DynaMoE: Dynamic Token-Level Expert Activation with Layer-Wise Adaptive Capacity for Mixture-of-Experts Neural Networks Paper • 2603.01697 • Published 5 days ago • 2
DynaMoE: Dynamic Token-Level Expert Activation with Layer-Wise Adaptive Capacity for Mixture-of-Experts Neural Networks Paper • 2603.01697 • Published 5 days ago • 2
Gabliteration: Adaptive Multi-Directional Neural Weight Modification for Selective Behavioral Alteration in Large Language Models Paper • 2512.18901 • Published Dec 21, 2025 • 3