llama.cpp

84250014 - gguf-split: add --no-tensor-first-split (#7072)