LLM — da5004_2024T3_ET_AN

Source: da5004_2024T3_ET_AN.pdf · Academic questions: 20 · Separately counted children: 0
Provenance: Large_Language_Models/da5004_2024T3_ET_AN.pdf · SHA-256 d736eeb92b600ca39f3def504e926f3882ea25aebae8a835e51b2682685f9ca6

Scope: LLM (Degree Level : Large Language Models (Computer Based Exam)), physical pages 1 to 10. The in-scope exam content starts on physical page 1 with Subject Confirmation and Question 175 and terminates on physical page 10 after Question 195.

Questions

Shared context 175

Question ID: 6406531046025

Type: General Note
PDF pages: 1

General Note:

If you find the given information is insufficient in any of the Numerical Answer Type (NAT) questions, enter -1 as the answer.
In all your calculations, take at least two digits. For example, if the intermediate result is 1.860875, then take it as 1.86 to the next step.

General Note specifying NAT insufficient information convention (-1) and rounding to at least two digits for intermediate calculations

Figure 1. General Note specifying NAT insufficient information convention (-1) and rounding to at least two digits for intermediate calculations

Question 176

Question ID: 6406531046026

Type: Short Answer Question · Marks: 3
PDF pages: 1, 2

Shared context: ctx-175

The input embeddings for the words “learning”, “brings” and “joy” are \(h_1 = [1.0, 0.5, 1]\), \(h_2 = [1, 0.25, 0]\), and \(h_3 = [0.1, 0.1, 0.9]\), respectively. Note that the embeddings are row vectors. The projection matrices are as follows:

\[W_Q = \begin{bmatrix} 1 & 1 \\ -1 & 1 \\ 0 & 1 \end{bmatrix}, \quad W_K = \begin{bmatrix} 1 & 1 \\ 1 & 0 \\ 0 & 1 \end{bmatrix}, \quad W_V = \begin{bmatrix} 0 & 0 \\ -1 & -1 \\ 1 & 0 \end{bmatrix}\]

The following quantities are computed as:

\[Q = H W_Q, \quad K = H W_K, \quad V = H W_V\]

Let \(e_{ij}\) denote the unnormalized attention score, \(\alpha_{ij}\) denote the normalized attention score (ignore the scaling by \(\sqrt{d_k}\)) and \(z_j\) denote the linear combination of the value vectors for the \(j\)-th word.

Enter the value of first element i.e. with index (0,0) of \(z_1\).

Self-attention query, key, value projection matrices and attention score calculation question

Figure 2. Self-attention query, key, value projection matrices and attention score calculation question

Correct answer: 0.2 to 0.3

Source evidence: [{"page": 2, "bbox": [22.87, 85.0, 180.0, 150.0], "type": "explicit field", "detail": "Possible Answers: 0.2 to 0.3"}]

Question 177

Question ID: 6406531046027

Type: Multiple Choice Question · Marks: 3
PDF pages: 2, 3

Assume that we have a large corpus of text. The vocabulary constructed from the text contains
10000 words. Of these, 100 words occurred only once in the entire corpus of text. The parameters
of the embedding layer and the output layer of the model are shared. Suppose we create a batch
of 256 samples (each sample is a sentence from the corpus). None of these samples contains any
of the 100 rare words. Suppose we pre-train the model for one iteration using the batch of
samples, then:

Options

it is certain that the embeddings of none of these 100 rare words will get
updated.

Source option ID: 6406533531262

there is a chance that the embeddings of all or some of these 100 rare words
will get updated

Source option ID: 6406533531263

the embeddings of all these 100 rare words will defintely get updated

Source option ID: 6406533531264

None of these

Source option ID: 6406533531265

Correct answer: there is a chance that the embeddings of all or some of these 100 rare words
will get updated

Source evidence: [{"page": 2, "bbox": [119.02, 706.74, 131.02, 718.74], "type": "green source option", "detail": "Option ID 6406533531263 marked in green (#008000)"}]

