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 · pagejtlh0227_10.png

What does TK-RK denote?

OCR text of the page · 1,890 characters; the scanned image itself is not published
a) TK-RK - Temik applied with RUSS-KEN incorporator
b) 0 Not significantly different from check
TK-1 -
Temik power incorporated
1 Significantly better than check
TK-D -
Temik Side dressed
2 Significantly better than fumigated
TK-S-
Temik scratched in
3 Significantly better than incorporated
FUM -
Fumigant
4 Significantly better than RUSS-KEN (Scratch in Neb.)
CHK -
Untreated check
5 Significantly better than Disc (side dress)
Method of
Method of
Location
application a
T/A
Rating b
Location
application
a
T/A
Rating
b
Prospect Valley, Colo.
TK-RK
15.97
Scottsbluff, Nebraska
TK -S
22.11
TK-1
15.71
TK-1
24.02
TK-D
17.29
TK-D
23.23
FUM
17.89
0.3
FUM
22.37
O - -
CHK
16.83
CHK
20.19
Sterling, Colorado
TK-RK
19.34
0
Scottsbluff, Nebraska
TK-S
28.31
TK-1
22.44
15
TK-I
28.64
TK -D
18.55
0
TK-D
30.29
1,2
FUM
20.39
FUM
28.58
1
CHK
17.16
CHK
23.56
Scottsbluff, Nebraska
TK-S
25.61
0,5
Stegall, Nebraska
TK-S
20.20
TK-1
25.94
0.5
TK-1
18.94
TK-D
22. 18
TK-D
17.82
0.4
FUM
25.61
0.5
FUM
15.70
O !
CHK
24.09
CHK
17.42
Greeley, Colorado
TK-RK
17.89
Berthoud, Colorado
TK-RK
17.75
TK -1
18.94
TK-1
18.94
TK -D
20.86
TK -D
17.82
DO
0,4
FUM
20.39
O
FUM
15.70
O
CHK
19.67
CHK
17.42
Prospect Valley
TK-RK
17.75
0
Melbeta, Nebraska
TK-S
29.30
TK-1
22 . 18
0,4
TK-1
26.66
TK-D
22.44
0.4
TK-D
24.75
FUM
20.46
0
FUM
30.03
15
CHK
21.58
CHK
26.07
Sterling, Colorado
TK-RK
16.30
O
Minatare, Nebraska
TK-I
25.34
TK-1
14.06
TK -S
26.80
TD-D
18.55
TK-D
26.66
O
0
FUM
19.27
1,3,4
FUM
28.78
0,3
CHK
15.58
CHK
30.36
3.4,5
Scottsbluff, Nebraska
TK -S
28.31
1,3,5
Platteville, Colorado
TK-1
26.73
TK- 1
23.17
0
TK-RK
26.53
TK-D
24.75
TR-D
28.12
FUM
26.80
13
FUM
28.97
- -
CHK
22.84
CHK
18.61
Scottsbluff, Nebraska
TK -S
30.23
0
Minatare, Nebraska
TK-S
29.57
TK-1
31.68
1,2
TK-1
26.40
15
TK-D
29.70
TK-D
23.29
FUM
28.31
FUM
29.04
1.5
CHK
27.32
CHK
23.03
9
Source: https://www.industrydocuments.ucsf.edu/docs/jtlh0227
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
Temik applied with RUSS-KEN incorporator
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

Temik applied with RUSS-KEN incorporator

40 characters11 tokens

Aux 2015

Wrong

The question asks for the abbreviation TK-RK (Temik applied with RUSS-KEN incorporator).

Looking at the OCR text:

  • "a) TK-RK - Temik applied with RUSS-KEN incorporator"
  • "b) 0 Not significantly different from check"
224 characters64 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

TK -1 TK -1 TK -1 TK -1 TK -1 TK -1 TK -1 TK -1 TK -1 TK -1 TK -1 TK -1 TK -1

77 characters64 tokens