This project focuses on fine-tuning four lightweight language models—Phi-1, Phi-1.5, Phi-2, and TinyL- lama—for novel recipe generation using a recipe dataset sourced from MongoDB. The process involved data preprocessing, ...