Interactive Memory Learning for Long-Term Conversations
InteraCtive Memory Learning is proposed, a multi-agent framework that transforms the memory mechanism from a passive archive into a learnable, interactive memory policy that significantly outperforms strong baselines.