A development team reports that their custom-trained Large Language Model (LLM) is "hallucinating"
- generating factually incorrect or nonsensical information, especially when asked questions outside the scope of its training data. The model was created by fine-tuning a foundation model on a large but static internal dataset. The team wants to improve the model's factual accuracy and reliability without embarking on a new, large-scale training project.
Which are the most appropriate strategies to mitigate this issue? (Choose 2.)