llama.cpp
convert : fix tensors using grad in some models
#4173
Merged

Loading