New3 developments · 1 source · over 2 days

llama.cpp b12065 fixes KV cache graph scheduling crash

How it developed

  1. llama.cpp ReleasesWhere it started

    llama.cpp b12065 fixes KV cache graph scheduling crash

  2. llama.cpp Releases

    llama.cpp fixes quantized KV cache state corruption

    llama.cpp fixes quantized KV cache state corruption caused by miscalculated element counts during on-device state copies

  3. llama.cpp Releases

    llama.cpp b12119 stabilizes realtime server

    llama.cpp b12119 stabilizes the realtime server by fixing exception crashes and enforcing stricter tool schema validation