Search papers, labs, and topics across Lattice.
1
0
Dynamical isometry can be the key to preserving neural network plasticity in continual learning, enabling models to maintain performance without sacrificing expressiveness.