A safety realignment framework via subspace-oriented model fusion for large language models

Documenti analoghi

• Unified attacks to large language model watermarks: spoofing and scrubbing in unauthorized knowledge distillation
  di: Yi, Xin, et al.
  Pubblicazione: (2025)
• NLSR: Neuron-Level Safety Realignment of Large Language Models Against Harmful Fine-Tuning
  di: Yi, Xin, et al.
  Pubblicazione: (2024)
• Latent-space adversarial training with post-aware calibration for defending large language models against jailbreak attacks
  di: Yi, Xin, et al.
  Pubblicazione: (2025)
• Hierarchical Divide-and-Conquer for Fine-Grained Alignment in LLM-Based Medical Evaluation
  di: Zheng, Shunfan, et al.
  Pubblicazione: (2025)
• Fine-Grained Detoxification via Instance-Level Prefixes for Large Language Models
  di: Yi, Xin, et al.
  Pubblicazione: (2024)