2 papers
cs.CL2026
PolyAlign: Conditional Human-Distribution Alignment
L. D. M. S. Sai Teja, Ufaq Khan, Sathira Silva +2
Post-training methods such as supervised fine-tuning (SFT) and preference optimization typically align language models toward a single global assistant behavior. While effective fo…
cs.CL2025
Sequence-level Large Language Model Training with Contrastive Preference Optimization
Zhili Feng, Dhananjay Ram, Cole Hawkins +3
The next token prediction loss is the dominant self-supervised training objective for large language models and has achieved promising results in a variety of downstream tasks. How…