llama.cpp

68f53485 - convert.py : start a new simplified implementation by removing old stuff