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 · pageppnf0023_1.png

what is the given document about?

OCR text of the page · 1,082 characters; the scanned image itself is not published
(Attachment #3 Region Monthly Telmark Program Evaluation Survey)
Region Name:
1. Retail Survey Summary
# Stores Surveyed
# Stores That Received Promo
#DIV/O! %Stores That Received Promo
# Stores That Placed Display in Visible Location
#DIV/O! % Stores That Placed Display In Visible Location
# Stores That Placed POS in Visible Location
#DIV/O! % Stores That Placed POS In Visible Location
# Stores Promotion Selling In
#DIV/0! % Stores Promotion Selling In
# Stores That Will Participate in Next Authorized Promotion
#DIV/0! % Stores That Will Participate in Next Authorized Promotion
Summary Notes:
Il. Wholesale Summary
# Accounts Sold by Telmark in the Region
# Telmark Sold Accounts Promotion Shipped To
#DIV/O! % Telmark Sold Accounts Promotion Shipped To
# Telmark Sold Accounts Promotion Shipped To That Refused It
#DIV/O! % Telmark Sold Accounts Promotion Shipped To That Refused it
Summary Notes:
*Also attach Recap of Attachment 2 for All Wholesalers Surveyed in the Region.
survey/Telmark.xis
52014 9612
- .-.
Source: https://www.industrydocuments.ucsf.edu/docs/ppnf0023
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
region monthly telmark program evaluation survey
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

Region Monthly Telmark Program Evaluation Survey

48 characters8 tokens

Aux 2015

Wrong

Region Name: Retail Survey Summary

Stores Surveyed

Stores That Received Promo

#DIV/O! %Stores That Received Promo

Stores That Placed Display in Visible Location

#DIV/O! % Stores That Placed Display In Visible Location

Stores That Placed POS in Visible Location

#DIV

274 characters64 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong
# Retail Survey Summary
# Retail Survey Summary
# Retail Survey Summary

The same 24-character fragment repeats 8 times until the token limit. Showing the first three.

264 characters64 tokens