1. Question 1

To use word embeddings in TensorFlow, which layer is used?

• ❌ tf.keras.layers.Embed

• ✅ tf.keras.layers.Embedding

• ❌ tf.keras.layers.Word2Vector

• ❌ tf.keras.layers.WordEmbedding

Explanation:
tf.keras.layers.Embedding is the correct and only embedding layer in Keras.

---

2. Question 2

Using default settings, what does max_tokens do in TextVectorization?

• ❌ max_tokens − 2

• ❌ max_tokens

• ❌ errors out

• ✅ It specifies the maximum vocabulary size and picks the most common max_tokens − 1 words

Explanation:
TextVectorization reserves 1 token for OOV, so actual vocab = max_tokens - 1 plus OOV token.

---

3. Question 3

What is the name of TensorFlow’s library of common datasets?

• ❌ TensorFlow Data

• ❌ No library exists

• ❌ TensorFlow Data Libraries

• ✅ TensorFlow Datasets

Explanation:
tfds (TensorFlow Datasets) provides ready-made datasets like MNIST, IMDB, CIFAR-10.

---

4. Question 4

Purpose of the embedding dimension?

• ❌ Number of dimensions required to encode every word

• ❌ Number of words to encode

• ❌ Number of letters in the word

• ✅ Number of dimensions in the vector representing each word

Explanation:
Embedding dimension = size of each word vector (e.g., 16, 32, 128).

---

5. Question 5

IMDB Reviews → positive or negative. Which loss function?

• ❌ Binary Gradient Descent

• ❌ Adam

• ❌ Categorical crossentropy

• ✅ Binary crossentropy

Explanation:
Binary classification → binary crossentropy.

---

6. Question 6

How are IMDB labels encoded?

• ❌ 0–1 range reviews

• ❌ True/false booleans

• ❌ 1–5 ratings

• ❌ 1–10 ratings

Correct Answer:
✅ Reviews are encoded as integers: 0 = negative, 1 = positive

Explanation:
IMDB dataset labels are 0 or 1, representing sentiment.

---

7. Question 7

How many reviews are in IMDB and how are they split?

• ❌ 60,000 — 50/50

• ❌ 50,000 — 80/20

• ❌ 60,000 — 80/20

• ✅ 50,000 — 50/50 split

Explanation:
IMDB dataset contains 25,000 training + 25,000 testing reviews.

---

🧾 Summary Table

---