Minlie Huang · page 2
Papers
- Crisp: Cognitive Restructuring of Negative Thoughts through Multi-turn Supportive Dialogues (2025 · 10 citations)
Cognitive Restructuring (CR) is a therapy technique in which a counselor helps someone uncover distorted negative thoughts and reshape them into more balanced ones through conversation. - How Should We Enhance the Safety of Large Reasoning Models: An Empirical Study (2026 · 6 citations)
Large Reasoning Models (LRMs) are strong at maths and code, but their safety often stays the same or gets worse. - IF-CRITIC: Towards a Fine-Grained LLM Critic for Instruction-Following Evaluation (2026 · 1 citation)
LLM-as-a-Judge is often used to supply rewards for training instruction-following. It has two problems. First, it is expensive, because strong proprietary or reasoning models must judge every constraint in a separate call. Second, it is unreliable: these judges miss errors and count poorly. IF-CRITIC is a 14B critic… - IF-RewardBench: Benchmarking Judge Models for Instruction-Following Evaluation (2026 · 1 citation)
Judge models (LLM-as-a-Judge and reward models) are widely used to check whether LLM outputs obey instructions and to supply reward signals for alignment. - Data Efficient RLVR via Off-Policy Influence Guidance (2026)
Choosing training data for reinforcement learning with verifiable rewards (RLVR) usually relies on heuristics such as difficulty or uncertainty. - LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety (2026)
LLMs tend to refuse harmful requests reliably in high-resource languages but are much easier to jailbreak in low-resource ones.
This page in Markdown → https://md.research.elamur.ai/free/en/authors/minlie-huang/page/2/