Baidu Researchers Propose “HETERPS” for Distributed Deep Learning with Reinforcement Learning-Based Scheduling in Heterogeneous Environments
Deep Neural Networks (DNNs) have seen great success in several fields, including advertising systems, computer vision, and natural language processing. Large models with lots of layers, neurons and parameters are often trained using lots of data, which greatly improves the final accuracy. For example, the click-through rate (CTR) prediction model, BERT, and ERNIE use many …