Implements dynamic and expandable architectures for continual learning — where networks grow or prune over tasks to balance plasticity and stability. Includes Progressive Neural Networks, Dynamically Expandable Networks (DEN), and PackNet, with utilities for parameter freez
pytorch incremental-learning neural-architecture-search lifelong-learning den continual-learning packnet dynamic-architecture progressive-networks model-expansion
-
Updated
Oct 15, 2025 - Python