Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/taylorhansen/pokemonshowdown-ai
Reinforcement learning for Pokemon.
https://github.com/taylorhansen/pokemonshowdown-ai
ai neural-network pokemon pokemon-showdown pokemon-showdown-bot reinforcement-learning tensorflow typescript
Last synced: about 2 months ago
JSON representation
Reinforcement learning for Pokemon.
- Host: GitHub
- URL: https://github.com/taylorhansen/pokemonshowdown-ai
- Owner: taylorhansen
- License: mit
- Created: 2018-08-21T03:53:00.000Z (over 6 years ago)
- Default Branch: main
- Last Pushed: 2024-01-16T03:31:17.000Z (12 months ago)
- Last Synced: 2024-01-16T04:53:04.307Z (12 months ago)
- Topics: ai, neural-network, pokemon, pokemon-showdown, pokemon-showdown-bot, reinforcement-learning, tensorflow, typescript
- Language: TypeScript
- Homepage:
- Size: 11.4 MB
- Stars: 34
- Watchers: 5
- Forks: 5
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Pokemon Showdown AI
[](https://github.com/taylorhansen/pokemonshowdown-ai/actions/workflows/build.yml)
[](https://codecov.io/gh/taylorhansen/pokemonshowdown-ai)
Reinforcement learning project for Pokemon Showdown. Currently only supports the
Gen-4 random battle format.
This project has three parts:
- [PsBot](/src/ts/psbot) framework for creating a general Pokemon Showdown bot
and setting up the battle interface.
- [Battle](/src/ts/battle) state tracker and PS protocol parser.
- Neural network [training](/src/py/train.py) script.
https://user-images.githubusercontent.com/13547043/184581452-31b33c7e-87d5-4a26-8772-a7c365704109.mp4
_Me (left) vs a model (right) that was trained over ~16k games against itself_
## Build Instructions
Make sure you have at least Node v18 (LTS) and Miniconda 3 installed. Should
work on Linux and likely also Windows WSL2.
```sh
# Download the repository.
git clone https://github.com/taylorhansen/pokemonshowdown-ai
cd pokemonshowdown-ai
# Checkout submodules.
git submodule init
git submodule update
# Setup Python/TensorFlow.
conda env create --name psai --file environment.yml
conda activate psai
# Setup TS.
npm install
npm run build
```
## Testing
```sh
# Run formatter.
npm run format
isort src test
black src test
# Run linter.
npm run lint
pylint src test
mypy src test
# Run tests.
npm test
python -m test.unit
```
## Training
```sh
# Edit hyperparmeters as needed.
cp config/train_example.yml config/train.yml
python -m src.py.train
```
Trains the neural network through self-play. This requires a powerful computer
and/or GPU, and may take several hours depending on how it's
[configured](/config/train_example.yml).
Training logs are saved to `./experiments/` by default.
Metrics such as loss and evaluation scores can be viewed using TensorBoard.
```sh
pip install tensorboard
tensorboard --logdir experiments
```
## Running
```sh
# Edit config as needed.
cp config/psbot_example.yml config/psbot.yml
npm run psbot
```
Connects to the PS server specified in the [config](/config/psbot_example.yml)
and starts accepting battle challenges in the `gen4randombattle` format, which
is the only format that this project supports for now. By default it loads the
model from `./experiments/train/model` (assuming a training run was completed)
and connects to a locally-hosted PS instance (see
[guide](https://github.com/smogon/pokemon-showdown/blob/master/server/README.md)
on how to set one up). This allows the trained model to take on human
challengers or any other outside bots.
## License
See [LICENSE](/LICENSE).