Work

• Giving a base model domain reasoning capabilities — Take a base model and improve its reasoning on a specific domain via RLHF/RLVR.