AI RESEARCH
LASER: Low-Rank Activation SVD for Efficient Recursion
arXiv CS.LG
•
ArXi:2604.17224v1 Announce Type: new Recursive architectures such as Tiny Recursive Models (TRMs) perform implicit reasoning through iterative latent computation, yet the geometric structure of these reasoning trajectories remains poorly understood. We investigate the activation manifold of TRMs during recursive unrolling and find that activations occupy an effectively linear, low-dimensional subspace whose principal directions can be tracked dynamically with cheap power iterations.