DICE: Structured Reasoning in LLMs through SLM-Guided Chain-of-Thought Correction

Explore the DICE framework, a novel approach that leverages Small Language Models to refine Large Language Model outputs, significantly boosting reasoning ac...

Level: advanced

By Unknown

Category: research