Enhanced Fine-Tuning with In-Context Learning for Better LLM Generalization

ICL outperforms fine-tuning in generalization; augmented fine-tuning combines both for improved LLM adaptation.