whats interesting is that I am able to run the model with ollama command on server directly though

found it, there is a param callaed llm where you can pass model