Question 178

Question ID: 6406531046028

Type: Multiple Choice Question · Marks: 3
PDF pages: 3

Suppose we use a pre-trained model for text generation with the given prompt “I am going to”.
Which of the following decoding strategies can be used such that the pre-trained model generates
same text completion each time it is executed

Options

Beam search with beam size 4

Source option ID: 6406533531266

Greedy approach

Source option ID: 6406533531267

Top-K with k = 2

Source option ID: 6406533531268

None of these

Source option ID: 6406533531269

Correct answer: Greedy approach

Source evidence: [{"page": 3, "bbox": [119.02, 230.66, 131.02, 242.66], "type": "green source option", "detail": "Option ID 6406533531267 marked in green (#008000)"}]

Question 179

Question ID: 6406531046036

Type: Multiple Select Question · Marks: 3
PDF pages: 3

Assume that we have 12 sufficiently large supervised NLP datasets for different tasks like
sentiment classification, textual entailment, language understanding and so on. Suppose we take
all the samples (dropping labels) from these datasets to train the GPT model using the CLM
objective.Then which of the following transfer learning approaches is (are) appropriate to transfer
the knowledge to any of these 12 downstream tasks?

Options

Zero shot learning

Source option ID: 6406533531287

Few-shot learning

Source option ID: 6406533531288

Full Fine-tuning

Source option ID: 6406533531289

None of the given approaches are appropriate as the model has already been
trained using samples from all the 12 datasets

Source option ID: 6406533531290

Correct answer: Zero shot learning; Few-shot learning; Full Fine-tuning

Source evidence: [{"page": 3, "bbox": [119.02, 482.12, 131.02, 494.12], "type": "green source option", "detail": "Option ID 6406533531287 marked in green (#008000)"}, {"page": 3, "bbox": [119.02, 502.94, 131.02, 514.94], "type": "green source option", "detail": "Option ID 6406533531288 marked in green (#008000)"}, {"page": 3, "bbox": [119.02, 523.76, 131.02, 535.76], "type": "green source option", "detail": "Option ID 6406533531289 marked in green (#008000)"}]

Question 180

Question ID: 6406531046037

Type: Multiple Choice Question · Marks: 2
PDF pages: 3, 4

What is the order of the language modeling pipeline?

Options

First, the model, which handles text and returns raw predictions.The tokenizer

then makes sense of these predictions and converts them back to text when needed.

Source option ID: 6406533531291

First, the tokenizer, which handles text and returns IDs. The model handles
these IDs and outputs a prediction, which can be some text.

Source option ID: 6406533531292

The tokenizer handles text and returns IDs. The model handles these IDs and
outputs a prediction. The tokenizer can then be used once again to convert these predictions back
to some text.

Source option ID: 6406533531293

None of these

Source option ID: 6406533531294

Correct answer: The tokenizer handles text and returns IDs. The model handles these IDs and
outputs a prediction. The tokenizer can then be used once again to convert these predictions back
to some text.

Source evidence: [{"page": 4, "bbox": [119.02, 85.0, 131.02, 97.0], "type": "green source option", "detail": "Option ID 6406533531293 marked in green (#008000)"}]

Question 181

Question ID: 6406531046041

Type: Multiple Choice Question · Marks: 2
PDF pages: 4

Consider following statment and mark if it is true or false:

Repeating examples in the pre-training datasets are completely harmless for downstream
performance.

Options

TRUE

Source option ID: 6406533531310

FALSE

Source option ID: 6406533531311

Correct answer: FALSE

Source evidence: [{"page": 4, "bbox": [119.02, 331.0, 131.02, 343.0], "type": "green source option", "detail": "Option ID 6406533531311 marked in green (#008000)"}]

Question 182

Question ID: 6406531046042

Type: Multiple Choice Question · Marks: 2
PDF pages: 4

What is the time complexity of the computing attention matrix A with strided local attention
(parameterized by c) and batch size of one? The symbols have the usual meaning.

