llama.cpp
Add attention and final logit soft-capping, update scaling factor to Gemma2
#8197
Merged

Loading