Fine-tuning llama3-instruct for Arabic Question Answering in the Medical and Mental Health Domain This work presents the fine-tuning of the llama3-instruct model for Arabic question answering in the medical and mental health domain. The approach leverages a custom dataset of Arabic questions and answers collected from medical and mental health websites.
Key aspects:
Model: unsloth/llama-3-8b-Instruct-bnb-4bit Fine-tuning Technique: LORA Dataset: Custom Arabic QA dataset from medical/mental health websites Quantization: Applied for efficiency Results:
The model successfully transitioned from answering solely in English to Arabic after fine-tuning. The fine-tuned model demonstrates good performance in generating relevant and informative answers to Arabic questions within the medical and mental health domain. Applications:
This work can serve as a foundation for building Arabic chatbots for healthcare applications. This approach highlights the effectiveness of fine-tuning large language models like llama3-instruct for domain-specific question answering in Arabic.
- Downloads last month
- 14