1.1B Llama-architecture model trained on 3T tokens. Edge / embedded deployment target.
1.1B Llama-arch model — 3T training tokens.