Jean Louis <bugs@gnu.support> writes:

>> Is the source code available anywhere?
>
> Of course it is available on request but my system is too complex. I would 
> like to make the simpler system and first I need to find some binary vector 
> database or easiest way to install the vector database maybe with the 
> PostgreSQL.
>
> postgresql-17-pgvector is used, so it is available on every system.
>
> To make it run for all users without complexities, I would first need to make 
> the database initiation. 
>
> - then there is text chunking application in Python

Do you parse Org files for chunking? What is the chunking strategy?

> - and there is embeddings model that runs on llama.cpp or any other inference 
> serving.

Which model did you use? How did you integrate it with pgvector?

> Elisp is easy.

Yet interesting. It is clear that you heavily tailored your code to your
use case, but the code itself could serve as inspiration for
others. Maybe someone can even write a package to be useful for all the
users.

> Prerequisits are rather more complex to install.
>
> The power that creats is great. Links can be expanded in different ways for 
> different outputs (I guess)

No doubt. You may even have special link types that will use the vector
search + extra LLM filtering technique with additional prompt.

-- 
Ihor Radchenko // yantar92,
Org mode maintainer,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Reply via email to