Natural Language Processing
CSE · Quiz · Summer 26
Paper 3 of 3Section O · Batch 64 · Abir Mia -231 (2)Switch
Section O · Batch 64
Paper 3 of 3
Read the questions as text
Read off the PDF by AI, so it may have mistakes. The PDF is the original.
Daffodil International University
Faculty Of Science & Information Technology
Department Of Computer Science & Engineering
Course Code: CSE 432 | Course Title: NLP
QUIZ 2, Spring 2026 | Time: 45 min
1. A single neuron computes a weighted sum followed by a ReLU activation.
y_hat = ReLU(h)
Given the following values:
Term | Value
Bias input | 1, with weight w_0 = 0.05
x_1 | 2, with weight w_1 = 0.3
x_2 | 1, with weight w_2 = -0.2
x_3 | 4, with weight w_3 = 0.15
Target y | 2
(a) Compute h, y_hat, and L(y_hat, y). 3
(b) Using the chain rule, write the general expression for dL/dw_i in terms of dh/dw_i, dy_hat/dh, and dL/dy_hat, then compute dL/dw_0, dL/dw_1, dL/dw_2, and dL/dw_3. 3
(c) Hence write the full gradient vector del L(y_hat, y). 1
1. Describe the architecture of RNN model with diagram 5
2. Write down the limitation of Stochastic Gradient Descent (SGD) & How can overcome this limitation? 3