You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
New type of element should be introduced into DocumentSearch: ImageElement.
ImageElement can be originate either from:
bigger document like pdf
be a standalonde .img / .png document
ImageElement should consist:
summary / description generated by multi-modal llm
OCR extracted text
image bytes
Motivation
A lot of technical documentation / manuals consist a lot of technical drawings, schemas, etc. which may be really important part of context while answering user questions
Additional context
In the future, we may try directly embedding images rather than creating descriptions with LLM: #85
The text was updated successfully, but these errors were encountered:
Feature description
New type of element should be introduced into DocumentSearch:
ImageElement
.ImageElement can be originate either from:
ImageElement should consist:
Motivation
A lot of technical documentation / manuals consist a lot of technical drawings, schemas, etc. which may be really important part of context while answering user questions
Additional context
In the future, we may try directly embedding images rather than creating descriptions with LLM: #85
The text was updated successfully, but these errors were encountered: