QAT Models supported?

Thanks for the wonderful repo. It's a pleasure to work with it.

Are the new 4-bit quantized models for gemma 3 supported by gemma.cpp? Judging from the press-release https://developers.googleblog.com/en/gemma-3-quantized-aware-trained-state-of-the-art-ai-to-consumer-gpus/ and looking through the gemma.cpp repo, it doesn't seem like it, but I am bit surprised about it.