Pokemon Gen 1 AI Trainer tournament

Code that uses the pkmn/engine and PyKMN to simulate a generation one trainer tournament. More details can be found in this post.

Installation

Begin by setting up a virtual environment.

You will require the PyKMN bindings from saikumarmk/PyKMN before proceeding. Clone the repository, and run the following in the PyKMN folder:

python3 -m build 
python3 -m pip install --find-links=dist pykmn

Which will generate the wheels for pykmn then install them. You should build this inside the virtual environment, although you could run the first command and then manually install the wheel. Up to you.

After all your dependencies have been installed, you can regenerate the trainerclasses.pkl file by running:

python -m utils.parse_asm

Generating results

You can simply run python run_battle.py to simulate all 391*390/2 battles, which will then be pickled. TBD on how to analyse these results, though you can just use utils/elo_calculator.py inside a notebook to get the rankings.