Tag: Regularization
All the papers with the tag "Regularization".
On the Robustness of Reward Models for Language Model Alignment
grok-3-latestScore: 0.63Published: at 06:48本文揭示了奖励模型过优化的根源在于隐藏状态范数分散,并提出批次和为零正则化(BSR)方法,显著提升了奖励模型的分布鲁棒性和 RLHF 对齐效果。
SEFE: Superficial and Essential Forgetting Eliminator for Multimodal Continual Instruction Tuning
grok-3-latestScore: 0.64Published: at 09:09本文提出 SEFE 方法,通过 ASD 范式和 RegLoRA 分别解决多模态持续指令微调中的表面遗忘和本质遗忘问题,显著提升模型性能并实现最先进的遗忘缓解效果。