A LoRA fine-tuned instruction-following version of Google’s Gemma-3-4b-it model specifically optimized for Sinhala language tasks using the Sinhala FLAN dataset. This model is a Parameter-Efficient Fine-Tuning (PEFT) adaptation of Google’s Gemma-3-4b-it using Low-Rank Adaptation (LoRA) technique. The model has been instruction-tuned on the Sinhala FLAN dataset to improve performance on instruction-following tasks in Sinhala language while maintaining English capabilities. The training focused on enhancing the model’s ability to understand and respond to instructions in Sinhala. — Developed by: Sulakna Weerasinghe, Ovindu Gunathunga, Supun Edirisuriya, Sayuru Bopitiya — Model type: Instruction-tuned Causal Language Model (LoRA Adapter) — Language(s): Sinhala (primary), English (secondary) — License: Apache-2.0 — Finetuned from model: google/gemma-3-4b-it — Base model size: 4B parameters — Adapter parameters: LoRA with rank 16 — Base Repository: google/gemma-3-4b-it — Training Dataset: 0xAIT/sinhala-flan This model is designed for instruction-followin
Модальности:
Генерация текста
Области применения:
Диалог / чат Мультиязычность
Задача: Генерация текста
Автор: iCIIT
Теги: peft, lora, sinhala, multilingual, conversational, si, en
Лайков: 4 | Загрузок: 3
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.