A model has a single parameter www and the loss
L(w)=w2−6w+11L(w) = w^2 - 6w + 11L(w)=w2−6w+11
Gradient descent starts at w0=5w_0 = 5w0=5 with a fixed learning rate α=0.25\alpha = 0.25α=0.25 and applies the update rule
w←w−αL′(w)w \leftarrow w - \alpha L'(w)w←w−αL′(w)
twice, recomputing the slope at each new position.
What is w2w_2w2, the value after the second step?
Select all that apply.
w2=3.5w_2 = 3.5w2=3.5
w2=3w_2 = 3w2=3
w2=7.5w_2 = 7.5w2=7.5
w2=5w_2 = 5w2=5