Options

\(O(cTd)\)

Option A complexity expression O(cTd)

Figure 3. Option A complexity expression O(cTd)

Source option ID: 6406533531312

\(O(cTd^2)\)

Option B complexity expression O(cTd^2)

Figure 4. Option B complexity expression O(cTd^2)

Source option ID: 6406533531313

\(O(c + Td)\)

Option C complexity expression O(c + Td)

Figure 5. Option C complexity expression O(c + Td)

Source option ID: 6406533531314

\(O(T + cd)\)

Option D complexity expression O(T + cd)

Figure 6. Option D complexity expression O(T + cd)

Source option ID: 6406533531315

None of these.

Option E None of these

Figure 7. Option E None of these

Source option ID: 6406533531316

Correct answer: \(O(cTd)\)

Source evidence: [{"page": 4, "bbox": [119.02, 502.94, 131.02, 514.94], "type": "green source option", "detail": "Option ID 6406533531312 marked in green (#008000)"}]

Question 183

Question ID: 6406531046047

Type: Multiple Choice Question · Marks: 2
PDF pages: 4, 5

Suppose we take a pre-trained model GPT-2 model and use it for grammatical acceptance test (based on perplexity score). Say the starting position of the input is shifted \(k\) units to the right where \(k \ge 0\) is some integer, i.e. if the index of a token in the input is 1 (0-based indexing) then its new position is \(1 + k\). In general token at \(i\) index is shifted to \(i + k\).

Choose the correct statement:

Pre-trained GPT-2 model grammatical acceptance test and input position shifting question

Figure 8. Pre-trained GPT-2 model grammatical acceptance test and input position shifting question

Options

k = 0 will perform same as compared to k = 600

Source option ID: 6406533531320

k = 0 will perform worse compared to k = 600

Source option ID: 6406533531321

k = 0 will perform better compared to k = 600

Source option ID: 6406533531322

Value of k is irrelevant as the difference between position embeddings of two
consicutive tokens remain the same.

Source option ID: 6406533531323

Correct answer: k = 0 will perform better compared to k = 600

Source evidence: [{"page": 5, "bbox": [119.02, 216.34, 131.02, 228.34], "type": "green source option", "detail": "Option ID 6406533531322 marked in green (#008000)"}]

Question 184

Question ID: 6406531046048

Type: Multiple Choice Question · Marks: 2
PDF pages: 5

Which of the following is correct expression for \(q_{ij}\)? The symbols have usual meaning.

Attention score q_ij expression question

Figure 9. Attention score q_ij expression question

Options

\(x_i^T W_Q W_K^T x_j + x_i^T W_Q W_K^T p_j + p_i^T W_Q W_K^T x_j + p_i^T W_Q W_K^T p_j\)

Option A expanded attention score expression with positive cross terms

Figure 10. Option A expanded attention score expression with positive cross terms

Source option ID: 6406533531324

\(x_i^T W_Q W_K^T x_j - x_i^T W_Q W_K^T p_j + p_i^T W_Q W_K^T x_j - p_i^T W_Q W_K^T p_j\)

Option B attention score expression with negative cross terms

Figure 11. Option B attention score expression with negative cross terms

Source option ID: 6406533531325

\(x_i^T W_K W_Q^T x_j + x_i^T W_K W_Q^T p_j + p_i^T W_K W_Q^T x_j + p_i^T W_K W_Q^T p_j\)

Option C attention score expression with z_i and z_j terms

Figure 12. Option C attention score expression with z_i and z_j terms

Source option ID: 6406533531326

\(x_i W_Q W_K x_j^T + x_i W_Q W_K p_j^T + p_i W_Q W_K x_j^T + p_i W_Q W_K p_j^T\)

Option D attention score expression with transposed weight variants

Figure 13. Option D attention score expression with transposed weight variants

Source option ID: 6406533531327

None of these

Option E None of these

Figure 14. Option E None of these

Source option ID: 6406533531328

Correct answer: \(x_i^T W_Q W_K^T x_j + x_i^T W_Q W_K^T p_j + p_i^T W_Q W_K^T x_j + p_i^T W_Q W_K^T p_j\)

Source evidence: [{"page": 5, "bbox": [119.02, 435.87, 131.02, 447.87], "type": "green source option", "detail": "Option ID 6406533531324 marked in green (#008000)"}]

