#mixedcot
Language‑Mixed Chain‑of‑Thought applied to Korean used a 5.79 million‑prompt Yi‑Sang dataset; the 35B KO‑REAson‑35B model reached an average score of 64.0 on nine multilingual benchmarks. https://getnews.me/language-mixed-chain-of-thought-boosts-korean-ai-reasoning-models/ #mixedcot #koreanai
October 7, 2025 at 11:41 PM