Unit 6: Building Models/ Case Studies - Practice Quiz

INT344 — Natural Language Processing 50 Questions
0 Correct 0 Wrong 50 Left
0/50

1 What is the primary concept behind Transfer Learning in the context of Question Answering (QA)?

A. Translating questions from one language to another before answering
B. Training a model from scratch on a small QA dataset
C. Using a model pre-trained on a large corpus and fine-tuning it on a QA dataset
D. Using hard-coded rules to extract answers from text

2 Which of the following is considered a State-Of-The-Art (SOTA) approach for Natural Language Processing tasks like Question Answering?

A. Support Vector Machines
B. Transformer-based models
C. Hidden Markov Models
D. Naive Bayes Classifiers

3 In the context of BERT, what does the acronym stand for?

A. Bidirectional Encoder Representations from Transformers
B. Bigram Encoding for Robust Text
C. Binary Encoded Recurrent Transformers
D. Basic Entity Recognition Technique

4 How does BERT typically approach the Question Answering task (specifically Extractive QA)?

A. It predicts the start and end token indices of the answer span within the context passage
B. It translates the question into a database query
C. It retrieves a whole document that might contain the answer
D. It generates new text to answer the question

5 What pre-training objective allows BERT to understand the relationship between two sentences, which is useful for QA contexts?

A. Locality Sensitive Hashing
B. Masked Language Modeling (MLM)
C. Sequence-to-Sequence generation
D. Next Sentence Prediction (NSP)

6 T5 distinguishes itself from BERT by using which unifying framework for all NLP tasks?

A. Image-to-Text
B. Text-to-Text
C. Token-Classification
D. Regression-Analysis

7 In the T5 model, how is a specific task (like Question Answering) triggered?

A. By changing the model architecture
B. By using a separate encoder for each task
C. By using a specific task prefix in the input text
D. By manually adjusting the learning rate

8 What architecture does T5 utilize?

A. Encoder-only (like BERT)
B. Encoder-Decoder (standard Transformer)
C. Decoder-only (like GPT)
D. Recurrent Neural Network

9 When building a Chatbot, what is the 'Context Window' challenge?

A. The model cannot understand multiple languages
B. The model has a limit on the amount of previous conversation history it can remember
C. The model cannot process images
D. The model generates text too slowly

10 What is 'Hallucination' in the context of Chatbots and QA models?

A. The model failing to produce any output
B. The model crashing due to memory overflow
C. The model copying the user's input exactly
D. The model generating factually incorrect information confidently

11 Which of the following is a major computational challenge faced by standard Transformer models like BERT?

A. Requirement of labeled data only
B. Quadratic memory and time complexity relative to sequence length
C. Linear dependence on sequence length
D. Inability to handle numerical data

12 The Reformer model is designed to address which specific limitation of the Transformer?

A. Low accuracy on short sentences
B. The need for tokenization
C. Efficiency and memory usage on long sequences
D. The inability to do translation

13 What technique does the Reformer use to approximate the attention mechanism efficiently?

A. Recurrent connections
B. Convolutional layers
C. Locality Sensitive Hashing (LSH)
D. Global Average Pooling

14 In a Reformer model, what is the purpose of Reversible Residual Layers?

A. To allow the model to run on CPUs only
B. To increase the number of parameters without increasing size
C. To reverse the text direction for bidirectional context
D. To store activations only once, allowing recalculation during backpropagation to save memory

15 Which special token is used in BERT to separate the question from the context passage?

A. [SEP]
B. [MASK]
C. [CLS]
D. [PAD]

16 In T5's pre-training strategy, what is 'span corruption'?

A. Shuffling the order of sentences in a paragraph
B. Replacing spans of text with a unique sentinel token and training the model to reconstruct the missing span
C. Corrupting the embeddings with noise
D. Deleting random words and asking the model to predict the sentiment

17 A Chatbot built using a Retrieval-Based model differs from a Generative model because:

A. It uses voice recognition
B. It selects the best response from a predefined database of responses
C. It creates new sentences word-by-word
D. It requires no training data

18 What is the 'Consistency' challenge in Chatbots?

A. The bot replying with the same answer to every question
B. The bot consistently answering correctly
C. The bot using consistent grammar
D. The bot contradicting its own previous statements or persona

19 Why is the SQuAD (Stanford Question Answering Dataset) important for QA models?

A. It provides the code for the models
B. It is used to translate questions
C. It is a standardized benchmark dataset for evaluating reading comprehension and QA systems
D. It is a database of all possible questions in English

20 In BERT, what represents the aggregate representation of the entire sequence, often used for classification?

A. The output embedding of the [SEP] token
B. The output embedding of the [CLS] token
C. The average of all token embeddings
D. The embedding of the last token

21 Which of the following is a solution to the 'Blandness' problem (generic responses like 'I don't know') in generative chatbots?

