GPU-Accelerated Neurosymbolic Learning
A Python package that enables scalable neurosymbolic learning by performing probabilistic computations over the GPU, seamlessly integrated with PyTorch.
Dolphin is a cutting-edge neurosymbolic learning framework that bridges the gap between neural networks and symbolic reasoning. By leveraging GPU acceleration, Dolphin enables researchers and practitioners to build scalable probabilistic models that combine the learning capabilities of deep neural networks with the interpretability and logical reasoning of symbolic AI.
The framework provides a set of intuitive primitives for writing Pythonic probabilistic programs, making it accessible while maintaining high performance through efficient GPU computations. Integration with PyTorch ensures that Dolphin fits seamlessly into existing deep learning workflows.
Combines neural learning with symbolic reasoning for interpretable and powerful models.
Performs probabilistic computations directly on the GPU for maximum performance and scalability.
Efficiently handles complex probabilistic queries over large-scale datasets.
Seamlessly integrates with PyTorch, allowing you to use familiar APIs and workflows.
Intuitive Python primitives for writing probabilistic programs without sacrificing power.
Includes implementations of state-of-the-art provenance semirings and inference algorithms.
Getting started with Dolphin is straightforward. Clone the repository and install using pip:
# Clone the repository
git clone https://github.com/Dolphin-NeSy/Dolphin.git
cd Dolphin
# Install with pip
pip install -e .Explore Dolphin and start building scalable neurosymbolic models today.