How to print intermediate result in llama.cpp

[torettoluke]

For example in llama2.c

I make some print, and the result is ok
But I try this method in llama.cpp
It seems not work
Is there some setting relate to print?

[slaren]

Check the eval-callback example.

[torettoluke]

Can download the intermediate results in chat mode？

[torettoluke]

I want to get the results in chat mode， how to use eval-callback ？