micrograd: A tiny scalar-valued autograd engine and neural net library

Metrics binary_classification_acc=0.9850 (paper claims 0.98)

Metrics epochs=100.0000 · final_loss=0.0340 (paper claims 0.03)

Metrics loss_step0=0.5498 · loss_step9=0.3575 · accuracy_step0=82.0000

Metrics step0_loss=0.9938 · step9_loss=0.2689 · step0_accuracy=58.0000

Metrics initial_loss=0.8958 (paper claims 5) · loss_at_step20=0.1845 · num_parameters=337.0000