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 · pagehtxg0227_3.png

What is the last column heading?

OCR text of the page · 613 characters; the scanned image itself is not published
Number of Patients in the Subdivisions of each Group
Group
Subdivisions
Starting with
Ba 18189
Starting with
Placebo
0.)
Patients with laboratory
data limited to the initial visit
(i.e. relatively useless lab.data for this
11
study)
2-4.) Patients with lab. data obtained
initially, after 2 months and in some
instances after 4 months.
(i.e. Patients receiving Placebo or Ba 18189,
not both)
6-8.) Patients with lab. data obtained
initially, after 2 mo., 4 mo. , 6 mo. and
in some instances after 8 mos.
(i.e. Patients receiving both Pacebo & Ba 18189)
Source: https://www.industrydocuments.ucsf.edu/docs/htxg0227
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
Starting with Placebo
Models
4 of 4 columns · click a model to add or remove it

Ours

Partial · ANLS 0.64

Starting with Ba 18189

22 characters10 tokens

Aux 2015

Wrong

Number of Patients in the Subdivisions of each Group

52 characters12 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Placebo Placebo Placebo Placebo Placebo Placebo Placebo

55 characters20 tokens