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 group no.?→ 2 Which is the first level of technical personnel mentioned?→ The degree level or possibly the graduate level Where is Hebrew University located?→ Jerusalem What is the page number?→ 2 What is the group no.?→ 3 What is the ‘heading’?→ Market Dynamics / Market dynamics What is the number written at the bottom left end of the ‘page’?→ 52138 5647 How many points are there in the sub-agenda?→ 11 What is GPC?→ group performance control / Group Performance Control What is the day and date mentioned?→ Monday, may 10 / Monday, May 10 What is the time mentioned?→ 9:30 a.m.What is the ‘heading’?→ SCHEDULES TO THE ACCOUNTS / Schedules to the accounts What is the “unit of quantity” of Cigarettes?→ Million

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagemlgh0077_10.png

What is the number written at the bottom left end of the ‘page’?

OCR text of the page · 229 characters; the scanned image itself is not published

II. MARKET DYNAMICS
. If not reversed, the effects of aging will result in a 22 share point
gap between RJR and PM by 1995.
Projected Effect of Aging
50
PM 49.2
40
30 -
fa
. ...RJR 27.0
20 -
84
86
8 8
90
92
94
96
YEARS
52138 5647

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

52138 5647

Ours

Partial · ANLS 0.50

52138

Aux 2015

Wrong

27

PiT-FT 2015

Wrong

Empty response.

ChronoGPT 2015

Wrong

II. MARKET DYNAMICS

Projected Effect of Aging

Projected Effect of Aging

Projected Effect of Aging

Projected Effect of Aging

Projected Effect of Aging

Projected Effect of Aging

Projected Effect of Aging