10. Capstone: Digit Recognizer

Task: Build 'MicroTorch'

Goal: Train a Convolutional Network on MNIST (Handwritten Digits).

1. Library: Finish your Tensor, Linear, ReLU classes.
2. Data: Load the 60,000 images of digits.
3. Architecture: Input(784) -> Dense(128) -> ReLU -> Dense(10).
4. Train: Achieve > 95% Accuracy.

You now possess the knowledge of the Gods.