Parameter Alignment Mitigates Catastrophic Forgetting in Multilingual Expert Language Models 文章

ArXiv CS.CL2026-06-02NEWSen作者: Sanchit Ahuja, Terra Blevins
查看原文 →
关系图谱

Parameter Alignment Mitigates Catastrophic Forgetting in Multilingual Expert Language Models · 相关技术

相关技术

catastrophic forgettingmodel mergingPost-hoc Weight ReversionSoft RegularizationHard Layer FreezingContinual Pretraining (CPT)