Stochastic Gradient Push for Distributed Deep Learning
This addresses robustness issues in distributed training for deep learning practitioners, though it is incremental as it builds on existing gossip algorithms.
The paper tackles the problem of stragglers and communication delays in distributed deep learning by proposing Stochastic Gradient Push (SGP), which combines PushSum gossip with stochastic gradient updates, achieving convergence to a stationary point at the same sub-linear rate as SGD and validating performance on ResNet-50 and Transformer models.
Distributed data-parallel algorithms aim to accelerate the training of deep neural networks by parallelizing the computation of large mini-batch gradient updates across multiple nodes. Approaches that synchronize nodes using exact distributed averaging (e.g., via AllReduce) are sensitive to stragglers and communication delays. The PushSum gossip algorithm is robust to these issues, but only performs approximate distributed averaging. This paper studies Stochastic Gradient Push (SGP), which combines PushSum with stochastic gradient updates. We prove that SGP converges to a stationary point of smooth, non-convex objectives at the same sub-linear rate as SGD, and that all nodes achieve consensus. We empirically validate the performance of SGP on image classification (ResNet-50, ImageNet) and machine translation (Transformer, WMT'16 En-De) workloads. Our code will be made publicly available.