support for settings.n in ollama

[cognitivetech]

ollama/ollama#7518

trying to figure out why I get only 1 generation from ollama.

I suspect this is due to constraints of its api, but noting here for future discussion

perhaps related to this same issue with llama.cpp server? #25

[deltanym]

yup llama.cpp and ollama do not have n. we need to take it up with them