Taeyoung Yun

Diffusion Alignment as Variational Expectation-Maximization

Active Attacks: Red-teaming LLMs via Adaptive Environments

Adaptive teachers for amortized samplers

Posterior Inference with Diffusion Models for High-dimensional Black-box Optimization

Improved Off-policy Reinforcement Learning in Biological Sequence Design

Improved Off-policy Reinforcement Learning in Biological Sequence Design