Fine-tuning an LLM to answer Arxiv Questions about AI/ML