hi!

i have installed privatechatgpt (https://docs.privategpt.dev/installation) on raspberry pi 4b Bookworm and qmkl6 also.
and looking how to speed it up a bit.
may be there is any flags to compile like follows :
CMAKE_ARGS="-DNEON=on" pip install --force-reinstall --no-cache-dir llama-cpp-python
to use qmlkl6 ?