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_12.png

What is the page number?

OCR text of the page · 2,380 characters; the scanned image itself is not published
force agency inaction is often exactly what industry and OMBmay seek to
accomplish with such proposals. The tobacco industry introduced the technique
of manufactured doubt as a means to deny health impacts and delay
regulation of its products: "Doubt is ourproduct since it is the best means of
competing with the body offact that exists in the mind of the generalpublic . It
is also a means of establishing controversy." (1969 intemal tobacco industry
memo, stamped "confidential" ) Studies of documents from the tobacco industry
archives have revealed evidence of concerted industry efforts to obscure the
contribution of secondhand smoke and otherenvironmental toxic s to disease
through the development of theirown version of "good epidemiologic al
practices" and "so und science", there by infusing the scientific lite ra ture with
"anti-data" intended to obfuscate scientific consensus (Ong and Glantz 2001).
The OMB Bulletin stands in a long tradition of infusing uncertainty, subjecting
evidence of harm to repeated challengesad infinitum, and derailing ordelaying
re gula tory actions.
OMB has not presented a compelling empiric al justification forforcing a
one - size - must- fit- all approach forall agencies. OMB insists on empiric al
evidence in the rule making process. But the Bulletin lacks any empiric al
evidence about the nature and extent of the problems with risk assessment
practices in each of the agencies. The Bulletin only contains general
pro no uncements that the risk assessment process can be improved to be better
understood, transparent, and more objective. Without knowing the specific ity of
the problems that the agencies and interested stake holders are confronting, it
becomes difficult to craft an appropriate so lution or so lutions.
Instead of presenting empiricalevidence of the problems, OMBleaped to
a so lution, but again it failed to provide any evidence demonstrating the
efficacy of its proposed one-size -fits-all solution forallagencies.
By rushing to judgment, OMB's Bulletin would effectively force the
goverment to engage in a vast, unwieldy experiment. What is appropriate and
necessary for the Food and Drug Administration in calculating and conducting
risk a sse ssments may not be appropriate for the Environmental Protection
NRDC on OMBRA Bulle tin June 2006
12
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
12
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

12

2 characters3 tokens

Aux 2015

Wrong

The page number is 12.

22 characters9 tokens

PiT-FT 2015

Wrong

<|<|<|<|<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

64 characters64 tokens

ChronoGPT 2015

Wrong

OCR text:

Question:

What is the page number?

OCR text:

Question:

What is the page number?

OCR text:

Question:

What is the page number?

155 characters64 tokens