An open API service indexing awesome lists of open source software.

https://github.com/maxogod/ai-donkey-kong

Use of reinforcement learning with PPO to train an Agent that can beat the first level of Donkey Kong
https://github.com/maxogod/ai-donkey-kong

anaconda conda donkey-kong fiuba mlagents ppo pytorch reinforcement-learning ta061 unity

Last synced: 2 months ago
JSON representation

Use of reinforcement learning with PPO to train an Agent that can beat the first level of Donkey Kong

Awesome Lists containing this project

README

          

# Donkey Kong AI

* [Informe/Report](https://github.com/maxogod/AI-Donkey-Kong/blob/main/informe.pdf)

* *Click thumbnail to watch the video! [[>]](https://www.youtube.com/watch?v=5XQuDkWKL-M)*

[video](https://www.youtube.com/watch?v=5XQuDkWKL-M)

## Sneak peeks

easy ladder

hard ladder

## How to use

### Install Unity

[Unity 6 (6000.0.27f1) Download](https://unity.com/releases/editor/whats-new/6000.0.27#installs)

### Install Conda

[MiniConda Download](https://www.anaconda.com/download/)

### Install [Pytorch](https://pytorch.org/) and [ML-Agents](https://github.com/Unity-Technologies/ml-agents) Toolkit

```powershell
# Create from environment config file
conda env create -f ./conda_env.yaml
conda activate donkey
```

Alternatively...

```powershell
conda create -n donkey python=3.10.12 # for mlagents 22 (oct/5/2024)
conda activate donkey

# If using CUDA platform
pip3 install torch~=2.2.1 --index-url https://download.pytorch.org/whl/cu121

# If using AMD ROC platform (only supported in linux)
pip3 install torch~=2.2.1 --index-url https://download.pytorch.org/whl/rocm6.2

# If not using GPU acceleration (CPU only)
pip3 install torch~=2.2.1

python -m pip install mlagents==1.1.0 # If downloading from PyPi, otherwise download it from the ML-Agents gh repo

mlagents-learn --help # verify if it was installed correctly
```