Federico Danieli

ParaRNN: Unlocking Parallel Training of Nonlinear RNNs for Large Language Models

GenCtrl -- A Formal Controllability Toolkit for Generative Models

Normalized Rewards for Preference Learning

Attention to Mamba: A Recipe for Cross-Architecture Distillation

Understanding Input Selectivity in Mamba: Impact on Approximation Power, Memorization, and Associative Recall Capacity

Theory, Analysis, and Best Practices for Sigmoid Self-Attention