llama.cpp

d0f77b13 - convert.py : try to determine n_ctx automatically for CodeLlama (#2770)