A news organization is exploring the use of generative Al to summarize articles for their readers. They want the summaries to be concise yet compr ensure the generated summaries meet their requirements?

Providing prompts with specific keywords and key phrases

Incorporating human$-$written summaries as prompts for model training

Adjusting the diversity penalty parameter during generation

Fine$-$tuning the model with a diverse dataset of news articles from various sources