My output is too low

[paulocoutinhox]

Hi,

Im trying implement simple.cpp into my app, but the output is less than 20 characters.

Im using the mistral 7b model.

Im checking the simple.cpp file and see this var:
https://github.com/ggerganov/llama.cpp/blob/master/examples/simple/simple.cpp#L30

What it represent?

How can i get the biggest number of tokens removing my own prompt and using this number as tokens to be returned, based on simple.cpp?

Thanks.