Model
Luganda-gemma-1B-it
A Gemma 3 1B Instruct model adapted using QLoRA for English to Luganda and Luganda to English translation.
- Luganda
- Translation
- Open Model
About this model
Advancing English and Luganda translation
Built on Google's Gemma 3 1B Instruct architecture, this model uses QLoRA fine-tuning to improve English-to-Luganda and Luganda-to-English translation. It was trained primarily on the Sunbird SALT parallel corpus, covering agriculture, health, and society, adapting the base model with just 13.05 million trainable parameters in a 52 MB adapter that runs on consumer GPUs.
- English to Luganda translation
- Luganda to English translation
- Luganda conversational AI
- Educational tools for Luganda language learning
- Low-resource African language NLP research
Model Information
- Language
- Luganda
- Task
- Translation
- Parameters
- 1B · 13.05M trained
- Method
- QLoRA · 4-bit NF4 · r16
- Direction
- en → lg, lg → en
- License
- Apache-2.0 adapter · Gemma terms (base)
- Released
- May 23, 2026
- Base Model
- google/gemma-3-1b-it ↗
- Available on
- Hugging Face, Ollama, Kaggle
Playground
Try the Model
Explore how the model translates civic discourse between English and Luganda
Luganda-gemma-1B-it
English0/500
Luganda
The translation will appear here