How can hallucinations be avoided when using language models?
A. By providing factual answers
B. By adding extra prompts
C. By training the models on complete information
D. By avoiding biases in the training data