Shared context 185-189

Question ID: 6406531046030

Type: Comprehension
PDF pages: 5, 6

Consider the following dictionary with the number of word occurrences in a corpus:

\[w_0 = \{\text{“blast”} : 2, \; \text{“bitterest”} : 1, \; \text{“we”} : 4\}\]

Note: Append to each word at the end.
You will be working with word piece algorithm, answer the given sub questions in that context:

WordPiece algorithm vocabulary dictionary with word counts and end-of-word marker instructions

Figure 15. WordPiece algorithm vocabulary dictionary with word counts and end-of-word marker instructions

Question 185

Question ID: 6406531046031

Type: Short Answer Question · Marks: 2
PDF pages: 6

Shared context: ctx-185-189

How many tokens are there in the initial vocabulary?

Correct answer: 10

Source evidence: [{"page": 6, "bbox": [22.87, 440.0, 150.0, 500.0], "type": "explicit field", "detail": "Possible Answers: 10"}]

Question 186

Question ID: 6406531046032

Type: Multiple Choice Question · Marks: 3
PDF pages: 6

Shared context: ctx-185-189

Which of the following pairs has the least score before any merge?

Options

('e', '')

Option A pair ('e', '</w>')

Figure 16. Option A pair ('e', '')

Source option ID: 6406533531275

('s', 't')

Option B pair ('s', 't')

Figure 17. Option B pair ('s', 't')

Source option ID: 6406533531276

('o', 'b')

Option C pair ('o', 'b')

Figure 18. Option C pair ('o', 'b')

Source option ID: 6406533531277

('l', 'e')

Option D pair ('l', 'e')

Figure 19. Option D pair ('l', 'e')

Source option ID: 6406533531278

('w', 'e')

Option E pair ('w', 'e')

Figure 20. Option E pair ('w', 'e')

Source option ID: 6406533531279

Correct answer: ('e', '')

Source evidence: [{"page": 6, "bbox": [119.02, 577.62, 131.02, 589.62], "type": "green source option", "detail": "Option ID 6406533531275 marked in green (#008000)"}]

Question 187

Question ID: 6406531046033

Type: Short Answer Question · Marks: 3
PDF pages: 7

Shared context: ctx-185-189

Which pair will be merged in the very first merge? Say the pair is (‘a’,‘b’), then enter “ab” (without
quotes and white spaces). If there is a tie between two or more candidates, pick the one that
occurs first in the original vocabulary.

NOTE: Enter the exact answer without any space in the beginning or at the end.

Correct answer: la

Source evidence: [{"page": 7, "bbox": [22.87, 430.0, 150.0, 480.0], "type": "explicit field", "detail": "Possible Answers: la"}]

Question 188

Question ID: 6406531046034

Type: Short Answer Question · Marks: 3
PDF pages: 7

Shared context: ctx-185-189

What is the score of the pair merged in the very first merge?

Correct answer: 1.0

Source evidence: [{"page": 7, "bbox": [22.87, 560.0, 150.0, 610.0], "type": "explicit field", "detail": "Possible Answers: 1.0"}]

Question 189

Question ID: 6406531046035

Type: Multiple Choice Question · Marks: 3
PDF pages: 7, 8

Shared context: ctx-185-189

Assuming that the algorithm was stopped after one merge only, how will the word “last” be tokenized with the given vocabulary?

Please note that is already appended to the word.

Tokenization of the word last</w> after one merge with the given vocabulary

Figure 21. Tokenization of the word last after one merge with the given vocabulary

Options

l, as, t,

Option A tokenization: l, as, t, </w>

Figure 22. Option A tokenization: l, as, t,

Source option ID: 6406533531282

la, s, t,

Option B tokenization: la, s, t, </w>

Figure 23. Option B tokenization: la, s, t,

Source option ID: 6406533531283

l, a, s, t,

Option C tokenization: l, a, s, t, </w>

Figure 24. Option C tokenization: l, a, s, t,

Source option ID: 6406533531284

l, a, st,

Option D tokenization: l, a, st, </w>

Figure 25. Option D tokenization: l, a, st,

Source option ID: 6406533531285

None of these

Option E None of these

Figure 26. Option E None of these

Source option ID: 6406533531286

Correct answer: la, s, t,

Source evidence: [{"page": 8, "bbox": [119.02, 44.32, 131.02, 56.32], "type": "green source option", "detail": "Option ID 6406533531283 marked in green (#008000)"}]

