A benchmark of online learning algorithms for RNNs in a continual learning setting. The models are evaluated on the task of incrementally learning a repertoire of 20 continuous sequences.
