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 · pagehrlg0227_1.png

What is the name of the University?

OCR text of the page · 654 characters; the scanned image itself is not published
WASHINGTON UNIVERS ITY
SPECIAL
ACCOUNT DISTRIBUTION
CHECK EMP.
EMPLOYEE NAME
ACCOUNT NUMBER
AMOUNT
PERCENT AUTH AUTH. SERVICE
SOCIAL
rm 6415
DATE NO.
AUT HOR IZED
PER NO.
SECURITY
MO DAY
CL DEPT BG CB FUND
PER ACCOUNT
ACCOUNT
FROM
TO
CHARGE
02 28 124232 FRANCES ELLEN BROCK 22 3235 11 00 64117 Res Asst
465.00 100.00 3334
02 28 483378 IRENE E KARL
22 3235 11 00 64117 Res Asst
333.33
2/1-2/28/66
798.33
50.00 3334
2/1-2/28/66
Journal 2442 dated 2/16/66:
Frances E. Brock 22 3235 11 00 64117 Res Asst 465.00
Salary for period 1/1-1/31/66
Irene Karl
22 3235 11 00 64117 Res Asst 333.33
Salary for period 1/1-1/31/66
798.33
Source: https://www.industrydocum
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
WASHINGTON UNIVERSITYWashington UniversityWASHINGTON
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

University of Washington

24 characters4 tokens

Aux 2015

Partial · ANLS 0.57

The University

14 characters3 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

4 characters1 tokens