3 papers
cs.LG2026
Fisher-IRG: Fisher-Induced Local Invariant Representation Geometry across Language and Vision Models
Abdullah All Tanvir, Xin Zhong
Semantic-preserving transformations can induce substantial motion in learned representations, while small changes may strongly affect model predictions, raising a basic question: w…
cs.LG2026
Invariant Atoms: Sparse Coordinates of Local Semantic Geometry in Language Model Representations
Muhammad Ahtesham, Xin Zhong
Large language models often preserve meaning despite substantial changes in wording, style, and syntax, while small semantic edits can systematically alter their hidden representat…
cs.LG2026
Invariant Features in Language Models: Geometric Characterization and Model Attribution
Agnibh Dasgupta, Abdullah Tanvir, Xin Zhong
Language models exhibit strong robustness to paraphrasing, suggesting that semantic information may be encoded through stable internal representations, yet the structure and origin…