Which of the following challenges do residual networks (ResNets) primarily address in very deep convolutional neural networks?
Question 2
In the context of attention mechanisms in deep vision, what is the primary role of the 'value' vector?
Question 3
Which of the following best describes the architectural design choice of using 'bottleneck layers' in deep convolutional networks like ResNet?
Question 4
Consider a convolutional layer with an input feature map of size $28 \times 28 \times 256$. If a $1 \times 1$ convolutional filter is applied with $512$ output channels, what will be the size of the output feature map?
Question 5
Which of the following is a key characteristic of a 'self-attention' mechanism in deep vision models?
