This research compares the effectiveness of prompt engineering versus fine-tuning approaches for adapting large language models to perform code-related tasks.

• Evaluates different strategies for querying LLMs like ChatGPT
• Assesses the performance benefits of fine-tuning pre-trained models like CodeBERT
• Provides empirical evidence to guide engineering decisions between these two approaches
• Offers practical insights for implementing LLMs in software development workflows

For engineering teams, this research offers critical guidance on selecting the most effective and resource-efficient approach when implementing AI solutions for code generation, summarization, and translation tasks.

Prompt Engineering or Fine-Tuning: An Empirical Assessment of LLMs for Code