Support Google Gemma-7b

[moqimoqidea]

Please describe the feature you want
Support the latest Gemma coder models. There are two models: base and instruct tuned.

• https://huggingface.co/google/gemma-7b
• https://huggingface.co/google/gemma-7b-it

llama.cpp already support, see ggml-org/llama.cpp#5631

Please reply with a 👍 if you want this feature.

[jehrhardt]

Google also provides an official interface: https://github.com/google/gemma.cpp

[wsxiaoys]

With #1805 in 0.10, Gemma model will be supported (though we might not adding it to official registry as it's not supporting FIM)

[jdekarske]

CodeGemma has FIM https://ai.google.dev/gemma/docs/formatting#formatting-fim

[wsxiaoys]

Yes - it's tracked in #1807