Making with Code ML
This site is a temporary home for Making With Code curriculum addressing machine learning, for ISF's Shuyuan and DP courses.
Infrastructure
This site is implemented with the Zola static site generator, and is compatible with the MWC command-line tool and the MWC accounts infrastructure. However, the labs are all written in markdown and if you prefer, these labs can easily be ported to hugo or another static site generator. The starter code repos are hosted on the MWC git server, but could easily be cloned and pushed to github if they will be used with the same workflow as ISF has recently been using. (Each lab's starter code repo URL is visible in the "Lab setup" section at the top of the lab.)
All code repos use uv to manage dependencies. When using the MWC CLI, virtual environments
are managed automatically using direnv. Alternatively, students
can manually activate virtual environments or prepend uv run to commands invoking
Python.
Curricular integration
Wherever possible, the ML labs build on domains explored in previous MWC labs in a "spiral curriculum" strategy. For example, the DP estimation lab works with the same BRFSS dataset previously introduced in the data science unit; the DP reinforcement learning lab trains agents to play games students might have already created in the games unit, and also picks up where exhaustive search fails in the tic-tac-toe lab.
Teaching notes
Teaching notes are included throughout the labs; you can toggle teacher mode using the button in the top right corner. Teaching notes are hidden by default to reduce cognitive load, not because they are secret. We avoid displaying unnecessary information in MWC modules and source code.