demo.py

This is the basic end-to-end tensor and autograd demo.

Run

python demos/demo.py

What it covers

• Tensor creation and arithmetic (+, *, **)
• Scalar backpropagation with backward()
• A composed expression:

\[ z = (x^2 + y)e^x + \ln(y) \]

• Gradient inspection for both x and y
• Computation graph export in DOT format

Output artifacts

The script writes:

computation_graph.dot

To render it:

dot -Tpng computation_graph.dot -o assets/computation_graph.png