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 · pagekjgf0227_4.png

What is the page number?

OCR text of the page · 886 characters; the scanned image itself is not published
4
6. Hypervitaminosis D
C. Woodruff
July, 1960
7. Calcium and Phosphorus
D. Watkin
Sept. , 1960
in the Diet
and
G. Whedon
8.
Dietetic Use of Sorbitol
W. Olmsted
Sept. , 1960
9 .
Nutrition Surveys and
W. Darby
Fall, 1960
Recommended Dietary
Allowances
10. Function of FDA in
Regard to Medicine
P. Day
11. Addition of Vitamin A
0. C. Johnson
to Skim Milk
12. Addition of Milk Solids
0. C. Johnson
to Skim Milk
13. Multivitaminized Milk
0. C. Johnson
14. Emulsifiers in Foods
0. C. Johnson
15. Antioxidants in Food
0. C. Johnson
16. Nutrition Consequences of
0. C. Johnson
the "Continuous" Bread
and
Making Process
D. B. Hand
17. Vitamin E
M. Horwitt
18. Editorial on Folic Acid,
W. Darby
Vitamin B6, Pantothenic
Acid and Vitamin B12 in
Human Dietaries
Articles to be Rewritten
1.
Use of Aluminum
Cooking Utensils
2 .
Importance of Amino Acid
Harper
Balance in Nutrition
.edu/docs/kjgi0227
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
4- 4 -
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

4

1 characters2 tokens

Aux 2015

Exact match

4

1 characters2 tokens

PiT-FT 2015

Wrong

<|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|>

120 characters64 tokens

ChronoGPT 2015

Wrong

OCR text: 4 6. Hypervitaminosis D C. Woodruff July, 1960 7. Calcium and Phosphorus D. Watkin Sept. , 1960

Question:

What is the page number?

OCR text:

157 characters64 tokens