Stop llama.cpp spamming stderr

[andreib1]

Is there any way to mute all of the chatter coming from llama.cpp?

[amorroxic]

I do this if that helps, not sure this is the right way. In binding.cpp:

#include "log.h"

and then in void* load_model function adding this (before return load_binding_model):

log_disable();

or from what I read in llama.cpp/common/log.h another avenue would be defining some flags like so (this would prevent creation of .log files for instance, not tested):

#define LOG_TARGET stderr
#include "log.h"