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 · pagefqdg0072_10.png

What is the last ‘guidelines’ mentioned?

OCR text of the page · 488 characters; the scanned image itself is not published
APPENDIX
B . PAGE 10 OF 11 , REVISION 1 .
Furniture:
o Construction & Material:
1. Plastic
2. Metal
3. Wood
- 4. Cloth Covered
-5. Vinyl Covered
Approximate number of seats:
Notes :
Rough Floor Plan:
Guidelines :
o Draw quickly, but label clearly, locations of: vents, doors, and
windows, your location, smoking/non-smoking areas, kitchen, fans, tile
areas.
o Estimate distances.
o Label areas you could not observe.
1601 62905
Source: https://www.industrydocuments.ucsf.edu/docs/fadg0072
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
Label areas you could not observe
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

Draw quickly, but label clearly, locations of: vents, doors, and windows, your location, smoking/non-smoking areas, kitchen, fans, tile areas.

142 characters34 tokens

Aux 2015

Wrong

APPENDIX B. PAGE 10 OF 11, REVISION 1.

Furniture: o Construction & Material:

  1. Plastic
  2. Metal
  3. Wood
    1. Cloth Covered -5. Vinyl Covered Approximate number of seats: Notes :
179 characters64 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Appendices: o Construction & Material:

  1. Plastic
  2. Metal
  3. Wood
    1. Cloth Covered -5. Vinyl Covered -6. Furniture -7. Furniture -8. Furniture -9. Furniture -10
163 characters64 tokens