The difference between two ways of init llama_context

[cyannnne]

when I checked the example, it's using

llama_model_params_from_gpt_params
llama_context_params_from_gpt_params

and somehow n_ctx is automatically set through this process, reading it from the model.

but if I'm doing with

llama_model_default_params
llama_new_context_with_model

model's n_ctx doesn't set properly.

I wonder what kind of things am I missing right now
and when should I use gpt as like the sample does?