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 · pagerkgl0228_3.png

What is the page number of coal sector?

OCR text of the page · 903 characters; the scanned image itself is not published
TABLE OF CONTENTS (cont'd)
41
APPENDIX 1
Results by Gas Type and Equipment Type
41
41
CO2
42
Methane
43
VOC
44
NOx
44
N20
44
CFC
APPENDIX 2 (Detailed Methodology)
45
Methodology All Sectors
45
Combustion CO2
45
Combustion CH4
46
Combustion VOC
46
Combustion NOx
46
Combustion N20
48
Surface Facility Equipment Leaks
48
Leaks from Surface Casing
53
Well Servicing
54
Characterization of Emissions
55
Natural Gas Sector
58
Oil Sector
60
Heavy Oil Sector
62
Crude Bitumen Sector
62
Oil Sands Mining Sector
63
Coal Sector
63
Other
64
Data Input Sheet
64
VOLUME 2
APPENDIX 3 (Summary of Site Specific Data and Assumptions)
Introduction
Summary Results by Division and Area .
Site Specific Assumptions and Calculations
Summary of Social Reduction Cost Data
Sample Social Cost of Service Calculation
VOLUME 3
APPENDIX 4 (Detailed Area specific Data)
Source: https://www.industrydocuments.ucsf.edu/docs/rkg10228
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
63
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

63

2 characters3 tokens

Aux 2015

Exact match

63

2 characters3 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Appendix 1 Appendix 2 Appendix 3 Appendix 4

Methodology:

Methodology:

Methodology:

Methodology:

Methodology:

Methodology:

Methodology:

Methodology:

186 characters64 tokens