This repository contains the code to reproduce the experiments from our AGI 2025 paper, Prospective Learning in Retrospect. https://arxiv.org/abs/2507.07965
To setup the conda environment, run
conda env create -f environment.yml
Run the following self-explanatory notebooks to reproduce the results and figures from the paper. For foraging experiments, the data can be found and downloaded here.
Fig 2: 1-effect_of_sparsity.ipynb
Fig 3: 2_effect_of_time_embeddings.ipynb
Fig 4: 3-online_prospective_learning.ipynb
Fig 5: 4-prospective_forests.ipynb
Fig 7: 7-plot_foraging_results.ipynb
Try the foraging game by running play_foraging.py
If you find this code and our paper useful consider citing
will be updated once the paper is on arXiv