[P] I trained a language model from scratch for a low resource language and got it running fully on-device on Android (no GPU, demo)

About this article

Hi Everybody! I just wanted to share an update on a project I’ve been working on called BULaMU, a family of language models trained (20M, 47M, and 110M parameters) trained entirely from scratch for a low resource language, Luganda. The models are small and compute-efficient enough to run offline on a phone without requiring a GPU or internet connection. I recently built an Android app called E.A.S.T. (Expanding Access to Systems of Learning and Intelligence) that allows you to interact with t...