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 · pagepjgn0226_8.png

what was the first date on which he reviewed Fehring depo ?

OCR text of the page · 922 characters; the scanned image itself is not published
10/20 1 hour working on draft report citations/literature review
10/21 0.5 hours telecom with Fujimoto
10/22 2 hours review technical literature on Pinnicle and
review of Schmalzried depo
10/23 1 hour review of literature for draft report
10/24 3 hours draft clean up and citations
10/28 2 hours review case depos for Lay and Paoli
10/29 2 hours reiview Fehring depo
10/30 2 hours review Paoli Case depo
11/2 3 hours review Fehring depo
11/5 2 hours review Fehring depo
11/6 2 hours review Griffin Depo
11/7 3 hours review of Paoli records
11/8 2 hours review of Lay records
11/8 3.5 hours team with attorneys at OSA
Total 36 hours of work
@ $850.00 / hour
Total payment $30,600.00
Payment is due as previously discussed upon delivery of
invoice and is non-refundable.
PROTECTED DOCUMENT. DOCUMENT SUBJECT TO PROTECTIVE ORDER.
HEINRICH-000000008
Source: https://www.industrydocuments.ucsf.edu/docs/pjgn0226
8LT-03832-00008
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
10/29
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

10/29

5 characters6 tokens

Aux 2015

Wrong

20/10 2 hours review draft report citations/literature review
21/10 0.5 hours telecom with Fujimoto
22/10 2 hours review technical literature on Pinnicle and
review of Schmalzried depo
23/10 1

200 characters64 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

10/20 1 hour working on draft report citations/literature review 10/21 0.5 hours working on draft report 10/22 2 hours review of literature for draft report 10/23 1 hour review of literature for draft report 10/24 3 hours review of literature for draft report 10

262 characters64 tokens