Question 190

Question ID: 6406531046029

Type: Multiple Select Question · Marks: 2
PDF pages: 8

Consider the following statements about decreasing the size of the vocabulary

Options

It increases number of tokens required to represent a sentence

Source option ID: 6406533531270

It Increases the computational complexity computing attention matrix

Source option ID: 6406533531271

It increases the context length T

Source option ID: 6406533531272

None of these

Source option ID: 6406533531273

Correct answer: It increases number of tokens required to represent a sentence; It Increases the computational complexity computing attention matrix; It increases the context length T

Source evidence: [{"page": 8, "bbox": [119.02, 391.31, 131.02, 403.31], "type": "green source option", "detail": "Option ID 6406533531270 marked in green (#008000)"}, {"page": 8, "bbox": [119.02, 412.13, 131.02, 424.13], "type": "green source option", "detail": "Option ID 6406533531271 marked in green (#008000)"}, {"page": 8, "bbox": [119.02, 432.95, 131.02, 444.95], "type": "green source option", "detail": "Option ID 6406533531272 marked in green (#008000)"}]

Question 191

Question ID: 6406531046038

Type: Multiple Select Question · Marks: 2
PDF pages: 8

Choose the correct statements regarding an encoder-decoder transformer model:

Options

Multi head masked attention block in decoder makes use of a mask.

Source option ID: 6406533531295

Transformer model is an improvement over its RNN based predecessor for
machine translation.

Source option ID: 6406533531296

For machine tranlation task the input and output embedding layers have to be
identical.

Source option ID: 6406533531297

None of these.

Source option ID: 6406533531298

Correct answer: Multi head masked attention block in decoder makes use of a mask.; Transformer model is an improvement over its RNN based predecessor for
machine translation.

Source evidence: [{"page": 8, "bbox": [119.02, 595.01, 131.02, 607.01], "type": "green source option", "detail": "Option ID 6406533531295 marked in green (#008000)"}, {"page": 8, "bbox": [119.02, 615.83, 131.02, 627.83], "type": "green source option", "detail": "Option ID 6406533531296 marked in green (#008000)"}]

Question 192

Question ID: 6406531046039

Type: Multiple Select Question · Marks: 2
PDF pages: 8, 9

Consider C4 pipeline. Which of the following will NOT pass through it as it is?

Options

ir a la escuela es un buen habito.

(Above sentence is spanish translation of “Going to school is a good habit”)

Option A Spanish sentence ir a la escuela es un buen habito

Figure 27. Option A Spanish sentence ir a la escuela es un buen habito

Source option ID: 6406533531299

Following function computes n — th Fibonacci number:

def fibo(n):
if n <= 1:
return n
return fibo(n - 1) + fibo(n - 2)

Option B Python code snippet computing n-th Fibonacci number

Figure 28. Option B Python code snippet computing n-th Fibonacci number

Source option ID: 6406533531300

Lorem ipsum dolor sit amet.

Option C Latin placeholder text Lorem ipsum dolor sit amet

Figure 29. Option C Latin placeholder text Lorem ipsum dolor sit amet

Source option ID: 6406533531301

Please enable JavaScript to use our site.
Home
Products
Shipping
Contact
FAQ

Option D Website navigation placeholder text Please enable JavaScript to use our site

Figure 30. Option D Website navigation placeholder text Please enable JavaScript to use our site

Source option ID: 6406533531302

D Gukesh defeated defending champion Ding Liren at the World Chess Championship. D Gukesh defeated defending champion Ding Liren at the World Chess Championship.

Option E Duplicate text sentences regarding D Gukesh defeating Ding Liren

Figure 31. Option E Duplicate text sentences regarding D Gukesh defeating Ding Liren

Source option ID: 6406533531303

Sardar Vallabhbhai Patel was the first deputy prime minister of independent India.