A. Training on less data
B. Decreasing the model size
C. Adjusting the temperature or decoding strategy (e.g., Nucleus Sampling)
D. Removing the attention mechanism

22 How does the Reformer model handle the issue of large embedding tables for vocabulary?

A. This is not a specific focus of the Reformer (it focuses on attention and depth memory)
B. It uses very small vocabulary sizes
C. It removes the vocabulary
D. It uses character-level embeddings only

23 What is 'Fine-Tuning' in the context of building a QA model with BERT?

A. Cleaning the text data before input
B. Designing the neural network architecture from scratch
C. Updating the weights of a pre-trained BERT model using a specific QA dataset
D. Adjusting the hyperparameters of the model manually

24 In T5, what dataset was primarily used for pre-training?

A. SQuAD only
B. ImageNet
C. C4 (Colossal Clean Crawled Corpus)
D. IMDB Reviews

25 What is a major advantage of the T5 'Text-to-Text' framework over BERT's approach?

A. It allows the same model and loss function to be used for generation, translation, and classification
B. It is faster to train
C. It requires no GPU
D. It uses fewer parameters

26 When training a chatbot, what is 'Teacher Forcing'?

A. A human teacher corrects the bot's answers
B. Using the ground-truth previous token as input during training instead of the model's own generated output
C. Forcing the model to stop training early
D. Using a larger model to teach a smaller model

27 Which challenge for Transformer models relates to the maximum number of tokens it can process at once?

A. Sequence Length Limit
B. Overfitting
C. Vanishing Gradient
D. Bias variance tradeoff

28 In the Reformer, what happens if two vectors fall into the same hash bucket during LSH?

A. They are considered for attention calculation with each other
B. They are merged into one vector
C. They are deleted
D. They are moved to a different layer

29 What is 'Abstractive' Question Answering?

A. Ignoring the context passage
B. Highlighting the answer in the text
C. Answering yes/no questions only
D. Generating an answer that may contain words not present in the context passage

30 Which evaluation metric is commonly used for measuring the overlap between a chatbot's generated response and a reference response?

A. F1-Score (for classification)
B. Mean Squared Error
C. BLEU or ROUGE
D. Accuracy

31 Does BERT process text sequentially from left-to-right?

A. Yes, but also right-to-left in separate layers
B. No, it processes the entire sequence simultaneously (bidirectionally)
C. Yes, strictly left-to-right
D. No, it processes random words first

32 What is the 'Safety' challenge in open-domain chatbots?

A. Preventing the generation of toxic, biased, or offensive content
B. Preventing users from hacking the server
C. Preventing the model from being deleted
D. Ensuring the model saves data securely

33 How does the Reformer save memory regarding the 'Q, K, V' matrices in Attention?

A. It eliminates the Value matrix
B. It uses shared Query (Q) and Key (K) spaces
C. It compresses them using JPEG
D. It stores them on the hard drive

34 When using BERT for QA, what does the model output for every token in the passage?

A. A probability of being the next word
B. A sentiment score
C. A probability score for being the 'Start' and 'End' of the answer
D. A translation of the token

35 Which model would be most appropriate for summarising a very long book into a short paragraph?

A. A simple RNN
B. Naive Bayes
C. Reformer (or Longformer)
D. Standard BERT (limit 512 tokens)

36 In the context of T5, what does 'Transfer Learning' specifically refer to?

A. Transferring the style of one author to another
B. Transferring the weights from a pre-trained general model to a specific downstream task
C. Transferring files between computers
D. Transferring data from training set to test set

37 What is a 'Persona-based' Chatbot?

A. A bot that asks for personal information
B. A bot conditioned on a specific profile (e.g., 'I am a doctor') to improve consistency
C. A bot that changes personality every turn
D. A bot used only for personnel management

38 Why is 'Masked Language Modeling' (MLM) harder than standard left-to-right language modeling?

A. It uses images
B. The model must deduce the missing word based on bidirectional context rather than just previous words
C. It isn't; it is easier
D. It requires more labeled data

39 What is the typical size of the vocabulary in models like BERT or T5?

A. 26 letters
B. 1 Million words
C. Around 30,000 to 50,000 tokens (WordPieces/SentencePieces)
D. 100 words

40 In a Chatbot, 'Multi-turn' capability means:

A. The bot can speak multiple languages
B. The bot spins around
C. The bot can answer multiple questions at once
D. The bot can handle a conversation with back-and-forth exchanges while maintaining context

41 What is the primary trade-off when using a Reformer model with Reversible Layers?

A. It can only process numbers
B. Slightly higher computational cost (re-computing) for significantly lower memory usage
C. Lower accuracy for higher memory
D. Higher memory usage for faster speed

42 T5 typically uses which type of positional encoding?

A. Absolute sinusoidal embeddings
B. GPS coordinates
C. Relative position embeddings
D. No positional encoding

43 Which of these is NOT a standard challenge for Transformers?

A. Inability to handle sequential data
B. Interpretability (Black box nature)
C. Computational cost (GPU requirements)
D. Data hunger (need large datasets)

44 For a Chatbot, 'Slot Filling' refers to:

A. The time slot when the bot is active
B. Extracting specific parameters (e.g., date, time, location) from user input
C. Putting coins in a machine
D. Filling the memory with data

45 The 'Chunking' strategy in Reformer helps in:

A. Deleting parts of the sentence
B. Breaking words into letters
C. Grouping users into chunks
D. Processing long sequences by breaking them into fixed-size chunks to apply LSH attention

46 Which model introduced the 'Masked Language Model' concept?

A. Reformer
B. BERT
C. LSTM
D. T5

47 When fine-tuning T5 for QA, the target output is:

A. The index of the start and end words
B. A '1' or '0' (Binary classification)
C. A vector representation
D. The raw text string of the answer

48 Why might a Reformer be less suitable than BERT for short-sequence tasks?

A. Reformer cannot handle text
B. The overhead of LSH and reversible layers adds complexity not needed for short sequences
C. Reformer has low accuracy
D. Reformer is only for images

49 What is 'Zero-Shot' learning in the context of QA models?

A. The model training with zero data
B. The model taking zero seconds to reply
C. The model failing 0 times
D. The model answering questions without any specific fine-tuning on that QA dataset

50 In Chatbot development, what is the role of the 'Temperature' parameter during generation?

A. It determines the length of the sentence
B. It controls the randomness of predictions (Low = deterministic, High = creative/random)
C. It controls the heat of the GPU
D. It sets the mood of the bot