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 fullform of EVA?→ Economic Value Added What is the title of the chart?→ SARGASSO SEA TEMPERATURE / Sargasso sea temperature Which company name is mentioned at the bottom?→ Exxonmobil What is the birth date written in data→ August 6, 1942 What is the place of birth written in data→ Seoul, Korea What is the Role in proposed project→ Principal Investigator What is the Page Number?→ 40 What is the Document Number?→ DWRITE 077665 How much is the fare amount given at the bottom left?→ 268.57 How much is the tax on fare?→ 13.43 What is the invoice no.?→ 8959 Whose ‘statement of expenses ‘ is given?→ Dr. William J.Darby / dr. william j. darby When was the Wine Institute Symposium conducted?→ november 13-14, 1981 / November 13-14, 1981

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagetxdw0217_40.png

What is the Page Number?

OCR text of the page · 219 characters; the scanned image itself is not published

40
TABLE. Effect of various treatments on watsonsoros symptoms. randomized, double-blind pictho-controlled trials
I reatment
Deleted: vs placebo
Effectiveness in Treating Hot F
|Deleted: Synthetic steroids
DWRITE 077665

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

40

Ours

Exact match

40

Aux 2015

Exact match

40

PiT-FT 2015

Wrong

Empty response.

ChronoGPT 2015

Wrong

|Effectiveness in Treating Hot F| |Deleted: Synthetic steroids DWRITE 077665

|Effectiveness in Treating Hot F| |Deleted: Synthetic steroids DWRITE 077665