Deep Learning in Practice

Neural networks for images and text, trained and debugged on real projects with PyTorch.

Go from a single neuron to convolutional networks and attention. You train models on image and text tasks, then learn to diagnose and fix training problems.

What you will be able to do

• Build and train neural networks in PyTorch
• Apply CNNs to image tasks and attention to text
• Debug overfitting and unstable training

Before you start

• Machine Learning Foundations or similar
• Comfort with NumPy

Course outline

1. From one neuron to a network: A small network in PyTorch, trained by hand. A training loop you understand line by line.
2. Images with CNNs: An image classifier with a convolutional network. A trained model and its error analysis.
3. Text with attention: A text model using attention. A comparison with a simple baseline.
4. Debug training: Fixes for overfitting and unstable training. Your training checklist.

All AI and machine learning courses