Tag: 蒸馏
All the articles with the tag "蒸馏".
- 5.6
Prune-Quantize-Distill: An Ordered Pipeline for Efficient Neural Network Compression
arXiv:2604.04988v1 Announce Type: cross Abstract: Modern deployment often requires trading accuracy for efficiency under tight CPU and memory const...
- 7.0
Revisiting On-Policy Distillation：实证失败模式与简单修复
重新审视 OPD 在长程场景下的脆弱性，揭示采样 token 变体将分布匹配简化为单 token 信号的系统性问题。
- 7.0
WriteBack-RAG：通过证据蒸馏和回写增强训练知识库
提出将 RAG 系统的知识库视为可训练组件，通过标注样本蒸馏和回写机制持续改进检索质量。