PANews reported on December 22nd that, according to the official Google blog, Google released FunctionGemma, a fine-tuned version of Gemma 3 270M specifically designed for function call models, targeting local/offline agent scenarios. Its features include unified chat and tool execution, support for custom fine-tuning (Mobile Actions accuracy improved from 58% to 85%), miniaturization for edge devices (such as NVIDIA Jetson Nano and mobile phones), and optimized JSON/multilingual input. Official downloads for Hugging Face and Kaggle are provided, along with fine-tuning guides for Transformers, Unsloth, Keras, and NeMo, and deployment guides for LiteRT-LM, vLLM, MLX, Llama.cpp, Ollama, Vertex AI, and LM Studio. Multiple demos and datasets/Colabs are also available on the Edge Gallery.



Copy linkX (Twitter)LinkedInFacebookEmail
XRP weakens after repeated price-action fail