Skip to main content

FunctionGemma 270M IT

A lightweight 270M parameter model from Google, built on Gemma 3 270M and trained for function calling, intended as a foundation for specialized fine-tuned models.

Supported processors

  • SL2619

Performance

Torq Compiler/Runtime v2.3.0 · lm_head

53.4ms
Inference Time
235.8MB
DRAM Footprint

Torq Compiler/Runtime v2.3.0 · transformer

58.727ms
Inference Time
230.4MB
DRAM Footprint

FunctionGemma 270M

Model Overview​

FunctionGemma is a lightweight, open model from Google built on Gemma 3 270M and trained specifically for function calling. It provides a foundation for developing specialized models through further fine-tuning, rather than serving as a general-purpose conversational model. Its compact architecture makes it well suited to adapting function-calling capabilities to specific tasks.

FunctionGemma 270M has been optimized for the Synaptics Astra® SL2610-Series processors with Torq NPU.

Model Features​

  • Model Type: Function-calling language model
  • Input: Text
  • Output: Text, including structured function calls

Deployment​

The compiled model files are available for download on Huggingface at Synaptics/functiongemma-270m-it-torq.

Usage Tutorial​

A usage example is provided at Torq Examples / functiongemma. Check out the README for instructions.

License​

Both the source model and the compiled model for on-device deployment are licensed under Gemma Terms of Use.

Learn More​