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 · pagegmgl0228_5.png

What was a serious problem in large metropolitan areas in both 1965 and 1967?

OCR text of the page · 983 characters; the scanned image itself is not published
46
43.
Variations Among Population-Density Areas
As probably would be expected, air and water pollution was an
especially serious problem in the large metropolitan areas in
both 1965 and 1967.
Even in farm areas, although less serious than in other areas
in 1967, air and water pollution still is a serious problem
compared to the other problems.
Consider Additional Government Action
" Absolutely Essential" For:
Auto-
Air And
mobile
Mistreat-
Water
Minimum
Safety
Child
ment Of
Pollution
Wages
Features
Welfare
Animals
1967 1965
1967 1965
1967 1965
1967 1965
1967 1965
%
%
%
All Canada
47
46
35
40
34
35
31
32
12
16
Over 450M
54
53
39
51
35
4 4
30
40
14
16
30-450M
45
45
32
38
30
28
29
28
8
14
1-30M
44
47
34
37
41
38
40
37
10
21
Rural non-farm
49
39
40
35
39
28
36
30
15
15
Farm
34
42
25
27
21
33
19
21
11 15
Total urban
48
48
36
43
35 37
32
35
11 16
Total rural
43
40
34 32
32 30
29
26
13 15
(See Table 2B)
Scanned by CamScannerttps://www.industrydocuments.ucsf.edu/docs/gmg10228
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
air and water pollutionAir and water pollution
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

an especially serious problem in the large metropolitan areas in both 1965 and 1967.

84 characters24 tokens

Aux 2015

Wrong

The problem of air and water pollution in large metropolitan areas was a serious problem in both 1965 and 1967.

116 characters34 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong
Air pollution
Air pollution
Air pollution

The same 14-character fragment repeats 21 times until the token limit. Showing the first three.

297 characters64 tokens