LSTMem: Hierarchical Long Short-Term Online Memory for Large Language Models
LSTMem is proposed, an LSTM-inspired online memory that equips each layer of a frozen LLM with two matrix-valued states: a cell state that accumulates history and a hidden state whose readouts correct the backbone's attention.
Xiang-Long Shi, Rui-Jie Yang, Si-Rui Zhao et al.
· 0 citations