Iterative Foundation Model Fine-Tuning on Multiple Rewards | Read Paper on Bytez