Continual Learning: Three Approaches That Work

Three approaches to continual learning in AI: reinforcement learning, LoRAs and differentiable plasticity, plus what a fruit fly reveals about weights.