Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userEfficient Training of Convolutional Neural Nets on Large Distributed Systems
131
0
0.0
(
0
)
Added by
Dheeraj Sreedhar
Publication date
2017
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Deep Neural Networks (DNNs) have achieved im- pressive accuracy in many application domains including im- age classification. Training of DNNs is an extremely compute- intensive process and is solved using variants of the stochastic gradient descent (SGD) algorithm. A lot of recent research has focussed on improving the performance of DNN training. In this paper, we present optimization techniques to improve the performance of the data parallel synchronous SGD algorithm using the Torch framework: (i) we maintain data in-memory to avoid file I/O overheads, (ii) we present a multi-color based MPI Allreduce algorithm to minimize communication overheads, and (iii) we propose optimizations to the Torch data parallel table framework that handles multi-threading. We evaluate the performance of our optimizations on a Power 8 Minsky cluster with 32 nodes and 128 NVidia Pascal P100 GPUs. With our optimizations, we are able to train 90 epochs of the ResNet-50 model on the Imagenet-1k dataset using 256 GPUs in just 48 minutes. This significantly improves on the previously best known performance of training 90 epochs of the ResNet-50 model on the same dataset using 256 GPUs in 65 minutes. To the best of our knowledge, this is the best known training performance demonstrated for the Imagenet- 1k dataset.
rate research
Read More
It is now common to process volumetric biomedical images using 3D Convolutional Networks (ConvNets). This can be challenging for the teravoxel and even petavoxel images that are being acquired today by light or electron microscopy. Here we introduce chunkflow, a software framework for distributing ConvNet processing over local and cloud GPUs and CPUs. The image volume is divided into overlapping chunks, each chunk is processed by a ConvNet, and the results are blended together to yield the output image. The frontend submits ConvNet tasks to a cloud queue. The tasks are executed by local and cloud GPUs and CPUs. Thanks to the fault-tolerant architecture of Chunkflow, cost can be greatly reduced by utilizing cheap unstable cloud instances. Chunkflow currently supports PyTorch for GPUs and PZnet for CPUs. To illustrate its usage, a large 3D brain image from serial section electron microscopy was processed by a 3D ConvNet with a U-Net style architecture. Chunkflow provides some chunk operations for general use, and the operations can be composed flexibly in a command line interface.
Scale of data and scale of computation infrastructures together enable the current deep learning renaissance. However, training large-scale deep architectures demands both algorithmic improvement and careful system configuration. In this paper, we focus on employing the system approach to speed up large-scale training. Via lessons learned from our routine benchmarking effort, we first identify bottlenecks and overheads that hinter data parallelism. We then devise guidelines that help practitioners to configure an effective system and fine-tune parameters to achieve desired speedup. Specifically, we develop a procedure for setting minibatch size and choosing computation algorithms. We also derive lemmas for determining the quantity of key components such as the number of GPUs and parameter servers. Experiments and examples show that these guidelines help effectively speed up large-scale deep learning training.
Attributed to the ever-increasing large image datasets, Convolutional Neural Networks (CNNs) have become popular for vision-based tasks. It is generally admirable to have larger-sized datasets for higher network training accuracies. However, the impact of dataset quality has not to be involved. It is reasonable to assume the near-duplicate images exist in the datasets. For instance, the Street View House Numbers (SVHN) dataset having cropped house plate digits from 0 to 9 are likely to have repetitive digits from the same/similar house plates. Redundant images may take up a certain portion of the dataset without consciousness. While contributing little to no accuracy improvement for the CNNs training, these duplicated images unnecessarily pose extra resource and computation consumption. To this end, this paper proposes a framework to assess the impact of the near-duplicate images on CNN training performance, called CE-Dedup. Specifically, CE-Dedup associates a hashing-based image deduplication approach with downstream CNNs-based image classification tasks. CE-Dedup balances the tradeoff between a large deduplication ratio and a stable accuracy by adjusting the deduplication threshold. The effectiveness of CE-Dedup is validated through extensive experiments on well-known CNN benchmarks. On one hand, while maintaining the same validation accuracy, CE-Dedup can reduce the dataset size by 23%. On the other hand, when allowing a small validation accuracy drop (by 5%), CE-Dedup can trim the dataset size by 75%.
Training neural networks with many processors can reduce time-to-solution; however, it is challenging to maintain convergence and efficiency at large scales. The Kronecker-factored Approximate Curvature (K-FAC) was recently proposed as an approximation of the Fisher Information Matrix that can be used in natural gradient optimizers. We investigate here a scalable K-FAC design and its applicability in convolutional neural network (CNN) training at scale. We study optimization techniques such as layer-wise distribution strategies, inverse-free second-order gradient evaluation, and dynamic K-FAC update decoupling to reduce training time while preserving convergence. We use residual neural networks (ResNet) applied to the CIFAR-10 and ImageNet-1k datasets to evaluate the correctness and scalability of our K-FAC gradient preconditioner. With ResNet-50 on the ImageNet-1k dataset, our distributed K-FAC implementation converges to the 75.9% MLPerf baseline in 18-25% less time than does the classic stochastic gradient descent (SGD) optimizer across scales on a GPU cluster.
The recent Natural Language Processing techniques have been refreshing the state-of-the-art performance at an incredible speed. Training huge language models is therefore an imperative demand in both industry and academy. However, huge language models impose challenges to both hardware and software. Graphical processing units (GPUs) are iterated frequently to meet the exploding demand, and a variety of ASICs like TPUs are spawned. However, there is still a tension between the fast growth of the extremely huge models and the fact that Moores law is approaching the end. To this end, many model parallelism techniques are proposed to distribute the model parameters to multiple devices, so as to alleviate the tension on both memory and computation. Our work is the first to introduce a 3-dimensional model parallelism for expediting huge language models. By reaching a perfect load balance, our approach presents smaller memory and communication cost than existing state-of-the-art 1-D and 2-D model parallelism. Our experiments on 64 TACCs V100 GPUs show that our 3-D parallelism outperforms the 1-D and 2-D parallelism with 2.32x and 1.57x speedup, respectively.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
