A new way of writing the Yoruba language. Introducing the Kikọ script. More details can be found through the profile link.

I am excited to be presenting MAGNET 🧲at NeurIPS 2024 next week. Subword tokenizers have been shown to overly segment text in non-Latin script languages. Our work presents an approach to train tokenizer-free multilingual LMs via efficient byte-level modeling. 1/n

Introducing Sagalee: an Open Source ASR Dataset for Oromo language. Happy to share that our work on Sagalee is accepted for IEEE ICASSP 2025! 🎉 I will be attending the conference in April. Link to paper and dataset👇

Fantastic work

Oreen Yousuf is doing fantastic work with minority scripts - read more in this interview via SEI. sei.berkeley.edu/2025/07/01/ore…

We are seeking applicants for our 2025 Language Revitalization Cohort! Work within your community to design and implement online language learning lessons using free language learning technology tools. Learn more at our link in bio!