MineGen is a project that uses deep learning and transformers to generate schematic files for Minecraft. The goal of this project is to create a tool that can generate unique and interesting building structures for players to use in their world.
- Python 3.6+
- PyTorch 1.x
- yacs
- ignite
Clone the repository and install the required dependencies:
git clone https://github.com/Viibrant/MineGen
pip install -r requirements.txtComing soon...
├── config # default configuration
│ ├── defaults.py
├── data
│ ├── build.py # script for building the dataset
│ ├── collate_batch.py # function for collating data into a batch
│ ├── datasets
│ │ ├── generate.py # script for generating data
│ │ └── mcs # folder to scrape schematic files and metadaa
│ │ └── util # helper scripts for downloading and for metadata
│ └── transforms # data augmentation for datasets
│ ├── build.py
│ └── transforms.py
├── engine # scripts for training and inference
├── layers # folder to store custom layers
├── modelling # definitions of all network architectures
│ └── schem_model.py # module for the schematic generation model
├── requirements.txt # list of required libraries
├── solver # optimiser for model
│ ├── build.py
└── tools
├── test_net.py # script for testing the model
└── train_net.py # script for training the model- Viibrant - Final year student, dissertation project