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 · pageyrmg0227_5.png

What is the largest quantity in the list?

OCR text of the page · 1,018 characters; the scanned image itself is not published
quantity
13 Ti prescription
2 Rud (12) 4-B0
Unit Price.
Total
T-14 table 26 x 26 x 24" h.
9 Reed
321-14
30. 30
333.30
128-1
342-1
5A -1
323-1
421-1-
422-1
5 44-1
539-2
417-1
X 14
C-185 Arm chair
120-2
39.40
551.60
354-16
306-2V
346-10
352-16
348-16
418-2V
307-1V
X ,
347-10
330-2"
235 - ?
C-143 side chair
120-2
20.30
142.10
355-1
347 -1V
345-1x
348-16
352-1
X
?
#14061=52, 50 x 30"
G8 -1
346-1/
354-1
103.30
309.90
#14031=32 55 x 30" desk
125-1
106.00
106.00
143280, 60 x 19 3/4"
131-1
124.50
124.50
X
#14328D, 60 x 19 3/4"
125-1-
124.50
124.50
#1598D telephone cabinet
127-1?
234-1.
343-1
419-1v
56.40
225.60
?
#16024, table 60 x 30
125-1
531-1
532-1
73.35
220.05
X ,
#315 cabinet, storage
125-1V
126-1V
531-10
8.00
288.00
C-177 posture chair
54 -17?
418-17
37.70
678.60
C--35 side chair
531-26-
532-26
110.40
13 Tax
124 45 x 19 x 15" h
? < 417-1
36.40
36. 40
1822, file for 4 x 6" cards
54 -1
144.45
144.45
#14071, 45 x 30" desk
54 -1 7
98.75
98.75
1414, table % x 38
5A -1
165.00
165.00
ents.ucsf.edu/docs/yrmg0227
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
18
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

531-26-

7 characters8 tokens

Aux 2015

Wrong

quantity

8 characters2 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

quantity 13 Ti prescription 2 Rud (12) 4-B0 Unit Price. Total T-14 table 26 x 26 x 24" h. 9 Reed 321-14 30.30 333.30 328-1 341-1 420-1 5

136 characters64 tokens