Few-shot Steerable Alignment: Adapting Rewards and LLM Policies with Neural Processes

Documenti analoghi

• Preference Learning for AI Alignment: a Causal Perspective
  di: Kobalczyk, Katarzyna, et al.
  Pubblicazione: (2025)
• Discovery of Hidden Miscalibration Regimes
  di: Kobalczyk, Katarzyna, et al.
  Pubblicazione: (2026)
• Eliciting Numerical Predictive Distributions of LLMs Without Autoregression
  di: Piskorz, Julianna, et al.
  Pubblicazione: (2026)
• Active Task Disambiguation with LLMs
  di: Kobalczyk, Katarzyna, et al.
  Pubblicazione: (2025)
• Learning Reasoning Rewards from Expert Demonstrations with Inverse Reinforcement Learning
  di: Fanconi, Claudio, et al.
  Pubblicazione: (2025)