 2022