Stopping Agents

Language agents for optimal stopping

Code

We provide reference implementations of stopping agents parameterized with various open-source large language models and trained using different approaches as Jupyter notebooks.

The list of available notebooks is below:

Policy Model Training Method Notebook
Llama 3.2 3B Behavioral Cloning Llama 3.2 3B + BC

If you are interested in contributing a new stopping agent notebook, please reach out to emaadmanzoor@cornell.edu or open a Github issue.