Bochen Lyu

MoDr: Mixture-of-Depth-Recurrent Transformers for Test-Time Reasoning

Dissecting the Role of Positional Encoding in Length Generalization

Transformers with RL or SFT Provably Learn Sparse Boolean Functions, But Differently

Heavy-Ball Momentum Method in Continuous Time and Discretization Error Analysis

A Solvable Attention for Neural Scaling Laws

DyCAST: Learning Dynamic Causal Structure from Time Series

Continuous Approximation of Momentum Methods with Explicit Discretization Error