Gemma.cpp from google

[julmich]

I searched the forum here and did not find anything about this topic, so I put it here. Nothing so drastic in my opinion, proving 100% the value of llama.cpp and mainly of the brilliant work of Georgi Gerganov with his ggml.
On the chat with Gemini Pro about the usage of an LLM model with a C++ interface, I got this answer :
You would use a C++ LLM interface engine like gemma.cpp (the standalone C++ implementation) or the llama.cpp project(which supports Gemma). See the image attachment below. I searched github and found :
https://github.com/google/gemma.cpp
Again, I just want to inform the community.