4 papers
Black Box Absorption: LLMs Undermining Innovative Ideas
Wenjun Cao
Large Language Models are increasingly adopted as critical tools for accelerating innovation. This paper identifies and formalizes a systemic risk inherent in this paradigm: \textb…
The Alignment Bottleneck
Wenjun Cao
Large language models improve with scale, yet feedback-based alignment still exhibits systematic deviations from intended behavior. Motivated by bounded rationality in economics an…
Fight Fire with Fire: Defending Against Malicious RL Fine-Tuning via Reward Neutralization
Wenjun Cao
Reinforcement learning (RL) fine-tuning transforms large language models while creating a vulnerability we experimentally verify: Our experiment shows that malicious RL fine-tuning…
Dynamic Action Interpolation: A Universal Approach for Accelerating Reinforcement Learning with Expert Guidance
Wenjun Cao
Reinforcement learning (RL) suffers from severe sample inefficiency, especially during early training, requiring extensive environmental interactions to perform competently. Existi…