Finetune Google FunctionGemma in NeMo Automodel

[HuiyingLi]

FunctionGemma is a lightweight, 270M-parameter variant built on the Gemma 3 architecture with a function-calling chat format. It is intended to be fine-tuned for task-specific function calling, and its compact size makes it practical for edge or resource-constrained deployments.

• Gemma 3 architecture, updated tokenizer and function-calling chat format.
• Trained specifically for function calling: multiple tool definitions, parallel calls, tool responses, and natural-language summaries.
• Small/edge friendly: ~270M params for fast, dense inference on-device.
• Text-only, function-oriented model (not a general dialogue model), best used after task-specific finetuning.

Data

We use xLAM, a function-calling dataset containing user queries, tool schemas, and tool call traces, to showcase how to finetune FunctionGemma in NeMo Automodel. An example data entry is shown below:

{
  "id": 123,
  "query": "Book me a table for two at 7pm in Seattle.",
  "tools": [
    {
      "name": "book_table",
      "description": "Book a restaurant table",
      "parameters": {
        "party_size": {"type": "int"},
        "time": {"type": "string"},
        "city": {"type": "string"}
      }
    }
  ],
  "answers": [
    {
      "name": "book_table",
      "arguments": "{\"party_size\":2,\"time\":\"19:00\",\"city\":\"Seattle\"}"
    }
  ]
}

We provide a helper make_xlam_dataset converts each xLAM row into OpenAI-style tool schemas and tool calls.

Run SFT and PEFT

Use the following command to finetune FunctionGemma on xLAM. Adjust number of GPUs to your needs.

torchrun --nproc-per-node=8 examples/llm_finetune/finetune.py \
  --config examples/llm_finetune/gemma/functiongemma_xlam.yaml

To apply LoRA (PEFT), uncomment the peft block in the recipe and tune rank/alpha/targets per the SFT/PEFT guide. Example override:

peft:
  _target_: nemo_automodel.components._peft.lora.PeftConfig
  match_all_linear: true
  dim: 16
  alpha: 16
  use_triton: true

For more details, checkout our guide!