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 · pagejjxh0224_1.png

When is the Program End Date given?

OCR text of the page · 1,130 characters; the scanned image itself is not published
Program Detail
52410 1556
(Completed by SMMO-Inventory Mgr.)
Program Name Doral Frontline Supermarket Test
Date
Program DTS
4/1/01
PPS Plan #
9972
Program End Date
9/30/01
PPS Program #
805409
# of Offers
YA Job # (Continuity Only)
(If Continuity, also attach Forecast)
(If Retail, also attach Allocations)
Program Description
This is a short description of the Program (either typed or hand written). Doral will test signage, pricing and name generation at 2300 supermarkets where we have
contracts. Each supermarket will display 1 pricer (2-sided), 2 transparencyes, 2 bundles of name gen take ones (100/bundle), and 2 takeone riser cards. Each store
will also need 2 sets of 3" fonts for the pricers (1 set is extra so that when prices change they have numbers on hand).
Product Description
This is used to describe Product Requirements (either typed or hand written)
Other Program Details
This contains other pertinent program details (either typed or hand written). Used primarily by Continuity for communications to YA.
PITDor Fontline Test.xis Item Detail
page 1
Source: https://www.industrydocuments.ucsf.edu/docs/jjxh0224
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
9/30/01
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

9/30/01

7 characters8 tokens

Aux 2015

Wrong

Program End Date

16 characters4 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Program End Date

Input:

Program End Date

Output:

9/30/01

68 characters28 tokens