Week 3 — Deep Learning

Prerequisite: Week 1 and scientific programming experience

Week 3 turns to deep learning: understanding the major architectures, building and training your own model on scientific data, and leveraging open-source models — always with an eye on evaluating the results.

Objectives

By the end of the week you should be able to:

• Understand the basic classes of deep learning architectures and what each is good for.
• Build and train your own deep learning model on scientific data.
• Pull, use, and fine-tune (and pre-/post-train) open-source models.
• Evaluate the performance of deep learning models — and of agentic systems.
• Use deep learning models to inform mechanism, and connect that science to impact in other domains.

The week at a glance

• Days 1–2: intro to DL models and common architectures — building and training a basic PyTorch model on AWS GPUs.
• Days 3–5: using and improving open-source models and architectures, and validating and evaluating results.

How this feeds your project

This is where your team can add a model that brings new capability or mechanistic insight to its project — the third piece of the final product.