3 papers
cs.LG2026
Time Without Timesteps: Simulating Coupled Dynamical Systems via Self-Consistency
Liyu Zerihun, Mark Shinyoung Lee
Numerical simulation of dynamical systems is usually organized as a causal march through time: each state is computed from the previous one. We explore a different formulation for…
cs.LG2026
Invertible Memory Flow Networks
Liyu Zerihun, Alexandr Plashchinsky
Long sequence neural memory remains a challenging problem. RNNs and their variants suffer from vanishing gradients, and Transformers suffer from quadratic scaling. Furthermore, com…
cs.LG2025
Estimating the Effective Rank of Vision Transformers via Low-Rank Factorization
Liyu Zerihun
Deep networks are heavily over-parameterized, yet their learned representations often admit low-rank structure. We introduce a framework for estimating a model's intrinsic dimensio…