Advancing Natural Language Formalization to First Order Logic with Fine-tuned LLMs