This does seem promising when playing with the demo. That says running on a T4 I believe. This probably means we are only months away from viable self hosted co

This does seem promising when playing with the demo. That says running on a T4 I believe. This probably means we are only months away from viable self hosted code generation and analysis that could compete with gpt 3.5 turbo. It seems like working on locally runnable LLM that has been fine tuned to focus on code should be a high priority.