Quiz Space

Large Language Models · Quiz 1 · 25 Feb 2024 · January 2024 term

Question 14: Consider a GPT model used for Causal language modelling.…

Question 14

+4 marksNumerical answer

Consider a GPT model used for Causal language modelling. We feed the input sentence “This is a cool idea” to the model by appending special staring [BOS] and ending [EOS] tokens ( that is, “[BOS] This is a cool idea [EOS]”). Assume the context length of the model is 7. The attention matrix computed in one of the attention layers is given below

1000000
0.50.500000
0.30.20.50000
0.10.10.80000
00.1500.60.2500
0.250.3500.20.10.10
0.10.10.10.10.10.40.1

Table 1: Attention score

Based on the above data, answer the given subquestions.

Assume the time step starts from t = 0 and ends at t = 6. Suppose the model is at time step t = 4, what is the attention value assigned for the word “is”?

Show answer

Correct answer: 0

Question 14 of 15 in the IIT Madras BS Large Language Models (LLM) Quiz 1 paper sat on 25 Feb 2024, in the January 2024 term (IIT M DEGREE AN2 EXAM QDB2 25 Feb 2024). It carries 4 marks.

More questions from this paper

  1. Q1Consider a vocabulary \mathcal{V} = (A, T, C, G) and the corresponding embedding matrix E = \begin{bmatrix} 1 & 0 \ 0 &…
  2. Q2Consider a vocabulary \mathcal{V} = (A, T, C, G) and the corresponding embedding matrix E = \begin{bmatrix} 1 & 0 \ 0 &…
  3. Q3Consider a vocabulary \mathcal{V} = (A, T, C, G) and the corresponding embedding matrix E = \begin{bmatrix} 1 & 0 \ 0 &…
  4. Q4Consider a vocabulary \mathcal{V} = (A, T, C, G) and the corresponding embedding matrix E = \begin{bmatrix} 1 & 0 \ 0 &…
  5. Q5Consider the following configuration for the Vannila transformer architecture with one encoder layer and one decoder la…
  6. Q6Consider the following configuration for the Vannila transformer architecture with one encoder layer and one decoder la…
  7. Q7Figure question
  8. Q8Suppose we run the pre-trained GPT model in an autoregressive fashion for generating a text sequence. Then, the stateme…
  9. Q9Suppose we have a dataset for machine translation tasks with thousands of samples. Suppose a team considers training th…
  10. Q10A team trained a Language model using the GPT architecture using a large corpus of text. The model configurations are g…
  11. Q11Choose the correct statements
  12. Q12Figure question
  13. Q13Consider a GPT model used for Causal language modelling. We feed the input sentence “This is a cool idea” to the model …
  14. Q15The statement that “the Next Sentence Prediction (NSP) task requires the BERT model to run autoregressively given the f…