DEBATE, TRAIN, EVOLVE: Self Evolution of Language Model Reasoning

Gaurav Srivastava, Zhenyu Bi, Meng Lu +1

Large language models (LLMs) have improved significantly in their reasoning through extensive training on massive datasets. However, relying solely on additional data for improveme…