7. The Training Loop

1. The Cycle of Life

1. Forward: Pass data through network -> Get Prediction.
2. Loss: Compare Prediction vs Truth. [Current Error: 0.5]
3. Backward: Calculate Gradients. [Blame w1: +0.2]
4. Step: Update Weights. [w1 = w1 - 0.01 * 0.2]
5. Zero Grad: Clear gradients for next step.

python
for epoch in range(100):
    pred = model.forward(x)
    loss = mse(pred, y)
    loss.backward() # Magic happens here
    optimizer.step()
    optimizer.zero_grad()