AI

I have recently seen lost of AI content from many big companies and have wanted to experiment with it my self. The most practical thing I have been able to think of is an AI assistant on my PC. I started when I saw this video (https://www.youtube.com/watch?v=XvbVePuP7NY&t) from network chuck. This was prity close to what I wanted so I gave it a go. The first issue I ran into was the inability to have add ons when home assistant is installed in docker. However this was a simple fix I just installed them as separate docker containers and manually added them to the whyoming intergration. I then installed the whyoming satalite on my main PC which went smoothly. However I did the extra step of adding sounds to the start and end. ...

I started with the docker compose file from https://github.com/open-webui/open-webui. I then edited this changing the volumes to mounts on my nas and changing the port. I then installed a range of models some designed for general conversations and others for coding. I then wanted to add tab auto complete to vscode so i used a section of technotim’s guide (https://technotim.live/posts/ai-stack-tutorial/). This worked fine and was all round a simple process. I look forward to trying to customise some of the models.

This project has been a roll-coaster and one of the hardest things I have attempted my goal when I set out was simple to run local Ollama models on a cheap £60 k80 GPU. My first attempted end very quickly when I found out that the GPU I got was faulty and it spewed out smoke within a second of first boot. After I confirmed the rest of my server was unharmed I got a full refund and purchased another K80. This one did not fail on first boot and so I was able to start setting up my GPU. I spent the next month trying on and off failing every time. ...