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 · pagemqbf0001_5.png

How many calls is possible to Banner ?

OCR text of the page · 1,163 characters; the scanned image itself is not published
RJR Marketing Merchandising Visit
Tentative Agenda
Date
Time
Event
Action
Tues. Feb. 4
7:45 A.M.
Arrive (Sydney) Airport
Meet and transport to hotel (limo or two
RPM
cars
1:00 P.M.
Pick up and go to Northpoint for lunch
RPM
boardroom)
2:00 P.M.
Commence Presentation
intro (market profile)
MMc
.
AK
. .
pa issues
legal issues
DG
marketing (specific issues)
MMc
sales execution issues
DMc
5:00 P.M.
Finish
RPM
Transport back to hotel via rocks!
botanical gardens, etc.
6:30 P.M.
Low key dinner (at hotel?)
RJR
Wed. Feb. 5
8:00 A.M.
Collect from hotel and to Granville
RPM
9:00 A.M.
.
POS presentation NSW and other
DW
areas (show and tell)
Individual marketing discussions
Camel
CB
- Other (Savings)
LR
- POS
DW
12:30 P.M.
Lunch (in conference room)
RJR/RPM
1:00 P.M.
Start field visit at Parramatta and work
towards city
(Break into two groups)
#1 2 RJR (brand)
3 RPM (Mkt/Rep)
#2 3 RJR (merchandising)
2 RPM (Des/Rep)
Supermarkets (4 calls if possible)
RJR and
Woolworths
nominated
Coles
RPM staff
Franklin
Jewel
Banner (2 calls if possible)
Foodtown
Festival
52920 0604
RJR0000000417073920
70032 0730
Source: https://www.industrydocuments.ucsf.edu/docs/mqbf0001
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
(2 calls if possible)2 calls2 calls if possible
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

2 calls if possible

19 characters5 tokens

Aux 2015

Wrong

2

1 characters2 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

RJR/RPM

Action:

RJR/RPM

Market profile

Low key dinner (at hotel?)

RJR/RPM

Low key dinner (at hotel?)

RJR/RPM

Low key dinner (at hotel?)

160 characters64 tokens