Since British mathematician Alan Turing proposed the Turing Test in 1950, humanity has been committed to exploring machines capable of passing this test, continuously advancing toward the goal of Artificial General Intelligence (AGI) to build machines with language intelligence comparable to humans. Language is inherently a complex human expression system governed by grammatical rules, making the development of powerful Artificial Intelligence (AI) algorithms that can understand and master language a significant challenge. Over the past two decades, language modeling approaches have been widely used for language understanding and generation, including statistical language models and neural language models.