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 · pagenlcn0226_4.png

What is the page number?

OCR text of the page · 2,292 characters; the scanned image itself is not published
observed range of data to the unobserved range, and with the use of
mathematicalmodels. All of these approaches rely heavily on expertjudgment,
assumptions, and extrapolations. The final risk assessment, including model
results, can vary widely depending on the built-in judgments, assumptions, and
data. Forexample, a modelmay assume an average resting breathing rate, or
a heavierbreathing rate to capture a working orexerc ising scenario ; the choice
may produce widely divergent predictions for the amount of an air pollutant that
enters the lungs in a given time.
Regulatory agencies know that the realities of constantly emerging new
science and the frailties inherent in available evidence dictate that it will never
e liminate all majorassumptions and judgments from its dec isio n-making. Our
public health and environmentalprograms, however, would not be effective if
incontrovertible evidence of harm were a prerequisite of regulatory action. To
quote Bradford Hill, the fatherof knowledge criteria forepidemiology:
"All sc ie ntific work is incomplete -whe the rit be observationalor
experimental. All scientific work is liable to be upsetormodified by
advancing knowledge. That does not conferupon us a freedom to ignore
the knowledge we already have, orto postpone the action it appears to
demand at a given time ." (Bradford Hill, 1965)
However, without any scientific support forregulatory decisions, courts will strike
down any proposed protections for lack of sufficient evidence. The te mble
paradox is that waiting for"evidence" is usually a matterof waiting foran
increase in disease and death among the exposed population. Thus, a
sig nific ant issue foragencies is how much analysis is necessary be fore
promulgateng a rule. Courts have consistently acknowledged the need to
proceed without full evidence, citing the precautionary goals of most
enviro nmental statutes, see Reserve Mining, Ethyl, recent DC Cir Clean Air
opinion. In the legaldecision of American Truc king on remand, the DC Cir
affirmed the propriety of "eming ] on the side of caution". American Truc king
Assns. v. EPA, 283 F.3d 355, 369 (D.C. Cir. 2002).
NRDC CONCERNS WITH THE OMB PROPOSED RA BULLETIN
NRDC on OMBRA Bulle tin June 2006
4
Source: https://www.industrydocuments.ucsf.edu/docs/nicn0226
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
4
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

4

1 characters2 tokens

Aux 2015

Wrong

The page number is 4.

21 characters8 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters1 tokens

ChronoGPT 2015

Wrong

OCR text: observed range of data to the unobserved range, and with the use of mathematicalmodels. All of these approaches rely heavily on expertjudgment, assumptions, and data. The final risk assessment, including model results, may vary widely depending on the built-

268 characters64 tokens