Keshav Ramji

Researcher, IBM Research AI

1 paper at NeurIPS 2025

Homepage· OpenReview· Semantic Scholar· Google Scholar

Poster Session 2

1 paper

Wednesday, December 3, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

Latent Principle Discovery for Language Model Self-Improvement

#1812 · Keshav Ramji, Tahira Naseem, Ramón Fernandez Astudillo

We discover model-generated constitutions and train language models to intrinsically self-correct their responses by using these principles; repeating this process iteratively enables self-improvement.