[QUESTION] bitnet.cpp using rockchip NPU hardware acceleration

[nichonaugle]

Bitnet.cpp seems to be a great lightweight model that would benefit from the hardware acceleration benefits available from many Rockchip processors. However given the unique nature of the LLM architecture it seems to be built using a custom framework outside of ONNX, Tensorflow, etc. Is there a way to run bitnet using the available NPU hardware acceleration?

https://github.com/microsoft/BitNet