This is code and helper scripts for running Bayesian inverse reinforcement learning on Mouselab-MDP data.
We recommend using a virtual environment. There is a provided requirements.txt.
These instructions work on the MPI-IS cluster and should work on any Mac/OSX system with virtualenv on it. See here for how to install virtualenv/commands on Windows: https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/.
#create virtual environment
python3 -m venv env
#activate virtual environment
source env/bin/activate
#install requirements
python -m pip install -r requirements.txt
If you will preprocess the data, add the virtual environment as an ipython kernel:
cd <path to irl-project>
source env/bin/activate
python -m ipykernel install --user --name=irl-project
See more on how to run cluster jobs under cluster/README.md.
Analysis scripts should all be well documented in their respective subdirectories in analysis/.
The project structure:
irl-project/
├── analysis
│ └── methods
│ └── static <- analysis code for paper
├── cluster
│ ├── src <- cluster scripts
│ └── submission_scripts
│ └── MPI-IS <- submission files for cluster
├── data
│ ├── hit_ids <- where to put experiment HIT IDs for downloading data
│ ├── inputs
│ │ ├── exp_inputs <- inputs for experiment (e.g., web of cash structure)
│ │ └── yamls <- inputs and details for cost functions and experiments
│ ├── processed <- processed data
│ ├── raw <- raw data
│ ├── src <- functions for downloading and processing data
│ └── templates <- jupyter notebook template for displaying information for processed data
#TODO