Revisiting Catastrophic Forgetting in Large Language Model Tuning | Read Paper on Bytez