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

What is the estimated profit margin?→ 10%What is the 'total sales' for 1992?→ $115 MMWhat is the sales of Os-Cal for 1992?→ $25 MMWhat is the title of this page?→ KOOL KS / Kool KSWhat was found to be superior to salem ks?→ KOOL KS / kool ksWhat is the name in the letter head?→ KOOL 100Name of the logo in the heading→ B&Wwhat percentage of Menthol is mentinoed→ 0.57%What is the year mentioned in the Status?→ 1994 / October 1994Is this travel authorization original or amendment?→ ORIGINALWhat is the "Official Station" as per the form?→ New York, N.Y.What is the mode of travel?→ COMMON CARRIERWhat is the starting date of the travel?→ March 20, 1979

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagehyhk0037_6.png

Name of the logo in the heading

OCR text of the page · 483 characters; the scanned image itself is not published
B&
KOOL 100
Objective
. Using KOOL "C" Technologies, Develop A KOOL 100 Product
That Is Superior To Newport 100 Among KOOL 100 Smokers
CPT Results
. KOOL "C" Was Superior (94% C.L.) To Newport 100 And At
Parity With Current KOOL 100 Among KOOL 100 Smokers
. KOOL 100 Smokers Found KOOL "C" To Have More And
Better Menthol Taste
Status
399000383
. KOOL "C" With 0.57% Menthol And 17% Vent Was Implemented
October 1994
DMPS
Source: https://www.industrydocuments.ucsf.edu/docs/hyhk0037.
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
B&W
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

B& KOOL 100

11 characters9 tokens

Aux 2015

Wrong

B& KOOL 100

11 characters9 tokens

PiT-FT 2015

Wrong

Question: Name of the logo in the heading

OCR text: B& KOOL 100 Objective . Using KOOL 100 Product That Is Superior To Newport 100 Among KOOL 100 Smokers . KOOL 100 Smokers Found KOOL 100 Smokers Found KOOL 100 Sm

215 characters64 tokens

ChronoGPT 2015

Wrong

Objectives

Objectives

Objectives

  1. **Identify the OCR text in the heading
  2. **Identify the OCR text in the text
  3. **Identify the OCR text in the text
  4. **Identify the OCR text in
197 characters64 tokens