Improving reflection of large language models: Addressing overconfidence and randomness with self-contrast to improve stability and accuracy
LLMs have been at the forefront of recent technological advances, demonstrating notable capabilities in various domains. However, improving the reflective ...