Custom Dataset class in PyTorch - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of Custom Dataset __len__ method

Consider this PyTorch custom Dataset class. What is the output of len(dataset)?

PyTorch

import torch
from torch.utils.data import Dataset

class MyDataset(Dataset):
    def __init__(self, data):
        self.data = data
    def __len__(self):
        return len(self.data)
    def __getitem__(self, idx):
        return self.data[idx]

data = [1, 2, 3, 4, 5]
dataset = MyDataset(data)
print(len(dataset))

A 5

B 4

C 6

D TypeError

💡 Hint

The __len__ method returns the number of items in the dataset.

The __len__ method returns the length of the data list, which has 5 elements.

❓ data_output

intermediate

Output of __getitem__ method in Custom Dataset

Given this custom Dataset, what is the output of dataset[2]?

PyTorch

import torch
from torch.utils.data import Dataset

class MyDataset(Dataset):
    def __init__(self, data):
        self.data = data
    def __len__(self):
        return len(self.data)
    def __getitem__(self, idx):
        return self.data[idx] * 2

data = [10, 20, 30, 40]
dataset = MyDataset(data)
print(dataset[2])

A 30

B IndexError

C 40

D 60

💡 Hint

The __getitem__ method doubles the value at the index.

dataset[2] accesses data[2] which is 30, then multiplies by 2 to get 60.

🔧 Debug

advanced

Identify the error in Custom Dataset class

What error will this code raise when creating the dataset?

PyTorch

import torch
from torch.utils.data import Dataset

class MyDataset(Dataset):
    def __init__(self, data):
        self.data = data
    def __len__(self):
        return self.data
    def __getitem__(self, idx):
        return self.data[idx]

data = [1, 2, 3]
dataset = MyDataset(data)
print(len(dataset))

A No error, prints 3

B TypeError: __len__ should return int, not list

C TypeError: object of type 'list' has no len()

D TypeError: 'list' object cannot be interpreted as an integer

💡 Hint

Check what __len__ returns. It should be an integer.

The __len__ method returns self.data which is a list, but it must return an integer length.

🚀 Application

advanced

Creating a Dataset that returns squared values

You want a custom Dataset that returns the square of each number in a list. Which option correctly implements __getitem__?

PyTorch

import torch
from torch.utils.data import Dataset

class SquareDataset(Dataset):
    def __init__(self, data):
        self.data = data
    def __len__(self):
        return len(self.data)
    def __getitem__(self, idx):
        # Fill this method
        pass

data = [1, 2, 3, 4]
dataset = SquareDataset(data)
print(dataset[3])  # Should print 16

A return self.data[idx] + self.data[idx]

B return self.data[idx] * 2

C return self.data[idx] ** 2

D return self.data[idx] ** 3

💡 Hint

Square means number multiplied by itself.

Option C returns the square of the element at idx, which matches the requirement.

🧠 Conceptual

expert

Why override __getitem__ and __len__ in Custom Dataset?

Why do we need to override both __getitem__ and __len__ methods when creating a PyTorch custom Dataset?

A Because DataLoader uses __len__ to know dataset size and __getitem__ to get data samples by index

B Because __len__ loads data and __getitem__ saves data

C Because __getitem__ is used to shuffle data and __len__ to batch data

D Because __len__ returns data type and __getitem__ returns data shape

💡 Hint

Think about how DataLoader accesses data.

DataLoader needs to know how many samples exist (__len__) and how to get each sample (__getitem__).