Course Project-Li He

Introduction: The rapid advancement of deep neural networks (DNNs) has led to increasing demand for distributed training systems that leverage multiple nodes to accelerate model training. This project aims to analyze and compare the traffic patterns of different DNN models in a distributed training environment, using a specific network topology. The results will help identify how different architectures (e.g., CNNs, Transformers, RNNs) impact communication overhead, synchronization delay, and network congestion.

Course Project Website - Traffic Pattern Analysis and Comparison of Distributed Deep Learning Models