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

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagefzdn0020_7.png

How many people did not answer question 11?

OCR text of the page · 1,336 characters; the scanned image itself is not published
10.
What percentage of your Inside Sales does the cigarette category currently represent?
/a's
0-10%
11-16% 17-25%
36-33%
34-40%
41-50%
# OF
RESPONSES
2
17
7
2
2
NOTE:
# of people who did not answer the question = 1
11.
What percentage of your Total Sales does the cigarette category currently represent?
0-5% 6-10%
11-16% 17-25%
26-33% 34-40%
41-50%
# OF
2
RESPONSES
10
2
NOTE:
# of people who did not answer the question = 8
12.
Is your organization currently in the carton sales business as well as pack?
If no, what is your level of interest in pursing carton sales?
YES: 27
NO:
4 (two of the respondents indicated their level of interest was very little, while ]
indicated they had no interest and the fourth did not indicate.)
13.
If known, what is the average $ ring when cigarettes are purchased?
(Below is a list of the prices given. Each was only mentioned once)
1 .
$3.50
$3.26
IN
$4.33
$1.50
$2.57
$3.00
$5.00
$2.05
Note: The number of people who did not answer this question = 23
14.
If known, what is the average $ ring when cigarettes are purchased, excluding the
value of the cigarettes purchased? (The following answers were each given once)
WN-
$2.50
$1.30
$1.79
$2. 10
50.90
$6.50
Note: The number of people who did not answer this question = 25
52426 6266
Source: https://www.industrydocuments.ucsf.edu/docs/fzdn0020
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
8
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

8

1 characters2 tokens

Aux 2015

Wrong

To answer the question, we need to determine the percentage of people who did not answer question 11.

Looking at the provided text, we see:

  • Question 11: "What percentage of your Total Sales does the cigarette category currently represent?"
  • The text states: "10. What
272 characters64 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

of people who did not answer the question = 1

of people who did not answer the question = 1

of people who did not answer the question = 1

of people who did not answer the question = 1

of people who did not answer the question = 1

of people who

255 characters64 tokens