#machine learning
Jul 2025
ESSA: Evolutionary Strategies for Scalable Alignment
ESSA (Evolutionary Strategies for Scalable Alignment), a gradient-free online alignment stage that follows supervised fine-tuning (SFT) and replaces the gradient loop with inference-only black-box search, is presented.
Daria Korotyshova, B. Shaposhnikov, A. Malakhov et al.
· arXiv.org · 3 citations