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 · pageqtjf0226_2.png

Who has lead responsibility of conducting literature search on renal/bladder cancer studies?

OCR text of the page · 986 characters; the scanned image itself is not published
Actos FDA Task Force: Action Plan
Update meetings daily, 5pm, Meeting Room 523
Task
Notes
Lead Responsibility
Due Date
1. DEVELOP CLINICAL MONITORING PLAN (Perez,]
August 8)
Identify studies with at least a
506
Perez
August 1 - DONE
12 month duration
Conduct literature search on
Haskins
August 2 - IN PROGRESS
renal/bladder cancer studies
Identify expert in
Andy Whelton, MD (John
Perez
August 2 - IN PROGRESS
urology/oncology
Hopkins)
Peter Scarred MSICCC - Rose
Set-up teleconference with
Perez '
August 5
expert
Contact UK about urinary tract
Hoos
Teleconference scheduled
safety study in development
for August
2. DEVELOP SEARCH TERM LIST
Develop list of search terms
Ramstack to forward list to
John Page
August 2 - DONE,
relating to development or
TCI and EU
`renal bladder/urology cancer
Last Update: August 1, 2002
Page 1 of 7
Confidential - Subject to Protective Order
TAK-THOMCL-00032210
Produced in MDL on 02/21/13
Source: https://www.indup538-00092ts.ucsf.edu/docs/qtjf0226
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
HASKINSHaskins
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

Haskins

7 characters4 tokens

Aux 2015

Wrong

Actos FDA Task Force: Action Plan

Update meetings daily, 5pm, Meeting Room 523

Task

Notes

Lead Responsibility

Due Date

  1. DEVELOP CLINICAL MONITORING PLAN (Perez,]

August 8)

Identify studies with at least a

216 characters64 tokens

PiT-FT 2015

Wrong

<|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|> <|<assistant|>

120 characters64 tokens

ChronoGPT 2015

Wrong

Chronos

Question:

What is the most important aspect of a literature search on renal/bladder cancer studies?

OCR text: Actos FDA Task Force: Action Plan

Input:

Question:

What is the most important aspect of a literature

233 characters64 tokens