Ways to calibrate.

[DunaSpice]

Please step forward if you would like to collaborate on this project. In what way would you like to give and what is in it for you?

[smithinoz]

Thanks to Mckay, this was a great project that kick started many of our AI journeys.
I still use it for sentimental reasons, and tinker with the code, but seriously it just can't compete with the larger open source projects with hundreds if not thousands of active developers, adding dozens of new features regularly. Most notable are artifacts and improvements to the rag pipelines.
Don't get me wrong, this project still has many advantages such has an extremely low foot print and I will likely continue to tinker.

Also when I started, I needed to use custom models, but now the generic ones more than exceed my needs and capability.
Like how good is the just updated Claude Sonnet 3.5.

Personally have found it far more cost effective now to use someone else's service. I was paying for Matt's back when usage was included, but he kept running out of credits. So now I use another (I deliberately won't mention the name) but it has unlimited usage of all the main models (including music, videos and images) for around $6 a month.

I few calls to the new o1 series of models, I could easily spend that personally per day. Guess it's a swings and roundabouts thing.

I stopped pushing to this repo, as they were never approved. I guess if someone maintained a similar codebase I would likely contribute.
I am currently hosting on Docker and the standard supabase.co database. I simplified the UI, made small changes like allowing the users to set the default model from the model select drop down. But certainly nothing advanced.
And it's been super reliable, a few security issues but very small footprint.
If I was to allow users on, would need to setup some housekeeping.

[DunaSpice]

Thanks for reply I would also like to take this project further. I can sent up a repo or you. Maybe you can also help me figure out why I have errors if I host the project through NGINX. On the local, it works and on the remote it fails of logging in. Any clue?