Option F Sentence regarding Sardar Vallabhbhai Patel

Figure 32. Option F Sentence regarding Sardar Vallabhbhai Patel

Source option ID: 6406533531304

None of these.

Option G None of these

Figure 33. Option G None of these

Source option ID: 6406533531305

Correct answer: ir a la escuela es un buen habito. (Above sentence is spanish translation of “Going to school is a good habit”); Following function computes n — th Fibonacci number: def fibo(n): if n <= 1: return n return fibo(n - 1) + fibo(n - 2); Lorem ipsum dolor sit amet.; Please enable JavaScript to use our site. Home Products Shipping Contact FAQ; D Gukesh defeated defending champion Ding Liren at the World Chess Championship. D Gukesh defeated defending champion Ding Liren at the World Chess Championship.

Source evidence: [{"page": 9, "bbox": [119.02, 81.07, 131.02, 93.07], "type": "green source option", "detail": "Option ID 6406533531299 marked in green (#008000)"}, {"page": 9, "bbox": [119.02, 215.64, 131.02, 227.64], "type": "green source option", "detail": "Option ID 6406533531300 marked in green (#008000)"}, {"page": 9, "bbox": [119.02, 257.22, 131.02, 269.22], "type": "green source option", "detail": "Option ID 6406533531301 marked in green (#008000)"}, {"page": 9, "bbox": [119.02, 403.04, 131.02, 415.04], "type": "green source option", "detail": "Option ID 6406533531302 marked in green (#008000)"}, {"page": 9, "bbox": [119.02, 482.12, 131.02, 494.12], "type": "green source option", "detail": "Option ID 6406533531303 marked in green (#008000)"}]

Question 193

Question ID: 6406531046040

Type: Multiple Select Question · Marks: 2
PDF pages: 9

Which of the following are correct about GPT-1 model:

Options

It is a decoder only model.

Source option ID: 6406533531306

It uses absolute positional encoding.

Source option ID: 6406533531307

It uses a composite dataset for pre-training.

Source option ID: 6406533531308

None of these.

Source option ID: 6406533531309

Correct answer: It is a decoder only model.; It uses absolute positional encoding.

Source evidence: [{"page": 9, "bbox": [119.02, 736.32, 131.02, 748.32], "type": "green source option", "detail": "Option ID 6406533531306 marked in green (#008000)"}, {"page": 9, "bbox": [119.02, 757.14, 131.02, 769.14], "type": "green source option", "detail": "Option ID 6406533531307 marked in green (#008000)"}]

Question 194

Question ID: 6406531046043

Type: Short Answer Question · Marks: 2
PDF pages: 10

How many entries of unnormalized attention matrix QKT will be calculated with strided attention
mechanism with c = 5. The sequence length is T = 32

Correct answer: 164

Source evidence: [{"page": 10, "bbox": [22.87, 280.0, 150.0, 330.0], "type": "explicit field", "detail": "Possible Answers: 164"}]

Shared context 195-196

Question ID: 6406531046044

Type: Comprehension
PDF pages: 10

Assume that you are using block attention model. Sequence length \(T = 32\), number of blocks \(n = 4\), identity permutation for \(Q\) matrix and \(\pi = (2,0,3,1)\) for \(K^T\) matrix. The mask matrix is denoted by \(M\).

Answer the given sub question:

Block attention model configuration specifications with permutation pi = (2,0,3,1)

Figure 34. Block attention model configuration specifications with permutation pi = (2,0,3,1)

Question 195

Question ID: 6406531046045

Type: Short Answer Question · Marks: 2
PDF pages: 10

Shared context: ctx-195-196

How many non zero entries will be there in the mask matrix M?

