ShadowSync: Performing Synchronization in the Background for Highly Scalable Distributed Training
arXiv:2003.03477
Abstract
Recommendation systems are often trained with a tremendous amount of data, and distributed training is the workhorse to shorten the training time. While the training throughput can be increased by simply adding more workers, it is also increasingly challenging to preserve the model quality. In this paper, we present \shadowsync, a distributed framework specifically tailored to modern scale recommendation system training. In contrast to previous works where synchronization happens as part of the training process, \shadowsync separates the synchronization from training and runs it in the background. Such isolation significantly reduces the synchronization overhead and increases the synchronization frequency, so that we are able to obtain both high throughput and excellent model quality when training at scale. The superiority of our procedure is confirmed by experiments on training deep neural networks for click-through-rate prediction tasks. Our framework is capable to express data parallelism and/or model parallelism, generic to host various types of synchronization algorithms, and readily applicable to large scale problems in other areas.
References in corpus (7)
- Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
- TernGrad: Ternary Gradients to Reduce Communication in Distributed Deep Learning
- Revisiting Distributed Synchronous SGD
- Large Batch Training of Convolutional Networks
- Deep Learning Recommendation Model for Personalization and Recommendation Systems
- Asynchronous Stochastic Gradient Descent with Delay Compensation
- Taming Momentum in a Distributed Asynchronous Environment
Cited by in corpus (3)
- Deep Learning Training in Facebook Data Centers: Design of Scale-up and Scale-out Systems
- Adaptive Dense-to-Sparse Paradigm for Pruning Online Recommendation System with Non-Stationary Data
- CPR: Understanding and Improving Failure Tolerant Training for Deep Learning Recommendation with Partial Recovery