gemma-4-e2b-it-istara-ux-research
Merged Gemma 4 E2B instruction model fine-tuned for UX research workflows in Istara.
Base model: google/gemma-4-E2B-it
Fine-tuning method: LoRA adapter merged into base weights.
Training data blend:
- 732 Alpaca-style UXR examples from
henrique-simoes/ux-research-strategy-dataset
- 3,498 Istara SFT examples generated from live Istara skill definitions and UXR nugget sources
Training hardware: Apple M4 Max via PyTorch MPS.
