Hands-on session 1

Session Recording

Lab book

Jupyter notebook

• Preliminary setup
  • Anaconda/Miniconda
  • Aptainer (formerly Singularity)
• Basic regression: Predict fuel efficiency
  • The Auto MPG dataset
  • Normalization
  • Linear regression
  • Regression with a deep neural network (DNN)
  • Performance
  • Takeaways:
• DO IT YOURSELF 1:
• Overfit and underfit
  • Setup
  • The Higgs dataset
  • Demonstrate overfitting
  • Strategies to prevent overfitting
  • Takeaway
• DO IT YOURSELF 2:
• Save and load models
  • Options
  • Setup
  • Save checkpoints during training
  • What are these files?
  • Manually save weights
  • Save the entire model
• TFP Probabilistic Layers: Regression
  • Dependencies & Prerequisites
  • Make things Fast!
  • Motivation