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 · pageqnph0078_25.png

What is the name of the person to whom the receipt should be delivered ?

OCR text of the page · 737 characters; the scanned image itself is not published
CLASSIFIED MATERIAL RECEIPT
Title and/or Number of Document: GTC Testing Program Memo.
Number of Copies:
Classification:
Secret
Confidential O
Deliver To:
Charlie Green
From: Rhenda H. Steele
Date: August 30. 1994
Received By:
Date:
CLASSIFIED MATERIAL RECEIPT
Title and/or Number of Document: GTC Testing Program Memo
Number of Copies:
Classification:
Secret
Confidential
Deliver To:
Charlie Green
From: Rhenda H. Steele
Date: August 30, 1994
Received By:
Date:
INSTRUCTIONS FOR USE
A
When Transmitting:
B.
When Receiving:
Prepare top and bottom of form
Verify that material received is as described.
and forward form with material.
Sign and return bottom portion of form to
sender. Retain top portion of form for your files.
50940 6351
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
CHARLIE GREENCharlie Green
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

Charlie Green

13 characters3 tokens

Aux 2015

Wrong

CLASSIFIED MATERIAL RECEIPT

Title and/or Number of Document: GTC Testing Program Memo. Number of Copies: Classification: Secret Confidential Deliver To: Charlie Green From: Rhenda H. Steele Date: August 30, 1994 Received By

224 characters64 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Top

Bottom

Confidential

Secret

Confidential

Confidential

Confidential

Confidential

Confidential

Confidential

Confidential

Confidential

Confidential

Conf

221 characters64 tokens