Correct answer: 256

Source evidence: [{"page": 10, "bbox": [22.87, 770.0, 150.0, 820.0], "type": "explicit field", "detail": "Possible Answers: 256"}]

Answer key

Source question	Source identity	Correct answer	PDF pages
176	6406531046026	0.2 to 0.3	1, 2
177	6406531046027	there is a chance that the embeddings of all or some of these 100 rare words will get updated	2, 3
178	6406531046028	Greedy approach	3
179	6406531046036	Zero shot learning; Few-shot learning; Full Fine-tuning	3
180	6406531046037	The tokenizer handles text and returns IDs. The model handles these IDs and outputs a prediction. The tokenizer can then be used once again to convert these predictions back to some text.	3, 4
181	6406531046041	FALSE	4
182	6406531046042	\(O(cTd)\)	4
183	6406531046047	k = 0 will perform better compared to k = 600	4, 5
184	6406531046048	\(x_i^T W_Q W_K^T x_j + x_i^T W_Q W_K^T p_j + p_i^T W_Q W_K^T x_j + p_i^T W_Q W_K^T p_j\)	5
185	6406531046031	10	6
186	6406531046032	('e', '')	6
187	6406531046033	la	7
188	6406531046034	1.0	7
189	6406531046035	la, s, t,	7, 8
190	6406531046029	It increases number of tokens required to represent a sentence; It Increases the computational complexity computing attention matrix; It increases the context length T	8
191	6406531046038	Multi head masked attention block in decoder makes use of a mask.; Transformer model is an improvement over its RNN based predecessor for machine translation.	8
192	6406531046039	ir a la escuela es un buen habito. (Above sentence is spanish translation of “Going to school is a good habit”); Following function computes n — th Fibonacci number: def fibo(n): if n <= 1: return n return fibo(n - 1) + fibo(n - 2); Lorem ipsum dolor sit amet.; Please enable JavaScript to use our site. Home Products Shipping Contact FAQ; D Gukesh defeated defending champion Ding Liren at the World Chess Championship. D Gukesh defeated defending champion Ding Liren at the World Chess Championship.	8, 9
193	6406531046040	It is a decoder only model.; It uses absolute positional encoding.	9
194	6406531046043	164	10
195	6406531046045	256	10

Extraction notes

Status: verified-source-limitation
Count reconciliation: The portal header does not declare a Section Marks or Number of Questions table. The source PDF contains 1 administrative subject confirmation entry (Page 1, 0 marks), 1 shared-context calculation note entry (ctx-175, 0 marks), 2 comprehension passages (ctx-185-189, ctx-195-196), and 20 academic questions (Q176-Q195) totaling 48 marks across 8 portal sub-sections. The second comprehension block on page 10 (Question Id 6406531046044) declares Question Numbers (195 to 196), but the PDF terminates on page 10 after Question 195; Question 196 is absent from the source. The observed academic count of 20 questions reconciles all detected question entries in the source PDF without invented reconciliation.

Extracted all 20 academic questions (Q176-Q195) and 3 shared context entries (ctx-175, ctx-185-189, ctx-195-196) from the 10-page source PDF.
Preserved exact raster images and verified transcriptions for all mathematical formulas and diagrams.
Comprehension container Question Id : 6406531046044 on page 10 declares Question Numbers : (195 to 196), but physical page 10 terminates the document after Question 195. Question 196 is absent/truncated in the source PDF.
Question 175 is an administrative general note container carrying reference NAT instructions (insufficient info -1, two-digit rounding) and 0 marks with administrative acknowledgment options; preserved calculation instructions as shared context ctx-175 while excluding acknowledgment options.
No top-level portal section header table (Section Marks, Number of Questions) is present on page 1; exam begins directly with Subject Confirmation question.
Short answer Question 176 provides an accepted answer range 0.2 to 0.3.

Excluded source entries:

{"classification": "subject-confirmation", "page": 1, "bbox": [22.87, 55.39, 569.03, 246.16], "reason": "Administrative subject confirmation asking candidate to confirm subject and cross check hall ticket", "identity": "confirmation"}
{"classification": "administrative-acknowledgment-options", "page": 1, "bbox": [23.62, 280.0, 560.0, 360.0], "reason": "Administrative acknowledgment options (Instructions has been mentioned above / This Instructions is just for a reference & not for an evaluation) excluded semantically while preserving calculation note as shared context", "identity": "175"}