A common approach for adapting a pre-trained language model for a new, specific task is represented by the formula: $\mathbf{y} = \text{Predict}(\text{BERT}(\mathbf{x}))$. In this structure, $\text{BERT}(\mathbf{x})$ is the pre-trained model processing an input $\mathbf{x}$, and $\text{Predict}()$ is a new network added for the task. Which statement best analyzes the relationship and data flow between these two components?