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 · pagenrmg0227_13.png

What is the metal in room 544?

OCR text of the page · 465 characters; the scanned image itself is not published
ROOM SCHEDULE (Cont'd)
WALL
ROOM
TYPES
EQUIPMENT
REMARKS
METAL
TEXTOLITE
VINYL
Bases &
Desk
Uphol.
Chair Frames
Table Tops
539
15
Miller
544
3
Blue
Miller
545
1Ox
B.T.
WH
BLA
547
2x
Blue
WH
BLA
548
2x
Blue
BLA
549
2x
B.T.
BLA
550
2x
B. T.
BLA
551
2x
Blue
BLA
552
2x
Blue
WH
BLA
554
2x
B.T.
BLA
555
2x
B.T.
BLA
556
2x
B . T.
BLA
557
2x
B.T.
BLA
558
2x
B.T.
BLA
559
2x
B .T.
WH
BLA
561
7
B.T.
BLA
5 A
10
B.T.
WAL
PAR
-12-
/www.industrydocuments.ucsf.edu/docs/nrmg0227
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
Blue
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

Miller

6 characters2 tokens

Aux 2015

Wrong

B.T.

4 characters4 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

ROOM SCHEDULE (Cont'd) WALL ROOM TYPES EQUIPMENT REMARKS Metal Textolite VINYL Bases & Desk 539 15 Miller 544 3 Blue Blue BLA 555

129 characters64 tokens