Small language models (≤3B params) that can be fine-tuned or run locally for history-focused tasks. Curated for training on Chinese and Brazilian hist