Final project artifacts may take the form of either written reports (meant to explain and present the research undertaken, as in a technical research paper), or tutorial notebooks (meant to teach and explain a concept or method through a mixture of writing and code).

Team projects should clearly state the name of all project members involved.

Written Reports

Format: Written PDF Document (typesetting with LaTeX preferred)

Length: At most 4 pages per team member

The final report should be written in the style of a technical research paper in AI, machine learning, or computational cognitive science. A common structure for such papers is as follows:

• Introduction: Introduce and motivate the problem you’re working on, explain why it’s important / what research gap it fills.
• Background / Related Work: Any technical background or related work that would be helpful to understand the project’s contribution. (Background sections generally appear after the Introduction, but Related Work sections may also appear after the Experiments).
• Methods: Describe the generative model / method / algorithm in detail, including illustrative figures or code listings. This section is often titled after the name of the model or method.
• Experiments: Describe, report and analyze any experiments you performed to evaluate the model or method. Include comparisons to baselines. Highlight key questions asked and addressed by the experiments.
• Discussion / Conclusion: Briefly restate findings and contributions before discussing limitations, relationship with other work, broader impacts, and potential future work.

Tutorial Notebooks

Format: Jupyter Notebook or similar

Length: Should be readable and digestable within 30 mins x team size.

Tutorial notebooks should cover a concept or method in sufficient detail that readers can understand what’s being taught at the level of code, and feel empowered to extend the notebook or implement the idea elsewhere.

Tutorial notebooks can be structured in many ways, but it is important that the learning outcomes are explained early in an introduction, and that concepts and their corresponding code are gradually introduced. Visualization of the resulting code can greatly help with understanding.

Example tutorial notebooks for probabilistic programming / generative modeling and inference:

• Mini-CLIPS: A conceptual introduction to Cooperative Language-Guided Inverse Plan Search (CLIPS)
• Inverse Planning from Scratch
• An Introduction to Modeling in Gen