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 · pagenpvw0217_6.png

From the 100, how many of them had baseline angiogram?

OCR text of the page · 325 characters; the scanned image itself is not published
Subjects Signed Consent
(signed consent form)
N=103
Subjects Enrolled
Protocol violation
N=100
N=3
(Table 1)
Subject with baseline
Subjects refused to
angiogram
continue in study
N=81
N=19
(Table 2)
IVUS
No IVUS
N=63
N=18
All SVG occluded
DWRITE 045518
or diseased
Source: https://www.industrydocuments.ucsf.edu/docs/npvw0217
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
81N=81
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

N=81

4 characters5 tokens

Aux 2015

Wrong

N=100

5 characters6 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

N=103 N=3 N=3 N=3 N=3 N=3 N=3

29 characters27 tokens