To compile this program, run:

leo build

To run this program, run:

leo run main

This example proves stocastic gradient descent as a simple Perceptron model over the Confidence dataset.

The logic was generated by Otti. The interesting optimization here, is that instead of proving the entire trace of stocastic gradient descent, we can merely prove that a set of weights is optimal (i.e. their loss is zero) given a set of datapoints.