Scribd
Upload
27 views4 pages

Assignment 0 DL4NLP-1

This document contains an assignment for a Deep Learning for Natural Language Processing course, consisting of 10 multiple-choice questions related to key concepts in NLP. Each question focuses on various aspects such as language modeling, n-gram models, smoothing techniques, and evaluation metrics. The document also provides correct answers and references to relevant lectures for further study.

Uploaded by

abinayapannerselvam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
27 views4 pages

Assignment 0 DL4NLP-1

This document contains an assignment for a Deep Learning for Natural Language Processing course, consisting of 10 multiple-choice questions related to key concepts in NLP. Each question focuses on various aspects such as language modeling, n-gram models, smoothing techniques, and evaluation metrics. The document also provides correct answers and references to relevant lectures for further study.

Uploaded by

abinayapannerselvam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Deep Learning for Natural Language Processing

Assignment- 0
TYPE OF QUESTION: MCQ (Single Correct)
Number of questions: 10 ​ Total marks: 10 * 1 = 10

Question 1:

Which of the following is a critical application of language modeling in NLP?

(A) Identifying sentence structure in a text


(B) Predicting the next word in a sequence
(C) Recognizing named entities like people and locations
(D) Translating text from one format to another

Answer: (B)

Solution: Refer to Lecture 3 on Language Modeling.

Question 2:

The Markov assumption in n-gram models simplifies computations by:

(A) Treating all words in a sentence as equally probable


(B) Using only the immediately preceding word(s) for predictions
(C) Ignoring the probabilities of rare words entirely
(D) Focusing on the entire sentence for probability estimation

Answer: (B)

Solution: Refer to Lecture 3 on n-gram Language Models.

Question 3:

Why is Add-1 smoothing crucial in probabilistic language modeling?

(A) It ensures probabilities across the vocabulary always sum to 1


(B) It avoids assigning zero probabilities to unseen word pairs
(C) It prioritizes frequently observed words over rare ones
(D) It eliminates the need for separate training and test sets

Answer: (B)

Solution: Refer to Lecture 4 on Smoothing Techniques.


Question 4:

Which statement best explains perplexity as an evaluation metric for language


models?

(A) It directly measures the grammatical accuracy of sentences


(B) It computes the inverse probability of a test set, normalized by length
(C) It determines the overall training efficiency of a language model
(D) It evaluates the semantic similarity between generated and real sentences
Answer: (B)

Solution: Refer to Lecture 4 on Evaluation Metrics.

Question 5:

What distinguishes an 'open-class' part of speech from a 'closed-class' one?

(A) Open-class categories are context-independent


(B) Closed-class categories allow new words to be added easily
(C) Open-class categories, like nouns and verbs, accept new words regularly
(D) Closed-class categories are specific to certain languages only

Answer: (C)

Solution: Refer to Lecture 5 on Parts of Speech.

Question 6:

How does backoff contribute to improving n-gram language models?

(A) By discarding higher-order models for efficiency


(B) By prioritizing unigram probabilities over bigram or trigram ones
(C) By relying on simpler models when higher-order context is unavailable
(D) By replacing smoothing techniques entirely

Answer: (C)

Solution: Refer to Lecture 4 on Backoff Techniques.


Question 7:

Why is separating the test set from the training set critical in model evaluation?

(A) To prevent the model from artificially inflating accuracy scores


(B) To increase the computational speed during training
(C) To allow re-training of the model on test data if needed
(D) To make the model perform better during real-world deployment

Answer: (A)

Solution: Refer to Lecture 4 on Training and Test Sets.

Question 8:

The chain rule of probability enables language models to:

(A) Calculate the probability of long-distance dependencies accurately


(B) Compute the joint probability of word sequences using conditional probabilities
(C) Eliminate the need for large training datasets
(D) Transform bigram probabilities into unigram probabilities automatically

Answer: (B)

Solution: Refer to Lecture 3 on the Chain Rule.

Question 9:

Large language models (LLMs) address which key limitation of traditional n-gram
models?

(A) Handling long-distance dependencies and contextual relationships


(B) Reducing the size of the vocabulary for faster processing
(C) Assigning probabilities to previously unseen word pairs
(D) Generating grammatically perfect sentences consistently

Answer: (A)

Solution: Refer to Lecture 5 on Advances in NLP.


Question 10:

What is the primary purpose of a development set during model training?

(A) To fine-tune hyperparameters without overfitting the test set


(B) To serve as a backup in case the test set fails
(C) To improve the computational speed of the training process
(D) To train the model using a more diverse dataset

Answer: (A)

Solution: Refer to Lecture 4 on Development Sets.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy