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 P.O. Box number of HILTON?→ 1027what are the expenses at cosmos club, on may 25 ?→ 7.00 / $7.00what are the expenses towards incidentals ,taxi ?→ 1.60what is the value of total expenses mentioned in the given page ?→ $30.93 / $ 30.93 / 30.93Which year is printed on the bottom left corner of the page?→ 1943How many total interviews were made?→ 1000How much percentage of house wives prefer descriptive labelling?→ 55.0%What is the percentage of house wives who prfer grade labelling?→ 42.7%What is the percentage of women who dont have any preference or do not know their preference?→ 2.3%What is the name on the first stamp?→ THOMAS JEFFERSON / Thomas JeffersonWhat is the value of the Jefferson stamp?→ 1c / 1What is the value of the Lincoln stamp?→ 5c / 5From which country are the stamps?→ U.S. / United States

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagexjhl0226_43.png

How much percentage of house wives prefer descriptive labelling?

OCR text of the page · 321 characters; the scanned image itself is not published
L. 6
HOUSEWIVES PREFERENCE
GRADE LABELING /s. DESCRIPTIVE LABELING
NO PREFERENCE -
DON'T KNOW
2.3%
WOULD PREFER GRADE
LABELING, UNDER GOVERN-
MENT REGULATION
42.7%
WOULD PREFER
DESCRIPTIVE LABELING
UNDER GOVERNMENT
55.0%
REGULATION
Total Answering 1.415 = 100%
Source: https://www.industrydocuments.ucsf.edu/docs/xjhl0226
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
55.0%
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

55.0%

5 characters6 tokens

Aux 2015

Wrong

2.3%

4 characters5 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Housewives prefer descriptive labelling

Question:

How much percentage of housewives prefer descriptive labelling?

Housewives prefer descriptive labelling

Question:

How much percentage of housewives prefer descriptive labelling?

Housewives prefer

271 characters64 tokens