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 · pagethcn0226_1.png

Who is presenting this?

OCR text of the page · 217 characters; the scanned image itself is not published
. . . ... ...
June 30, 2016
COMMENTS AT THE EPA PUBLIC SCIENCE
MEETING ON TERT-BUTANOL
Nancy Beck, PhD, DABT
Senior Director, ACC
American
Chemistry
Council
Source: https://www.industrydocuments.ucsf.edu/docs/thcn0226
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
Nancy Beck, PhD, DABT
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

Nancy Beck, PhD, DABT

21 characters10 tokens

Aux 2015

Exact match

Nancy Beck, PhD, DABT

21 characters10 tokens

PiT-FT 2015

Wrong

Question:

11 characters4 tokens

ChronoGPT 2015

Wrong

June 30, 2016

13 characters4 tokens