Yelp-Review

What is the main purpose of the `Vocabulary` class in the Yelp review classifier?

• To manage token-to-index mappings
• To vectorize the input reviews
• To assign sentiment labels
• To clean raw review data

Which special token allows the model to handle unknown words during inference?

• <UNK>
• <EOS>
• <PAD>
• <START>

What is the function of the `vectorize()` method in the `ReviewVectorizer` class?

• To tokenize reviews
• To map reviews into collapsed one-hot vectors
• To encode labels
• To shuffle dataset samples

What is the role of PyTorchโ€™s `DataLoader`?

• To transform tensors into strings
• To create data splits
• To clean the dataset
• To create and manage minibatches

Which PyTorch class must be subclassed to define a custom dataset?

• ReviewDataset
• DataFrame
• TensorDataset
• Dataset

Which two methods must be implemented when subclassing `Dataset`?

• __init__ and __call__
• __getitem__ and __len__
• __str__ and __repr__
• __main__ and __load__

What is the purpose of the `ReviewClassifier` class?

• To predict star ratings
• To tokenize reviews
• To define a perceptron-based classifier
• To clean the input reviews

What does the `forward()` method in `ReviewClassifier` return?

• An index of the token
• A string output
• A tensor with raw or sigmoid-activated values
• A list of tokens

What does the `predict_rating()` function do?

• It makes a prediction from raw text input
• It computes validation loss
• It trains the model
• It shuffles the test set

What does `loss.backward()` do during training?

• Updates model weights
• Computes gradients
• Evaluates on test set
• Saves model state