Griot Edge: Smaller Speech Models, Better Questions
60% fewer parameters, selectable streaming lookahead, and an Akan result that changed what I want to work on next.
Writing about code and things I find interesting.
Welcome. I write about software engineering (Mostly Machine Learning). I am very opinionated like all devs are.
Links 👇
Latest Writing
View all posts →Building Griot Nano 1: Lessons from 44,000 Hours of Multilingual Speech Training
How we built and evaluated an open-source, locally deployable speech-recognition model for Akan, Dagbani, Ewe, Ghanaian English, and Ga.
Building a C++ Deep Learning Framework from Scratch: My Journey with CUDA
Building a PyTorch-like deep learning framework from the ground up.