AI RESEARCH

GeoLAN: Geometric Learning of Latent Explanatory Directions in Large Language Models

arXiv CS.LG

ArXi:2603.19460v1 Announce Type: new Large language models (LLMs) nstrate strong performance, but they often lack transparency. We