llama.cpp

fea95c68 - fix convert.py for codellama, add llama 34B to the list of recognized models (#2768)