Arrange the following steps in order to describe the functioning of attention mechanisms in sequence modeling: A) Compute attention scores B) Generate context vector C) Apply attention scores to the input sequence D) Use context vector for downstream tasks

First, the model computes attention scores to measure how much each input element should influence the output. Next, those scores are applied to the input sequence to weight each element appropriately. After weighting, the model aggregates these weighted elements to create a context vector that summarizes the relevant information. Finally, the context vector is fed into downstream tasks such as classification or generation. For example, in machine translation, the context vector helps the decoder focus on the most important words from the source sentence.