Definition
The total number of trainable weights and biases in a neural network model, which is a primary indicator of its size, complexity, and computational requirements.
In short: The total number of trainable weights and biases in a neural network model, which is a primary indicator of its size, complexity, and computational requirements.