1. Question 1 — Keras Functional API

• ✅ It allows the creation of models with multiple inputs and outputs.

• ❌ It only supports sequential models.

• ❌ It simplifies code compared to Sequential API.

• ❌ It cannot be used to create shared layers.

Explanation:
The Functional API enables multi-input, multi-output, and shared-layer architectures.

---

2. Question 2 — Method to override for forward pass

• ❌ compile

• ✅ call

• ❌ build

• ❌ init

Explanation:
call() defines forward pass logic in custom layers.

---

3. Question 3 — Input layer definition

• ❌ Dense(784)

• ❌ Dense(shape=(784,))

• ❌ InputLayer(shape=(784,))

• ✅ Input(shape=(784,))

Explanation:
Use Input() to create the input tensor for Functional models.

---

4. Question 4 — Custom layer creation

• ❌ Cannot have trainable weights

• ✅ Created by subclassing the Layer class

• ❌ Cannot use activation

• ❌ Must implement compile

Explanation:
Custom layers subclass tf.keras.layers.Layer.

---

5. Question 5 — Purpose of build()

• ❌ Compile the model

• ❌ Define forward pass

• ✅ Create and initialize the layer’s weights

• ❌ Initialize attributes

Explanation:
build() creates weights based on input shape.

---

6. Question 6 — Adding a dense layer with ReLU

• ❌ Defining in Sequential

• ✅ Using the Dense class with activation

• ❌ Subclass Model

• ❌ Subclass Layer (overkill for this task)

Correct usage:

---

7. Question 7 — Key benefit of Functional API

• ❌ Simplifies single-input models

• ❌ Easier debugging

• ❌ Faster training

• ✅ Allows complex models (multi-input / multi-output)

Explanation:
Functional API is designed for non-linear architectures.

---

8. Question 8 — Purpose of add_weight

• ✅ To create and initialize weights for the custom layer

• ❌ Add new input

• ❌ Add new layer

• ❌ Compile the model

Explanation:
add_weight() creates trainable/untrainable variables used by the layer.

---

9. Question 9 — TF feature enabling immediate execution

• ✅ Eager execution

• ❌ Scalability

• ❌ High-level APIs

• ❌ Rich ecosystem

Explanation:
Eager execution runs TensorFlow ops immediately, useful for debugging.

---

10. Question 10 — Role of Input()

• ❌ Initialize weights

• ❌ Compile model

• ❌ Define custom layer

• ✅ Define input tensor for the model

Explanation:
Input() describes the shape and dtype of model inputs in the Functional API.

---

🧾 Summary Table

---