Fork of llama.cpp with fused TurboQuant flash attention — the FA kernel reads raw TBQ4_0 K/V blocks directly from global memory and dequants via centroid lookup in the FWHT-rotated domain. No separate ...

NEW: Thonny now runs in single instance mode. Previously, when you opened a py file with Thonny, a new Thonny instance (window) was created even if an instance existed already. This became nuisance if ...