Frontier to PiT

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

How many people did not answer question 10?→ 1What is the highest number of responses for Question 10, for any given % of inside sales?→ 17How many people did not answer question 11?→ 8How many people have a percentage of Total sales between 11 to 16%?→ 10How many organisations are currently in the carton sales business as well as pack?→ 27What was the question# of the question unanswered by maximum number of people?→ 14Which checkbox is checked?→ FOR YOUR INFORMATIONWho is this slip from?→ ChrisWhat is the title of the ‘form’?→ Water Analysis / Water analysisWhat is the name of factory?→ Gering / Gering factoryWhat ‘kind of water’ is given for analysis ?→ Boiler feedWhat is the ‘pH’?→ 7.5What is the date on the ‘form’?→ May 4, 1960

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagefzkc0228_1.png

Which checkbox is checked?

OCR text of the page · 461 characters; the scanned image itself is not published
NOTE.-DO NOT USE THIS ROUTE SLIP TO
SHOW FORMAL CLEARANCES OR APPROVALS
DATE
10 June 64
TO:
AGENCY
BLDG. ROOM
APPROVAL
REVIEW
OPER CONVERSATION
SIGNATURE
NOTE AND SEE ME
AS REQUESTED
COMMENT
NOTE AND RETURN
NECESSARY ACTION
FOR YOUR INFORMATION
[] PREPARE REPLY FOR SIGNATURE OF
REMARKS:
(Fold here for return)
To
- - - -
From
PHONE
BUILDING
ROOM
FORM HEW-30 REV. 11/56
ROUTE SLIP
GPO : 1956-0-409606
Source: https://www.industrydocuments.ucsf.edu/docs/fzkc0228
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
FOR YOUR INFORMATION
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

PREPARE REPLY FOR SIGNATURE OF

30 characters10 tokens

Aux 2015

Wrong

To

2 characters2 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Input:

Question:

Which checkbox is checked?

OCR text:

Question:

Which checkbox is checked?

Question:

Which checkbox is checked?

Question:

Which check

176 characters64 tokens