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

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagehyhk0037_5.png

What was found to be superior to salem ks?

OCR text of the page · 481 characters; the scanned image itself is not published
B&W
KOOL KS
Monitor Results
. KOOL KS Vs. Newport KS
- KOOL KS Was Found To Be At Parity With Newport KS
Among KOOL KS And Newport KS Smokers
. KOOL KS Vs. Salem KS
-
KOOL KS Was Found To Be Superior To Salem KS Among
Salem KS Smokers And At Parity Among KOOL KS Smokers
Status
399000382
. KOOL KS Vs. Newport KS Monitor Will Be
Conducted Again Using A Greater Number Of
Total Smokers (n > 400 compared to n < 300)
DMP46
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
KOOL KSkool ks
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

KOOL KS

7 characters4 tokens

Aux 2015

Exact match

KOOL KS

7 characters4 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

OCR Text:

B&W KOOL KS Monitor Results

Question:

What was found to be superior to salem ks?

OCR text: B&W KOOL KS Monitor Results

Question:

What was found

169 characters64 tokens