Self-consistency improves chain of thought reasoning in language models. What aspect of language model performance benefits from self-consistency?
a) Coherence
b) Fluency
c) Creativity
d) Accuracy