AIBullisharXiv – CS AI · 7h ago6/10
🧠
To Intervene or Not: Guiding Inference-time Alignment with Probabilistic Model Blending
Researchers introduce BlendIn, an inference-time alignment framework for large language models that uses probabilistic model blending instead of binary intervention decisions. The method dynamically weights guidance from multiple models based on reliability, achieving up to 50% performance improvement by reducing ineffective interventions that typically degrade output quality.