15 papers
Rubric Judgments Lack Uncertainty: Conformal Weighting for LLM Evaluation
Licheng Pan, Haocheng Yang, Haoxuan Li +7
Rubric-based evaluation is widely used to assess large language model (LLM) outputs by decomposing quality into multiple criteria and assigning each a binary judgment from an LLM v…
CleanRM: Reward Modeling from Noisy Feedback with a Few Clean Feedback for RLHF
Licheng Pan, Haocheng Yang, Yuzhou Chen +7
Feedback noise is pervasive in preference data, arising from subjective human judgments, annotation mistakes, and imperfect LLM judges. Standard reward modeling trained on such fee…
RM-OPD: On-Policy Distillation for Generative Reward Models
Qinglin Su, Licheng Pan, Zhen Fang +5
Reinforcement Learning on Mixed-domain preference data (Mix-RL) has become the mainstream approach for training general-purpose Generative Reward Models (GRMs). However, Mix-RL suf…
Mitigating Cross-Modal Hallucinations by Correcting Modality Misrouting in Omnimodal LLMs
Yu Mi, Licheng Pan, Jiayi Zhu +7
Omnimodal large language models (OLLMs) are susceptible to cross-modal hallucinations under audio-visual interference. We find that, for a substantial fraction of incorrect joint p…
Uncertainty-Aware Reward Modeling for Stable RLHF
Licheng Pan, Haocheng Yang, Haoxuan Li +7
Reinforcement learning from human feedback (RLHF) aligns large language models by training reward models on preference data and optimizing policies to maximize predicted rewards. H…
Inducing Overthink: Hierarchical Genetic Algorithm-based DoS Attack on Black-Box Large Language Reasoning Models
Shuqiang Wang, Wei Cao, Jiaqi Weng +4
Large Reasoning Models (LRMs) are increasingly integrated into systems requiring reliable multi-step inference, yet this growing dependence exposes new vulnerabilities related to c…
Optimal Transport for LLM Reward Modeling from Noisy Preference
Licheng Pan, Haochen Yang, Haoxuan Li +8
Reward models are fundamental to Reinforcement Learning from Human Feedback (RLHF), yet real-world datasets are inevitably corrupted by noisy preference. Conventional training obje…
Robust Reward Modeling for Large Language Models via Causal Decomposition
Yunsheng Lu, Zijiang Yang, Licheng Pan +1
Reward models are central to aligning large language models, yet they often overfit to spurious cues such as response length and overly agreeable tone. Most prior work weakens thes…
Deep Time-Series Forecasting in 10 Years: A Survey
Hao Wang, Licheng Pan, Qingsong Wen +12
Autocorrelation is a common property of time-series, where each observation is dependent on its predecessors. In deep time-series forecasting, it raises two central challenges: (1)…
CausalRM: Causal-Theoretic Reward Modeling for RLHF from Observational User Feedbacks
Hao Wang, Licheng Pan, Zhichao Chen +7
Despite the success of reinforcement learning from human feedback (RLHF) in aligning language models, current reward modeling heavily relies on experimental feedback data collected…
Deep Time-series Forecasting Needs Kernelized Moment Balancing
Licheng Pan, Hao Wang, Haocheng Yang +12
Deep time-series forecasting can be formulated as a distribution balancing problem aimed at aligning the distribution of the forecasts and ground truths. According to Imbens' crite…
Lowest Span Confidence: Zero-Shot Hallucination Detection from a Single LLM Response
Yitong Qiao, Licheng Pan, Yu Mi +6
Hallucinations in Large Language Models (LLMs), i.e., plausible but non-factual generations, pose a significant challenge to reliable deployment in high-stakes environments. Howeve…
Quadratic Direct Forecast for Training Multi-Step Time-Series Forecast Models
Hao Wang, Licheng Pan, Yuan Lu +7
The design of learning objectives is central to training time-series forecasting models. Existing learning objectives such as mean squared error mostly treat each future step as an…
DistDF: Time-Series Forecasting Needs Joint-Distribution Wasserstein Alignment
Hao Wang, Licheng Pan, Yuan Lu +7
Training time-series forecasting models requires aligning the conditional distribution of model forecasts with that of the label sequence. The standard direct forecast (DF) approac…
Understanding and Mitigating Overrefusal in LLMs from an Unveiling Perspective of Safety Decision Boundary
Licheng Pan, Yongqi Tong, Xin Zhang +3
Large language models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks, yet they often refuse to answer legitimate queries--a phenomenon known as overr…