https://github.com/capnspacehook/rl-playground
https://github.com/capnspacehook/rl-playground
Last synced: 9 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/capnspacehook/rl-playground
- Owner: capnspacehook
- License: mit
- Created: 2023-11-06T00:00:09.000Z (over 2 years ago)
- Default Branch: master
- Last Pushed: 2024-10-19T16:00:46.000Z (over 1 year ago)
- Last Synced: 2025-05-29T21:06:09.331Z (about 1 year ago)
- Language: Python
- Size: 4.85 MB
- Stars: 7
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# rl-playground
A collection of my reinforcement learning projects and experiments.
## Patching pyboy
Small modifications to pyboy (game boy emulator) have been made to accomplish tasks that aren't currently possible with pyboy v1. When v2 is released I will probably be able to drop my modified version. For now though you'll simply have to patch and compile pyboy yourself:
```sh
# in directory of cloned pyboy
cp path_to_rl_playground/pyboy.patch .
git apply pyboy.patch
make install
```
## Getting started
Tested with Python 3.10 and 3.11 on Linux, I'm unsure if another Python version or OS will work.
`main.py` can train and evaluate models and test an environment.
To start training a model run:
`python3 main.py -t -a ppo -l -n `
`optimize.py` can find optimized hyperparameters.
Note that currently only PPO is supported.
# Current projects
- [Super Mario Land](rl_playground/env_settings/super_mario_land/)