Open mwinkens opened 10 months ago
While this issue still persists, this is now way better with the new "Chat with AI" feature, good change there :+1:
This poses an issue as well for timeouts. For really long responses, it sometimes takes a while to receive the entire thing. There's no feedback to the user and Nextcloud times out the request as well.
Describe the feature you'd like to request
Currently the LocalAI is only using the static endpoint instead of the streaming endpoint, where you would see the answer of the "in real time". I don't know, if the OpenAI endpoint is doing this as well. It would be real nice to have similar behaviour, (e.g. answers coming word by word) having in the nextcloud assistant, too!
Describe the solution you'd like
Describe alternatives you've considered
forcing this fields doesn't make sense, maybe switch to "Current behavior" and "Enhanced behavior". What alternatives do you expect?
I guess the only alternative is still using the static endpoint