Xinyuan Wang

large language model (4) adversarial attack (2) contrastive learning (2) model selection (1) text generation (1) model security (1) continual learning (1) self-supervised learning (1) in-context learning (1) kl divergence (1) importance sampling (1) dynamic routing (1) natural language understanding (1) feature transformation (1) reinforcement learning from human feedback (1) embedding space (1) low-rank estimation (1) reinforcement learning (1) diffusion model (1) unsupervised learning (1)