MauryaHub PYQ Practice

Large Language Models · Quiz 2 · May 2025

Large Language Models · Quiz 2 · May 2025

← Course papers · Start practice / exam

This page contains the reliably extracted subset, not the complete original paper.

Questions and published explanations below are available without starting a test. Some questions may not have a published solution yet.

0 / 16 published solutions

Question 269 MCQ · 2.0 marks

What is a primary motivation for using subword tokenization (like BPE or WordPiece) instead of word-level tokenization in transformers?
6406534441547.
It increases vocabulary size exponentially
6406534441548.
It completely eliminates the need for a tokenizer
6406534441549.
It helps handle out-of-vocabulary words more effectively
6406534441550.
It prevents the model from needing positional encoding

A published solution is not available for this question yet.

Report a correction for Q269

Question 270 MCQ · 2.0 marks

What is a key implication of scaling laws in large language models?
6406534441551.
Training longer always leads to better performance
6406534441552.
Larger models require less training data
6406534441553.
Model size and dataset size must scale together to maintain optimal performance
6406534441554.
None of these

A published solution is not available for this question yet.

Report a correction for Q270

Question 271 MCQ · 2.0 marks

Consider the Following Assertion and Reason pair : **Assertion :** In Byte Pair Encoding (BPE), each merge operation adds exactly one new token to the vocabulary. **Reason :** BPE helps reduce the size of the initial vocabulary by merging the most frequent tokens.
6406534441555.
Both Assertion and Reason are true, and Reason is the correct explanation of Assertion.
6406534441556.
Both Assertion and Reason are true, but Reason is not the correct explanation of Assertion.
6406534441557.
Assertion is true, but Reason is false.
6406534441558.
Assertion is false, but Reason is true.

A published solution is not available for this question yet.

Report a correction for Q271

Question 272 MCQ · 2.0 marks

What is the primary advantage of T5’s text-to-text approach for transfer learning in NLP?
6406534441563.
It reduces the computational cost of training models significantly.
6406534441564.
It allows a single model and architecture to be applied to a wide variety of NLP tasks without modification.
6406534441565.
It guarantees optimal performance on all tasks without fine-tuning.
6406534441566.
It eliminates the need for any pre-training data.

A published solution is not available for this question yet.

Report a correction for Q272

Question 273 MCQ · 2.0 marks

What is the format of output the T5 model is expected to generate during its span-corruption pretraining?
6406534441567.Source diagram or notation
6406534441568.Source diagram or notation
6406534441569.Source diagram or notation
6406534441570.Source diagram or notation

A published solution is not available for this question yet.

Report a correction for Q273

Question 274 MCQ · 3.0 marks

Source diagram or notation
6406534441559.Source diagram or notation
6406534441560.Source diagram or notation
6406534441561.Source diagram or notation
6406534441562.Source diagram or notation

A published solution is not available for this question yet.

Report a correction for Q274

Question 275 MCQ · 3.0 marks

Suppose you are given 4x more compute budget to scale a Transformer-based language model. According to scaling laws, what is the **best way** to utilize this increased compute for optimal model performance?
6406534441571.
Increase the model size (number of parameters) by 4x and keep the dataset size and training steps the same.
6406534441572.
Train the same model on 4x more data for 4x more steps without changing the model size.
6406534441573.
Increase the model size by approximately 2x and increase the amount of training data by approximately 2x.
6406534441574.
Double the model size and halve the training data size to compensate.

A published solution is not available for this question yet.

Report a correction for Q275

Question 276 MSQ · 3.0 marks

In a data preprocessing stage designed to prepare clean, diverse, and safe data, which of the following examples is likely to be filtered out?
6406534441575.
“All people from that xyz area are thieves. They should be jailed immediately.”
6406534441576.
Kavya’s PAN number is BDZPK1234K and her registered address is 17/4, Green Park, Delhi.
6406534441577.
[“Reinforcement learning allows agents to learn optimal strategies through interaction with an environment.”, “Reinforcement learning allows agents to learn optimal strategies through interaction with an environment.”]
6406534441578.
The Union Budget 2024 proposes increased spending on green energy infrastructure.
6406534441579.
According to WHO, global life expectancy has increased by 6 years since 2000.
6406534441580.
None of these.

A published solution is not available for this question yet.

Report a correction for Q276

Question 277 MSQ · 3.0 marks

What are the key challenges encountered when building a vocabulary for NLP models?
6406534441581.
A larger vocabulary leads to a bigger embedding matrix and increased softmax computation
6406534441582.
Difficulty in handling out-of-vocabulary (OOV) words
6406534441583.
The size of each embedding vector increases with the number of out-of- vocabulary (OOV) words
6406534441584.
More vocabulary entries always guarantee better model generalization

A published solution is not available for this question yet.

Report a correction for Q277

Question 278 MSQ · 3.0 marks

Select all the correct input-output representations used for different pretraining objectives based on the base sentence: **“The quick brown fox jumps over the lazy dog”**
6406534441585.Source diagram or notation
6406534441586.Source diagram or notation
6406534441587.Source diagram or notation
6406534441588.Source diagram or notation

A published solution is not available for this question yet.

Report a correction for Q278

Question 279 NAT · 2.0 marks

Source diagram or notation

A published solution is not available for this question yet.

Report a correction for Q279

Question 280 NAT · 3.0 marks

Source diagram or notation

A published solution is not available for this question yet.

Report a correction for Q280

Question 281 NAT · 2.0 marks

Source diagram or notation
Based on the above data, answer the given subquestions.
Using Byte Pair Encoding (BPE), how many distinct tokens are there in the initial vocabulary before any merges are performed?

A published solution is not available for this question yet.

Report a correction for Q281

Question 282 MCQ · 2.0 marks

Source diagram or notation
Based on the above data, answer the given subquestions.
Which token pair has the lowest frequency before any merges in Byte Pair Encoding?
6406534441592.
(‘b’, ‘r’)
6406534441593.
(‘t’, ‘l’)
6406534441594.
(‘p’, ‘e’)
6406534441595.
(‘e’, ‘r’)
6406534441596.
None of these

A published solution is not available for this question yet.

Report a correction for Q282

Question 283 NAT · 2.0 marks

Source diagram or notation
Based on the above data, answer the given subquestions.
What is the frequency of the most frequent byte-pair before the first merge in Byte Pair Encoding?

A published solution is not available for this question yet.

Report a correction for Q283

Question 284 NAT · 2.0 marks

Source diagram or notation
Based on the above data, answer the given subquestions.
Source diagram or notation

A published solution is not available for this question yet.

Report a correction for Q284