Tianlong Li

large language model (5) representation engineering (2) reinforcement learning from human feedback (2) parameter-efficient fine-tuning (2) parameter efficient fine-tuning (2) representation editing (2) neural network interpretability (1) model safety (1) representation learning (1) data privacy (1) reward model (1) distribution shift (1) preference alignment (1) parameter efficient (1) representation alignment (1) jailbreaking attack (1) pre-trained language model (1) soft prompt (1) human preference alignment (1) multi-task learning (1)