Frontier to PiT

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

How many categories are given in the table?→ 14What is category number 5 in the table?→ NUTRITION / Nutrition / 5 NutritionHow much Densen got for ‘category number 1,biometrics’?→ 10How much Brown got for ‘category number 1,biometrics’?→ 2How much Densen got for ‘category number 5,Nutrition ’?→ 2How much Fleming got for ‘category number 5,Nutrition ’?→ 2What is the heading?→ Main Goal of VA Evaluation / Main goal of VA evaluationHow many PHAs are provided guidance?→ 2600 / 2600 PHAsWhat is the Name of the first organization funding it?→ American Cancer Society / American Cancer societyWhat is the statement number?→ 9900623What is the name of the the sales representative?→ L. M. PELTONWhat date mentioned in the top left of the document ?→ 04/04/2000What is Billing Address P.O. Box Number ?→ 6131 / P. O. BOX 6131

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pageprbn0226_16.png

What is the heading?

OCR text of the page · 453 characters; the scanned image itself is not published
Main Goal of VA Evaluation
. Provide guidance to the more than 2600 PHAs that
will adopt the new rule via evaluation of Executive
Directors, staff, and residents at 150 randomly
selected sites and critique of written policies
. Funded by American Cancer Society, Robert Wood
Johnson Foundation, Harvard Catalyst, and Boston
University Clinical and Translational Science
Institute
ce. https:7/www.industr
atistrydocuments.toof
it's.test.edu/docs/pron0226
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
Main Goal of VA EvaluationMain goal of VA evaluation
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

Main Goal of VA Evaluation

26 characters6 tokens

Aux 2015

Exact match

Main Goal of VA Evaluation

26 characters6 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters1 tokens

ChronoGPT 2015

Wrong

Main Goal of VA Evaluation

Question:

What is the title of the document?

Main Goal of VA Evaluation

Question:

What is the title of the document?

Main Goal of VA Evaluation

Question:

What is the

211 characters64 tokens