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 · pagenhpj0226_1.png

Who has signed the letter?

OCR text of the page · 671 characters; the scanned image itself is not published
July 27, 1972
Mr . Bob Goldsmith
Marketing Director
Boys' Life
271 Madison Avenue
New York, NY 10016
Dear Mr. Goldsmith:
I have been studying the use and safety of food
additives and the nutritive value of foods, particularly
those consumed by children.
Boys' Life's ad in Advertising age mentioned the fact
that your Reader Panel drinks (on the average ) more than
three soft drinks each day. I would appreciate your
sending me further details on this study, such as how
many readers were involved, how large were the drinks, etc.
Looking forward to hearing from you, I am,
Sincerely,
/ Michael F/ Jacobson, Ph.D.
Source: https://www.industrydocuments.ucsf.edu/docs/nhpj
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
Michael f. Jacobson, Ph.d.Michael F. Jacobson, Ph.D.Michael f. Jacobson
Models
4 of 4 columns · click a model to add or remove it

Ours

Partial · ANLS 0.89

/ Michael F/ Jacobson, Ph.D.

28 characters11 tokens

Aux 2015

Wrong

Mr. Bob Goldsmith

17 characters6 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

ChronoGPT, a large language model trained by Manela Lab at WashU, has been trained to respond to the following questions:

  1. What is the average number of soft drinks consumed by a reader each day?
  2. What is the average number of soft drinks consumed by a reader
264 characters64 tokens