llama.cpp

gguf-split: add `--no-tensor-first-split` option

#7072

Merged