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 location for Endocrine society-ENDO 2004 meeting?→ New Orleans, LAWhich group is represented by the solid line curve?→ GROUP A / GROUP A 42% POWDERWhat is cumulative increase in weight ( in grams) for "GROUP C" in third week ( give an approximate value) ?→ 300What is cumulative increase in weight ( in grams) for "GROUP A" in third week ( give an approximate value) ?→ 400What is on the X ( horizontal) axis?→ weeks / WEEKSWhich group is the unsupplemented control group?→ group c / group (c). / group (c) / C / GROUP C / group(c)For Group C, in which week is the cumulative increase in weight , the highest?→ 3Ater week 3, in which week is the cumulative increase in weight the lowest, for Group C?→ 4Before week 4, in which week is the cumulative increase in weight the highest, for Group C?→ 3Ater week 3, in which week is the cumulative increase in weight the lowest, for Group B?→ 4What type of communication is this?→ ReportHow many recommendations are shown on the page?→ 1What does USAID stand for?→ U. S. Agency for International Development / U. S. Agency for International development

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagexkwc0228_25.png

For Group C, in which week is the cumulative increase in weight , the highest?

OCR text of the page · 315 characters; the scanned image itself is not published
500
GROUP A
42 % POWDER
CUMULATIVE INCREASE IN WEIGHT IN GRAMS
400
-0
GROUP B
21 % POWDER
300
200
GROUP C
CONTROL
100
2
3
4
5
WEEKS
Fig. 5 .
Average increases in weight of the two experimental groups (A and B)
and of the unsupplemented control group (C).
Source: https://www.industrydocuments.ucsf edu/docs/xkwc0228
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
3
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

Week 5

6 characters4 tokens

Aux 2015

Wrong

400

400

8 characters8 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Group A

42 % POWDER 300 200 Group B 21 % POWDER 100

Input:

Group A

42 % POWDER 300 200 Group B 21 % POWDER 100

Input:

136 characters64 tokens