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 · pagehslf0227_1.png

What is the weight % of a Tub of 18:1 margarine?

OCR text of the page · 433 characters; the scanned image itself is not published
ATTACHMENT
MARGARINE
(Weight "%)
Stick
Stick
Tub
16:0
11
11
11
18:0
8
6
7
18:1
57
48
37-47
18:2
21
32
31-44
18:3
1
2
cis, cis
11
29
23-44
Saturated
19
17
18
P/S
0. 6
1 . 7
1. 3-2. 5
COMPOSITION OF MARGARINES
Older
Current
Type
Current
Stick
Tub
Saturated
22
20
13
Monounsaturated
63
50
23
Cis
28
Trans
35
25
25
17
6
Polyunsaturated
15
30
64
Cis, cis
Cis, trans
27
61
3
3
Source: https:7/www.industrydocuments. ucsi. edu/docs/nsii0227
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
37-47
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

37-47

5 characters6 tokens

Aux 2015

Wrong

18:1

4 characters5 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER OCTOBER

103 characters64 tokens