llama.cpp

gguf : reject tensor size that wraps after padding

#26979

Merged