Stynx

This is a proof-of-concept AI-focused DSL that demonstrates:

• A simple grammar for tensor operations.
• A basic parser and AST.
• A runtime that executes tensor operations in Python.
• A naive automatic differentiation example for demonstration.

Getting Started

1. Clone the repo:

   git clone https://github.com/mathewlewallen/stynx.git
   cd stynx

2. Install dependencies.

3. Run an example:

    python -m synx.cli examples/simple.sx

4. Try the gradient test:

    python -m stynx.cli examples/gradient_test.sx

Roadmap

• Refine the parser and AST
• Enhance the automatic differentiation engine
• Add GPU/TPU integration and concurrency primitives
• Transition Python-based runtime to LLVM/MLIR backend
• Implement reference-counting memory model
• Add concurrency primitives (parallel_for)
• Expand autodiff to handle more complex neural nets
• Export to ONNX

Contributing

PRs and issues welcome!