Revisiting Softmax Masking: Stop Gradient for Enhancing Stability in Replay-based Continual Learning

Open in new window