Comment by tugdual

I actually did something similar using llama.cpp a while back, would be curious to see the speedup with this model.

https://github.com/TugdualKerjan/bunny/tree/main