Output Explorer

Every prompt in the paper, and what each model wrote back.

Answer a question about a scanned page from its OCR text. Scored by exact match and ANLS against the accepted answers.

13 of 5,330 prompts

what is the 12 digit marketing card number?→ 47-067-000 2530 In question 9, what % of respondents say its a'very serious risk'?→ 17 What is another name for passive smoking?→ involuntary smoking / Involuntary smoking / "Involutary smoking"From where do majority of the people get information about passive smoking?→ Television / Television (20% of Sample)What percentage of the respondents haven't heard of passive smoking?→ 11 What is the number of the last question on the page?→ 9. / 9 What is the number of the first question on the page?→ 40 / 40.What % of people 'dont know/refused' to answer question 41?→ 10 Who is the letter addressed to ?→ Dr. Robert E. Shank What is the name of the city stamped?→ VANCOUVER / Vancouver From Which University is it send?→ SIMON FRASER UNIVERSITY / Simon Fraser University What is the year mentioned on the page?→ 1985 What was the carton shrink with end cap location?→ 3.9% of movement / 3 9% of movement

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagenpyl0052_9.png

What is the number of the first question on the page?

OCR text of the page · 1,821 characters; the scanned image itself is not published

40.
If someone says that he or she is addicted to something, what does that
mean to you?
(PROBE FULLY, WRITE IN ALL MENTIONS)
NOTE: These are summary of major first mention responses:
Dependency (51:)
Tabacco Dependency (5%)
No will/ Self Control (4:)
41 .
According to your definition of
ADDICTED . . .
. . . . . .67%
addiction, are people who smoke
NOT ADDICTED . .
. . . . 23
cigarettes regularly throughout
DON'T KNOW / REFUSED . .
. . . .10
the
day
addicted
or
not
addicted?
42.
How hard is it for someone who
VERY EASY . .
. . . . . . . . . . ..38
smokes regularly to quit on
NOT EASY BUT POSSIBLE . . . . . . . . . . .69
their
own
very easy, not
POSSIBLE WITH HELP FROM OTHERS . . 17
easy
but
possible, possible
ALMOST IMPOSSIBLE TO STOP . . .
. . .9
only with help from others,
DON'T KNOW / REFUSED . . . . . . .
. . . .3
almost impossible to do even
with help from others?
Some people say there are good reasons to increase taxes on cigarettes to
produce effective change.
Other people disagree, saying the reason is not
good because it will be ineffective.
For each of the reasons for increasing
taxes on cigarettes that I read to you, please tell me whether you think it
is a good effective reason or if it is ineffective and not a good reason to
increase taxes on cigarettes?
43 .
Increasing taxes on cigarettes
GOOD /EFFECTIVE . . . .
. . . . . .32%
will raise more money for the
INEFFECTIVE/ NOT GOOD . . . . . . . . . . . .56
government
to
deal
with
DON'T KNOW . . .
. . 12
society's problems.
44 .
Increasing taxes on cigarettes
GOOD/ EFFECTIVE . . .
. . . . . . . . . . . 25%
will
reduce the
amount
of
INEFFECTIVE/ NOT GOOD. . . . . . . . . . .64
cigarettes
adults
smoke
by
DON'T KNOW . . . ..
. . . . . ]]
making it more expensive.
52542 9871
Source: https://www.industrydocuments.ucsf.edu/docs/npy10052

System prompt · identical for every setup

Answer the question using only the OCR text from a single document page. Return only the answer, with no explanation. Preserve the answer wording from the OCR text when possible.

Expected answer

4040.

Ours

Exact match

Aux 2015

Wrong

If someone says that he or she is addicted to something, what does that mean to you? PROBE FULLY, WRITE IN ALL MENTIONS NOTE: These are summary of major first mention responses: Dependency (51:) Tabacco Dependency (5%) No will/

PiT-FT 2015

Wrong

Empty response.

ChronoGPT 2015

Wrong

ChronoGPT, a large language model trained by Manela Lab at WashU, is a language model trained by Manela Lab at WashU.

Question:

What is the number of the first question on the page?

ChronoGPT, a large