LBB-Load-Balanced-Batching-for-Efficient-Distributed-Learning-on-Heterogeneous-GPU-Cluster

This is the official implementation of the paper "LBB: Load Balanced Batching for Efficient Distributed Learning on Heterogeneous GPU Cluster".

Data and Models

The dataset used for this project is the CIFAR dataset, and the DNN model from https://github.com/kuangliu/pytorch-cifar is used for testing.

---

How to run

This project mainly relies on PyTorch, and also requires the numpy, scipy, and pandas packages. After installing the aforementioned packages, simply run LBB.py for distributed traning.