上一条：Bridging the Domain Gaps in Context Representations for <i>k</i>-Nearest Neighbor Neural Machine Translation

下一条：Mitigating Catastrophic Forgetting in Large Language Models with Self-Synthesized Rehearsal