run-llama/llama_cloud_services
1
0
Fork 0
mirror of https://github.com/run-llama/llama_cloud_services.git synced 2026-08-27 01:15:06 -04:00
Code Issues 352 Packages Projects Releases 10 Wiki Activity

Inconsistent Record Extraction #565

New Issue
Open
opened 2026-02-16 00:18:17 -05:00 by yindo · 1 comment
yindo commented 2026-02-16 00:18:17 -05:00
Owner
Copy Link
Copy Source

Originally created by @LadyLegacyK on GitHub (Nov 10, 2025).

Hello,
I have been testing the extraction node and there are serious inconsistencies with data results.
Using the same pdf document with a table with headers and columns, a total record of 86 [as counted].

1st Run - Using Llama UI - The pdf document with 86 counted records, only extracted 76 [Document extraction]
2nd Run - Using Llama UI - The same pdf document with 86 records, only extracted 10 items [Document extraction]
3rd Run - Using LLama UI- The same pdf document with 86 records, only extracted 43 items [Page extraction]

Other than changing from 'Document' to 'Page' extraction, there is no reasonable explanation that I can conclude for the discrepancies.

If anyone can assists with explaining how to get consistent extraction results or help me with settings that would make the outcome of my extraction consistent and reliable.

Originally created by @LadyLegacyK on GitHub (Nov 10, 2025). Hello, I have been testing the extraction node and there are serious inconsistencies with data results. Using the same pdf document with a table with headers and columns, a total record of 86 [as counted]. 1st Run - Using Llama UI - The pdf document with 86 counted records, only extracted 76 [Document extraction] 2nd Run - Using Llama UI - The same pdf document with 86 records, only extracted 10 items [Document extraction] 3rd Run - Using LLama UI- The same pdf document with 86 records, only extracted 43 items [Page extraction] Other than changing from 'Document' to 'Page' extraction, there is no reasonable explanation that I can conclude for the discrepancies. If anyone can assists with explaining how to get consistent extraction results or help me with settings that would make the outcome of my extraction consistent and reliable.
yindo commented 2026-02-16 00:18:17 -05:00
Author
Owner
Copy Link
Copy Source

@neerajprad commented on GitHub (Nov 10, 2025):

@LadyLegacyK: Exhaustive extraction of table rows is hard for LLMs. We have a new experimental extraction target called "Table Row". Could you try that on Premium mode and let us know how it does? If it doesn't yield good results, please share your job_id. Note that you might have to modify your schema so that it corresponds to a single row in the table and you'll get results of type list[schema]

@neerajprad commented on GitHub (Nov 10, 2025): @LadyLegacyK: Exhaustive extraction of table rows is hard for LLMs. We have a new experimental extraction target called "Table Row". Could you try that on Premium mode and let us know how it does? If it doesn't yield good results, please share your job_id. Note that you might have to modify your schema so that it corresponds to a single row in the table and you'll get results of type `list[schema]`
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
LlamaCloud
LlamaExtract
LlamaParse
bug
client
complex layout
documentation
duplicate
enhancement
figure
font
header footer
help wanted
ocr
parser
pdf_debug
pdf_debug
pull-request
Mirrored from GitHub Pull Request
 python-lib
question
server_issue
server_issue
ui
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: run-llama/llama_cloud_services#565
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?