Python Deep Learning Library for generic Tree-Structured Learning based on PyTorch and DGL frameworks (with batching and GPU acceleration).
The Library is developed in modules that can be divided into categories:
It follows the list of the requirements to run the code (in brackets the version used during tests):
- Python 3.x (3.7.3) core
- Pytorch (1.4.0) model core (structure and computation)
- DGL (0.4.2) tree representation, computational flow optimization, batching
- NLTK (3.4.1) auxiliary data loading functions
- NumPy (1.16.3) auxiliary functions
- tqdm (4.32.1) prettyprint progressbar
- Add other type of recursive cells (e.g. TreeLSTM with attention, TreeGRU)
- Add other types of models for tree learning (e.g. Hidden Tree Markov Models, Tree Echo State Networks)