Damus
NonMetalCoin · 4d
Will you continue or use cloud LLM?
Leito · 4d
How does their performance compare?
Kieran · 4d
yes, the tool calls in gemma4 are not great yet unless you use ollama latest build, generally tool calling doesnt work well at release until engines support the model
ABH3PO · 4d
I can corroborate, have had a similar experience.
Tubii · 4d
Gemma is smart but not for agentic stuff. Best for chats and so on. Qwen coder next is the best stuff right now. Running it on CPU which is slow a bit but it is looocaaal!
jon martins · 4d
Running gemma 4 on ollama, latest llama.cpp or something else?
Reichard KΓΆnige · 4d
So I should then just try to get the biggest version I can fit into rtx 3060 12gb or get a rtx 3090 24gb