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 · pagejqbn0226_33.png

What is the question at line number 25?

OCR text of the page · 1,111 characters; the scanned image itself is not published
Case: 2:13-CV-00170-EAS-EPD Doc #: 133 Filed: 10/06/15 Page: 52 of 307 PAGEID #: 5053
Vol. 12 -
52
A.
Yes .
Study directors are always useful at meetings like
this .
3
Q.
During your personal observations during those two days
of the meetings, did any of the industry representatives
5
dominate the discussions or the voting?
A .
No
7
Q .
Were the industry folks lobbying for higher numbers for
8
the screening level?
A .
No .
10
Q .
Bottom line, Dr. Dourson, did it appear to you that the
11
CATT team process was an unbiased process with an open
12
discussion of the science based on the knowledge at the time?
13
A.
Absolutely.
14
Sir, did anyone coerce you or improperly influence you
15
in any way to arrive at 150 part per billion number?
16
A.
No .
17
Q .
Before I sit down, sir, have you ever testified for me
BT
before?
19
A.
No.
20
Q.
Have you ever testified for anyone from my law firm,
21
Squire Sanders or Squire Patton Boggs before?
22
A.
No.
23
Q.
Have you ever been employed by Dupont?
24
A.
No.
25
Q .
Do you own any stock in DuPont?
Source: https://www.industrydocuments.ucsf.edu/docs/jqbn0226
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
Do you own any stock in Dupont?
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

Do you own any stock in DuPont?

31 characters10 tokens

Aux 2015

Wrong

No

2 characters2 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

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

Question:

What is the question at line number 25?

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

Question:

203 characters64 tokens