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 · pagegpnn0081_3.png

What is the performance and index for Vantage?

OCR text of the page · 708 characters; the scanned image itself is not published
III. SEGMENT PERFORMANCE
VIRILE SEGMENT OVERDEVELOPED (150 INDEX) .
MODERATION SEGMENT IS THE ONLY OTHER SEGMENT
HOLDING SHARE.
ALL OTHER SEGMENTS SHOW CONSIDERABLE WEAK-
NESS.
SAVINGS SEGMENT AVERAGES 60 INDEX IN BOTH
OVER AND UNDER DEVELOPED SAVINGS MARKETS.
IS POPULATION REFERENCE POINT ADJUSTMENT
NEEDED?
IV.
COMPANY PERFORMANCE
RJR OVERDEVELOPED (AVERAGE INDEX 120)
PM OVERDEVELOPED (AVERAGE INDEX 115)
ALL OTHER COMPANIES LOSE SUBSTANTIALLY
V.
BRAND PERFORMANCE
WINSTON/CAMEL HIGHLY OVERDEVELOPED AVERAGING
220/180 RESPECTIVELY
VANTAGE PERFORMING WELL (AVG. 110)
ALL OTHER RJR BRANDS UNDERDEVELOPED
PM'S STRENGTH IS SOLELY DUE TO MARLBORO
(AVG. INDEX 150)
ALL OTHER PM BRANDS UNDERDEVELOPED
51338 6008
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
performing well (Avg. 110)Performing well (Avg. 110)
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

PERFORMING WELL (AVG. 110)

26 characters14 tokens

Aux 2015

Wrong

Vantage's VIRILE SEGMENT OVERDEVELOPED (150 INDEX) . MODERATION SEGMENT IS THE ONLY OTHER SEGMENT HOLDING SHARE. ALL OTHER SEGMENTS SHOW CONSIDERABLE WEAKNESS. SAVINGS SEGMENT AVERAGES 60 INDEX IN

196 characters64 tokens

PiT-FT 2015

Wrong

Question: What is the performance and index for Vantage?

OCR text: III. SEGMENT PERFORMANCE VIRILE SEGMENT OVERDEVELOPED (150 INDEX) . MODERATION SEGMENT IS THE ONLY OTHER SEGMENT H

184 characters64 tokens

ChronoGPT 2015

Wrong

VICTORIOUS SEGMENT OVERDEVELOPED (AVERAGE INDEX 120)

SEGMENT PERFORMANCE

VICTORIOUS SEGMENT OVERDEVELOPED (AVERAGE INDEX 120)

SEGMENT PERFORMANCE

155 characters64 tokens