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 · pagetqcg0072_1.png

When do they break for lunch?

OCR text of the page · 581 characters; the scanned image itself is not published
LONDON MEETING
ORIENTATION AGENDA
TUESDAY - JULY 7, 1987
Introduction
L. W. Pullen
10:00 - 10:10
Overview
T. Brock/R. Kampe
10:10 - 10:30
Product Development/R&D
K. Rush
10:30 - 11:30
Toxicology/R&D
J. Hayes
11:30 - 12:00
Manufacturing
L. Bass
12:00 - 1:00
LUNCH
1:00 - 2:00
purchasing
G. Mckenna
2:00 - 2:30
Legal/Regulatory
W. Juchatz
2:30 - 3:00
BREAK
3:00 - 3:15
Marketing
G. Mckenna
3:15 - 4:15
Communications
H. Osmon
4: 15 - 4:45
Finance
R. Summers
4:45 - 5:15
Closing
R. Kampe
5:15 - 5:30
EVENING
Free
50614 7851
Source: https://www.industrydocuments.ucsf.edu/docs/tacg0072
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
1:00 - 2:00
Models
4 of 4 columns · click a model to add or remove it

Ours

Partial · ANLS 0.83

12:00 - 1:00

12 characters12 tokens

Aux 2015

Partial · ANLS 0.55

5:15 - 5:30

11 characters11 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Lunch Lunch Lunch Lunch Lunch Lunch Lunch

41 characters20 tokens