Combining Domain and Alignment Vectors to Achieve Better Knowledge-Safety Trade-offs in LLMs

Documents similaires

• A Deep Dive into the Trade-Offs of Parameter-Efficient Preference Alignment Techniques
  par: Thakkar, Megh, et autres
  Publié: (2024)
• LLMs Can't Play Hangman: On the Necessity of a Private Working Memory for Language Agents
  par: Baldelli, Davide, et autres
  Publié: (2026)
• Probabilistic Calibration Is a Trainable Capability in Language Models
  par: Baldelli, Davide, et autres
  Publié: (2026)
• Fundamental Safety-Capability Trade-offs in Fine-tuning Large Language Models
  par: Chen, Pin-Yu, et autres
  Publié: (2025)
• Exploring Quantization for Efficient Pre-Training of Transformer Language Models
  par: Chitsaz, Kamran, et autres
  Publié: (2024)