llama.cpp

[Mirror] server: improve slots scheduling for n_cmpl

#80

Open