attashe

/

gemma-3-27b-tools-Q5_K_M-GGUF

like 1

I tried different methods with gemma-3, default llama-cpp-server with default template worked but not perfect
In this article described built-in gemma-3 function calling syntax
https://www.philschmid.de/gemma-function-calling

But it need some workaround to impement. After all I switched to qwen3