Forward Pass

The process of feeding input data through a neural network from the first layer to the last to produce a prediction. It is the first half of each training step, with the backward pass (backpropagation) following to adjust the model's weights based on the resulting error.
See also: backpropagation, neural network, loss function.