Open tomscoppock2 opened 6 months ago
Create an index schema and skillset that can work with any file size, inc. cracking the file and vectorizing content in a projection index.
Remove skills that fail if fed too much data (e.g. keywords and entities), or reduce the chunk size output from documents.
Goal is to have a complete end2end index build that successfully runs regardless of the number of pages / words in a document.
Include common fields & metadata, inc. classification, pages and co-ordinates.
Create an index schema and skillset that can work with any file size, inc. cracking the file and vectorizing content in a projection index.
Remove skills that fail if fed too much data (e.g. keywords and entities), or reduce the chunk size output from documents.
Goal is to have a complete end2end index build that successfully runs regardless of the number of pages / words in a document.