The proliferation of large language models (LLMs) has significantly advanced intelligent systems. Unfortunately, LLMs often face knowledge conflicts between internal memory and retrieved external information, arising from misinformation, biases, or outdated knowledge. These conflicts undermine response reliability and introduce uncertainty in decision-making. In this work, we analyze how LLMs navigate knowledge conflicts from an information-theoretic perspective and reveal that when conflicting and supplementary information exhibit significant differences, LLMs confidently resolve their preferences and alleviate the uncertainty during their response generation. When this difference is ambiguous, LLMs experience considerable uncertainty about their generation. Based on this insight, we propose Swin-VIB, a novel framework that integrates a pipeline of variational information bottleneck models to adapt the retrieved information difference, facilitating robust response generation of LLMs even in conflicting contexts. Extensive experiments confirm our theoretical analysis and demonstrate the performance of Swin-VIB. Notably, Swin-VIB outperforms all competitive baselines in terms of the accuracy of the multiple-choice task, while improving the EM values in the open-ended QA task by at least 11.14%.

Beyond maximum likelihood estimation (MLE), the standard objective of a language model (LM) that optimizes good examples probabilities, many studies have explored ways that also penalize bad examples for enhancing the quality of output distribution, including unlikelihood training, exponential maximizing average treatment effect (ExMATE), and direct preference optimization (DPO). To systematically compare these methods and further provide a unified recipe for LM optimization, in this paper, we present a unique angle of gradient analysis of loss functions that simultaneously reward good examples and penalize bad ones in LMs. Through both mathematical results and experiments on CausalDialogue and Anthropic HH-RLHF datasets, we identify distinct functional characteristics among these methods. We find that ExMATE serves as a superior surrogate for MLE, and that combining DPO with ExMATE instead of MLE further enhances both the statistical (5-7%) and generative (+18% win rate) performance.

In the rapidly evolving domain of artificial intelligence, safeguarding the intellectual property of Large Language Models (LLMs) is increasingly crucial. Current watermarking techniques against model extraction attacks, which rely on signal insertion in model logits or post-processing of generated text, remain largely heuristic. We propose a novel method for embedding learnable linguistic watermarks in LLMs, aimed at tracing and preventing model extraction attacks. Our approach subtly modifies the LLM's output distribution by introducing controlled noise into token frequency distributions, embedding an statistically identifiable controllable watermark.We leverage statistical hypothesis testing and information theory, particularly focusing on Kullback-Leibler Divergence, to differentiate between original and modified distributions effectively. Our watermarking method strikes a delicate well balance between robustness and output quality, maintaining low false positive/negative rates and preserving the LLM's original performance.