acon96/Home-FunctionGemma-270m
The acon96/Home-FunctionGemma-270m is a 0.3 billion parameter model, fine-tuned from Google's FunctionGemma, designed for smart home automation. It specializes in controlling devices via the "Assist" API and answering basic questions about home state. This model is optimized for low-resource environments, such as Raspberry Pis, through quantization using Lama.cpp.
Loading preview...
Model Overview
The acon96/Home-FunctionGemma-270m is a compact 0.3 billion parameter language model, derived from Google's FunctionGemma. Its core purpose is to facilitate smart home automation by enabling control of devices through the "Assist" API and providing basic question-answering capabilities regarding the home's current state.
Key Capabilities
- Smart Home Control: Directly interacts with home devices via the "Assist" API.
- Contextual Q&A: Answers questions about the provided home's state.
- Resource-Efficient: Quantized using Lama.cpp, making it suitable for deployment in low-resource environments like Raspberry Pis, commonly found in Home Assistant installations.
Training Details
The model was trained using the Axolotl framework. It leveraged the acon96/Home-Assistant-Requests-V2 dataset, specifically designed for home automation interactions. Key hyperparameters included a learning rate of 0.0002, a total batch size of 32, and 597 training steps.
Good For
- Developers building local smart home AI assistants.
- Integrating AI capabilities into Home Assistant setups on embedded devices.
- Applications requiring function calling for device control in resource-constrained environments.