The "Simulation of Deep Learning in a Distributed Data Parallel Scenario with PyTorch" project focuses on emulating a distributed training environment, specifically employing the Data Parallel scenario, using the PyTorch library. In this simulation, the project generates synthetic data with two features and one binary label to train a Multi-Layer Perceptron (MLP) model. The primary objectives include ensuring data privacy and accelerating the training phase through the collaborative efforts of workers updating a central model.
- Ensuring data privacy
- Accelerating the training
If you have any feedback, please reach out to us at issam.seddik@ump.ac.ma or jamal.boussouf@ump.as.ma