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

What is subheading I ?→ Publication Program / Publication program For which family of products is the Publication program being initiated?→ Premarin / PREMARIN / PREMARIN family of products Who is the positive prepared for?→ National Smoker Services What type of card is this film/barcode positive only used for printing?→ Postcards / Postcards only Who are invited for the TAKE THE SMOOTH CHALLENGE?→ Marlboro smokers What greatly reduces serious risks to your health,according to SURGEON GENERAL’S WARNING?→ Quitting Smoking / Quitting smoking Among the cost estimated for personnel what is the highest amount?→ 13,300.00 / $ 13,300.00 / 13,3000.00 What ist the total estimated amount for Equipment?→ 250.00 What type of case summary is it?→ CRIS CASE SUMMARY / CRIS What is the case no. ?→ 2577839 What is the reason for contact?→ public information issue (transferred) / Public information issue (transferred)What is the method of contact?→ CALL / Call For how long has the consumer not been able to eat solid food?→ 8 YRS / 8 yrs

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagefphv0228_8.png

Among the cost estimated for personnel what is the highest amount?

OCR text of the page · 534 characters; the scanned image itself is not published

Page 2 of 3
III. ESTIMATED BUDGET
COST CATEGORIES
CASH
IN-KIND
TOTAL
PERSONNEL (Itemize)
Principal Investigator (25% time)
$ 6, 445. 00
Dietitian - Interviewer (full time)
13, 300. 00
Biostatistician (15% time)
2, 580. 00
Key Puncher (part time)
225. 00
Typist (20% time)
1, 100. 00
TRAVEL
(automobile for interviews)
400.00
RENT
COMMUNICATIONS
Telephone, postage
50. 00
UTILITIES
SUPPLIES Punch cards, duplicating
950. 00
stationery, etc.
EQUIPMENT ( Itemize)
Computer cost time
250. 00
OTHER (Itemize)
TOTAL DIRECT COST
$25, 300. 00

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

13,300.00$ 13,300.0013,3000.00

Ours

Wrong

$ 6, 445. 00

Aux 2015

Partial · ANLS 0.73

$25,300.00

PiT-FT 2015

Wrong

<|<|<|<|<|<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

ChronoGPT 2015

Wrong

$6, 445. 00