Tag
This research paper extends FunctionGemma 270M for practical on-device Android workflows by introducing a synthetic dataset and fine-tuning the model, achieving improved accuracy for function calling while balancing performance and coverage.
A user shares their successful attempt to extend a fine-tuned Gemma model by inserting new layers, overcoming initialization failures to create extGemma4-40_5B without wrecking the original capabilities.
A user extends Google's Gemma4-31B model to 44B parameters with 88 layers, criticizing Google for not releasing larger versions.