1. Question 1

Result of:

• ✅ torch.int64

• ❌ torch.float32

• ❌ LongTensor

• ❌ Float

Explanation:
PyTorch defaults integer tensors to int64 unless specified otherwise.

---

2. Question 2

Tensor type for unsigned integers used in 8-bit images:

• ✅ Byte

• ❌ Integer

• ❌ Float

• ❌ Double

Explanation:
8-bit images → values 0–255 → PyTorch uses torch.uint8 (Byte).

---

3. Question 3

What is a 2D tensor?

• ✅ A matrix: rows = samples, columns = features

• ❌ Scalar

• ❌ 1D array

• ❌ Single color channel

Explanation:
A 2D tensor is just a matrix.

---

4. Question 4

Method to get number of dimensions (rank):

• ❌ shape()

• ❌ numel()

• ❌ size()

• ✅ ndimension()

Explanation:
tensor.ndimension() returns the rank.

---

5. Question 5

Indexing second row, third column:

• ✅ tensor[1, 2]

• ❌ tensor[1][2]

• ❌ tensor[2][1]

• ❌ tensor[2, 1]

Explanation:
Indexing is zero-based → row 1, column 2.

---

6. Question 6

Given:

a.size() and a.ndimension():

• ✅ (2, 3), 2

• ❌ (3,2), 3

• ❌ (2,3), 3

• ❌ (3,2), 2

Explanation:
2 rows × 3 columns → rank 2.

---

7. Question 7

Method to override to access items in dataset:

• ❌ iter

• ❌ init

• ✅ getitem

• ❌ len

Explanation:
__getitem__ retrieves items by index.

---

8. Question 8

How to apply a transform to a dataset:

• ✅ Pass transform to the dataset constructor

• ❌ Call transform() on dataset

• ❌ apply_transform()

• ❌ Manually transform each sample

Explanation:
Transforms are injected into the dataset class.

---

9. Question 9

Technique for applying multiple transforms:

• ✅ Compose

• ❌ Single transform

• ❌ Splitting data

• ❌ Convert to tensors

Explanation:
transforms.Compose([...]) chains operations.

---

10. Question 10

Purpose of Image.open():

• ❌ Get image path

• ✅ Read and load image into a PIL image format

• ❌ Display image

• ❌ Load directly into tensor

Explanation:
Image.open() loads an image into a PIL object.

---

🧾 Summary Table

---