I am trying to make use of llm so that users can ask xwiki question to get response based on the Wxiki pages I add.
I created a local model using gemma3:1b llm model. The XWiki AI Chat works fine when no collection is added to the context.
I added one wiki page (test) with a one line content to a collection (My docs). After this when I ask the XWiki AI Chat I only get an empty response.
Any help is appreciated. Thanks in advance.
Pako
3
Same here. Have you ever figured out how to fix this?
The LLM App is a great addition to XWiki and I would love to use it in production.
It works fine as long as no collection is added.
As soon as you add collections the response takes forever.
However, I noticed the Stop button disappearing while waiting for the response.
It seems like the connection is lost while waiting for an answer.
Any help would be appreciated.
Jubei
4
You need an embedding Model for the Collection.
LLama 3.2 with all-minilm or jina-embeddings-v2 is working for me.
