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 · pagekqch0227_3.png

how many conferences were held in the fall of 1968 ?

OCR text of the page · 1,352 characters; the scanned image itself is not published
FOREWORD
Many of the immediate questions confronting NIH in its support of
biomedical research relate to the allocation of funds in terms of
the scientific potential of a field. NIH has always believed that
the persons most capable of evaluating a scientific field are sci-
entists themselves -- on the NIH staff or elsewhere in the scien-
tific community. They are best qualified to examine the efforts
and objectives of their colleagues, to set priorities, and to
assist in obtaining an appropriate balance within the health-
related research effort. To investigate how NIH might avail it-
self of this expertise, Dr. Joseph C. Hwang, Chief of the Scientific
Evaluation Section, Office for Research Analysis and Evaluation in
the Division of Research Grants, suggested that the Division con-
duct a series of evaluation conferences. The proposal was endorsed
by the then Associate Director for Research Analysis and Evaluation,
Dr. Arley T. Bever, and four conferences were held in the fall of
1968. The conference dealing with Nutrition and Developmental Biol-
ogy forms the basis for this report.
It is hoped that this report and the others resulting from these
conferences will suggest a possible format for future evaluation
of scientific fields that will provide an improved rationale for
allocation choices.
Source: https://www.industrydocument
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
fourfour conferences
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

four

4 characters2 tokens

Aux 2015

Wrong

The fall of 1968 saw four conferences held by the NIH's Scientific Evaluation Section, Office for Research Analysis and Evaluation, to evaluate scientific fields.

162 characters33 tokens

PiT-FT 2015

Wrong

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

64 characters64 tokens

ChronoGPT 2015

Wrong

Conferences:

  • Nutrition and Developmental Biology
  • Nutrition and Developmental Biology
  • Nutrition and Developmental Biology

Conference:

  • Nutrition and Developmental Biology
  • Nutrition and Developmental Biology

Conference:

260 characters64 tokens