3 papers
cs.LG2026
Probe-Geometry Alignment: Erasing the Cross-Sequence Memorization Signature Below Chance
Anamika Paul Rupa, Anietie Andy
Recent attacks show that behavioural unlearning of large language models leaves internal traces recoverable by adversarial probes. We characterise where this retention lives and sh…
cs.LG2026
Neural Collapse Dynamics: Depth, Activation, Regularisation, and Feature Norm Threshold
Anamika Paul Rupa
Neural collapse (NC) -- the convergence of penultimate-layer features to a simplex equiangular tight frame -- is well understood at equilibrium, but the dynamics governing its onse…
cs.LG2026
A Systematic Empirical Study of Grokking: Depth, Architecture, Activation, and Regularization
Shalima Binta Manir, Anamika Paul Rupa
Grokking the delayed transition from memorization to generalization in neural networks remains poorly understood, in part because prior empirical studies confound the roles of arch…