First numerical observation of the Berezinskii-Kosterlitz-Thouless transition in language models (2412.01212v1)

Abstract: Several power-law critical properties involving different statistics in natural languages -- reminiscent of scaling properties of physical systems at or near phase transitions -- have been documented for decades. The recent rise of LLMs has added further evidence and excitement by providing intriguing similarities with notions in physics such as scaling laws and emergent abilities. However, specific instances of classes of generative LLMs that exhibit phase transitions, as understood by the statistical physics community, are lacking. In this work, inspired by the one-dimensional Potts model in statistical physics we construct a simple probabilistic LLM that falls under the class of context sensitive grammars (CSG), and numerically demonstrate an unambiguous phase transition in the framework of a natural LLM. We explicitly show that a precisely defined order parameter -- that captures symbol frequency biases in the sentences generated by the LLM -- changes from strictly 0 to a strictly nonzero value (in the infinite-length limit of sentences), implying a mathematical singularity arising when tuning the parameter of the stochastic LLM we consider. Furthermore, we identify the phase transition as a variant of the Berezinskii-Kosterlitz-Thouless (BKT) transition, which is known to exhibit critical properties not only at the transition point but also in the entire phase. This finding leads to the possibility that critical properties in natural languages may not require careful fine-tuning nor self-organized criticality, but is generically explained by the underlying connection between language structures and the BKT phases.