Data Mixing Optimization for Supervised Fine-Tuning of Large Language Models | Read Paper on Bytez