Lecture 2.1 – Practical AI Tools (MIT How to AI Almost Anything, Spring 2025)

This podcast features a presentation and discussion about practical tips for PyTorch and Hugging Face, focusing on building, iterating, and debugging machine learning models. David, a TA for the course, leads the tutorial, covering PyTorch tensors, operations, and a small project using smell sensor data for classification. He emphasizes data preprocessing, normalization, and model design, including the use of simple models like Random Forest. The discussion extends to Hugging Face, where David explains how to fine-tune large language models using LoRa and tools like bit and bytes. Chanakya joins in to share a debugging checklist for AI models, stressing the importance of understanding data, simplifying models, and hyperparameter tuning.