Use llama 3 instead of codellama

[magneticmojo]

Could you make it configurable to chose llama 3 instead of codellama?

[yusufcanb]

There is already a ticket regarding allowing other models. I will investigate and try to understand it's feasibility on computers with limited resources.

[Trapper4888]

There is also codeqwen and deepseek 7b that should be good.

I think it would be better to have a default model, let the user optionally change it, and hardcode the custom parameter into the request instead of making 2 modelfils and 2 new alias on ollama side for the request and suggest.