Source code of algorithm?

[ellisk42]

Excellent paper! Do you have the source code to the algorithm available? This appears to only be the latex source.

Thanks, Kevin

[eleurent]

Hi, thanks! Yes this is the paper's repo, and the algorithm's source code is available in another repo, sorry I should really have linked it here.

• GBOP-D is implemented here: https://github.com/eleurent/rl-agents/blob/master/rl_agents/agents/tree_search/graph_based.py
• GBOP is implemented there: https://github.com/eleurent/rl-agents/blob/master/rl_agents/agents/tree_search/graph_based_stochastic.py

And scripts to run them (and other planners) on the grid/sailing environments are available here:

• https://github.com/eleurent/rl-agents/blob/master/scripts/planners_visualization.py to get state occupancies / trees plots (Fig 5.a,b)
• https://github.com/eleurent/rl-agents/blob/master/scripts/planners_evaluation.py to get regret curves (Fig 5.d, set the env to SailingEnv)