
ZS
Zhuo-Yang Song, Zeyu Li, Qing-Hong Cao, Ming-xing Luo, Hua Xing Zhu
· 1 min read
ResearcharXiv cs.LG
A phase diagram for training and inference in language models
arXiv:2611.19848v1 Announce Type: new
Abstract: A language model is a complex system whose behavior is determined jointly by the Transformer and the corpus, yet how the representation geometry in the Transformer relates to the actual behavior of language models remains open. We analyze how the representation geometry changes along the forward propagation of actual language models and characterize the representation geometry of a document at a given layer by two collective coordinates, document distinguishability and document dimension. On the phase diagram spanned by these collective coordinates, inference is represented as a flow along depth. We find that, for untrained models, depending on their initialization scale, most of the 156 inference flows move toward the collapse and noise regions of the phase diagram. By contrast, the inference flows of 8 trained models from 5 architecture families all enter and reside in the central region of the phase diagram. Describing the evolution of the flows with a discrete Langevin equation, we find that the difference between the inference flows of untrained runs and those of trained models reduces to whether the effective force on the flows in this region is of saddle or stable-node type. On Pythia-1.4B, we interpret the training process as a phase transition. These results give a unified gateway to the relation between representation geometry and language-model behavior.
Original source
This story was published by arXiv cs.LG and written by Zhuo-Yang Song, Zeyu Li, Qing-Hong Cao, Ming-xing Luo, Hua Xing Zhu. SyncAI.news shows a preview; the complete article is on the publisher's site.
Read the full story on arxiv.org


