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 · pagenhgh0228_6.png

Bale at various moisture contents are for how many stalk positions?

OCR text of the page · 992 characters; the scanned image itself is not published
Plastic Management
Purpose:
Correlate curing rate with TSNA formation.
.
Use regular crop tobacco
Use 6" stick spacing
.
Black plastic on ground with plastic down after loading tobacco
Tyvek pulled down after loading tobacco
Test Curing Barns
Purpose:
Test controlled curing environment and effects on TSNA formation.
Use regular crop tobacco
.
.
Two cures per barn
Curing schedules to be determined
Post Harvest
Purpose:
Test tobacco moisture and bale density versus TSNA formation
Use regular crop tobacco
Sampling
Time 0 (stripping)
O
1 month
O
2 month
. Bale at various moisture contents for three stalk positions
Moisture
Stalk
Unitized
(%)
Position
Pallets
<14
L,M,H
3 (1 per stalk position)
18-20
L,M,H
3 (1 per stalk position)
>25
L,M,H
3 (1 per stalk position)
.
Bale at two densities for one stalk position (Mid) and one
moisture content (18% - 20%)
Density
Unitized
Variable
Ibs/bale
Pallets
Low
65
1
High
100
52414 6095
Source: https://www.industrydocuments.ucst.edu/docs/nhgn0228
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
three
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

three stalk positions

21 characters4 tokens

Aux 2015

Wrong

1

1 characters2 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

ChronoGPT, a large language model trained by Manela Lab at WashU, is a language model trained by Manela Lab at WashU.

ChronoGPT is a large language model trained by Manela Lab at WashU.

ChronoGPT is a language model

217 characters64 tokens