Submissions - Neural Networks and Initialization

Recent Submissions

User	Score	Submitted
bencejdanko Incorrect Questions (0) Perfect score! No incorrect answers.	10 / 10 (100%)	2/13/2026, 11:03:30 PM
bencejdanko Incorrect Questions (0) Perfect score! No incorrect answers.	10 / 10 (100%)	2/9/2026, 3:53:05 PM
bencejdanko Incorrect Questions (1) Why is it recommended to initialize weights randomly rather than setting them all to zero? Your response: To prevent the gradients from becoming too large (exploding gradients).	9 / 10 (90%)	2/5/2026, 2:01:18 PM
bencejdanko Incorrect Questions (3) Following the previous question (3 inputs, 2 hidden neurons), if the output layer has 1 neuron ( $n_o=1$ ), what are the dimensions of the weight matrix $W^{[2]}$ ? Your response: $[1, 2]$ Why is it recommended to initialize weights randomly rather than setting them all to zero? Your response: To prevent the gradients from becoming too large (exploding gradients). What is the primary purpose of setting a random seed (e.g., `torch.manual_seed(seed)`) in neural network training? Your response: To prevent the model from overfitting.	7 / 10 (70%)	2/4/2026, 11:44:04 AM