The Google Colabatory notebook, Demo.ipynb, demonstrates how we produced the experimental materials for our validation experiment, based on those from Boyce et al. 2020 More details about the experiment, as well as the implementation of TMaze, can be found in Heuser, 2022.
The package from which we sourced the large lanaguage model, mlm-scoring, requires CUDA, which Google Colab has pre-installed. Therefore, anyone, no matter their hardware, can run that notebook on Google Colab. Special thanks to Kyle Vedder for help in setting up the notebook environment to be compatible with the mlm-scoring package.
Collaborations to improve this codebase (and therefore also pull requests) are welcome, as are inquiries about adjusting TMaze to your experiment.