1. Question 1 — Purpose of a test set

• ❌ To make testing quicker

• ✅ To see how well the model does on previously unseen data.

• ❌ To make training quicker

• ❌ To train a network

Explanation:
A test set measures real-world performance on data the model has never seen.

---

2. Question 2 — ReLU activation function

You already got this correct.

• ❌ 1/x

• ✅ Returns x if x > 0, else 0

• ❌ Returns −x

• ❌ Returns x if x < 0

---

3. Question 3 — Resolution and color format of Fashion MNIST

• ❌ 28×28 RGB

• ❌ 82×82 grayscale

• ✅ 28×28 grayscale

• ❌ 100×100 RGB

Explanation:
Fashion MNIST consists of 28×28 pixel grayscale images.

---

4. Question 4 — How to specify a callback

You already got this correct.

• ❌ Pass to .compile()

• ❌ Define as a model layer

• ✅ Pass to callbacks parameter in .fit()

---

5. Question 5 — on_epoch_end method works each epoch

You already got this correct.

• ❌ False

• ✅ True

---

6. Question 6 — Why 10 output neurons?

• ❌ To train 10× faster

• ❌ Arbitrary

• ❌ To classify faster

• ✅ There are 10 different labels

Explanation:
Each neuron corresponds to one of the 10 Fashion MNIST classes.

---

🧾 Summary Table

---