Multilingual Health Large Language Model(LLM) for Low Resource Kenyan Languages
Background
Large Language Models(LLMs) are advanced AI models trained on vast amounts of data to understand and generate human-like language. They play a crucial role in many areas of Natural language processing(NLP) such as translation, text generation and summarization. However, most of the existing LLMs are trained on high resource languages leaving low resource languages(LRLs) underrepresented, this possess a significant challenge of language barrier in underserved communities The project aims to develop a multilingual health LLM for low resource Kenyan languages to enhance healthcare accessibility, The model will enable patients to access health information, communicate effectively and receive care in their native language.
Accomplishments
We have managed to fine tune the base model which demonstrates a solid understanding of language and it is able to translate text from English to kikuyu. The model has been trained on a curated dataset of English-kikuyu sentence pairs sourced from African-Next-Voices and JW300.
Next Steps
Future work involves refinement of the model and health data collection to improve the model accuracy and fluency, including other low resource kenyan languages to enhance multilingual support and incorporate multimodality.