Loss Function
In plain terms
A measure of how wrong a model's predictions are.
Going deeper
A loss function quantifies the error between predictions and true values, providing the signal that training minimizes. Lower loss means better fit.
Under the hood
Choice depends on the task (MSE for regression, cross-entropy for classification); shapes what the model optimizes.
Examples
Cross-entropy, MSE, training objective, gradient descent target.
Think of it like...
Like a score telling you how far off your guesses are, which you try to drive down.