# MoD-DPO: Towards Mitigating Cross-modal Hallucinations in Omni LLMs using Modality Decoupled Preference Optimization

arXiv: https://arxiv.org/abs/2603.03192 · HTML: https://mlsanity.com/paper/2603.03192/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion and multimodal model alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 49 of its 57 references that are in the map.
- First posted on arXiv 3 Mar 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Diffusion and multimodal model alignment](https://mlsanity.com/alignment/diffusion-and-multimodal-model-alignment/)

## Builds on (most cited first, 30 of 49)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [Evaluating Object Hallucination in Large Vision-Language Models](https://mlsanity.com/paper/2305.10355/) (2023)
- [LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models](https://mlsanity.com/paper/2403.13372/) (2024)
- [MVBench: A Comprehensive Multi-modal Video Understanding Benchmark](https://mlsanity.com/paper/2311.17005/) (2023)
- [MiniCPM-V: A GPT-4V Level MLLM on Your Phone](https://mlsanity.com/paper/2408.01800/) (2024)
- [VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs](https://mlsanity.com/paper/2406.07476/) (2024)
- [Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding](https://mlsanity.com/paper/2311.16922/) (2023)
- [Qwen3-Omni Technical Report](https://mlsanity.com/paper/2509.17765/) (2025)
- [Video-R1: Reinforcing Video Reasoning in MLLMs](https://mlsanity.com/paper/2503.21776/) (2025)
- [OPERA: Alleviating Hallucination in Multi-Modal Large Language Models via Over-Trust Penalty and Retrospection-Allocation](https://mlsanity.com/paper/2311.17911/) (2023)
- [RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback](https://mlsanity.com/paper/2312.00849/) (2023)
- [Recognize Anything: A Strong Image Tagging Model](https://mlsanity.com/paper/2306.03514/) (2023)
- [MMAU: A Massive Multi-Task Audio Understanding and Reasoning Benchmark](https://mlsanity.com/paper/2410.19168/) (2024)
- [Detecting and Preventing Hallucinations in Large Vision Language Models](https://mlsanity.com/paper/2308.06394/) (2023)
- [Woodpecker: Hallucination Correction for Multimodal Large Language Models](https://mlsanity.com/paper/2310.16045/) (2023)
- [Audio Flamingo 3: Advancing Audio Intelligence with Fully Open Large Audio Language Models](https://mlsanity.com/paper/2507.08128/) (2025)
- [VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interaction](https://mlsanity.com/paper/2501.01957/) (2025)
- [HALC: Object Hallucination Reduction via Adaptive Focal-Contrast Decoding](https://mlsanity.com/paper/2403.00425/) (2024)
- [Direct Preference Optimization of Video Large Multimodal Models from Language Model Reward](https://mlsanity.com/paper/2404.01258/) (2024)
- [VALOR: Vision-Audio-Language Omni-Perception Pretraining Model and Dataset](https://mlsanity.com/paper/2304.08345/) (2023)
- [Daily-Omni: Towards Audio-Visual Reasoning with Temporal Alignment across Modalities](https://mlsanity.com/paper/2505.17862/) (2025)
- [Human-centric Dialog Training via Offline Reinforcement Learning](https://mlsanity.com/paper/2010.05848/) (2020)
- [mDPO: Conditional Preference Optimization for Multimodal Large Language Models](https://mlsanity.com/paper/2406.11839/) (2024)
- [MM-RLHF: The Next Step Forward in Multimodal LLM Alignment](https://mlsanity.com/paper/2502.10391/) (2025)
- [Unified Hallucination Detection for Multimodal Large Language Models](https://mlsanity.com/paper/2402.03190/) (2024)
- [Open-Set Image Tagging with Multi-Grained Text Supervision](https://mlsanity.com/paper/2310.15200/) (2023)
- [V-DPO: Mitigating Hallucination in Large Vision Language Models via Vision-Guided Direct Preference Optimization](https://mlsanity.com/paper/2411.02712/) (2024)

## Built on by (most cited first, 6 of 6)

- [Buried in Textual Debt: Context Pruning with Visual Evidence Preservation for MLLM Agents](https://mlsanity.com/paper/2608.22963/) (2026)
- [Seeing What Should Be Heard: Diagnosing and Repairing Cross-Modal Shortcuts in Omni-Modal LLMs](https://mlsanity.com/paper/2609.36798/) (2026)
- [Devils in Question Relay: Source-Conditioned Relay Steering to Mitigate Hallucinations in Audio-visual Large Language Models](https://mlsanity.com/paper/2609.37568/) (2026)
- [Relevant Evidence Decoding for Audio-Visual Hallucination Mitigation](https://mlsanity.com/paper/2610.02976/) (2026)
- [OmniConfess: Eliciting Token Confessions to Mitigate Omni-Modal Hallucination](https://mlsanity.com/paper/2610.02999/) (2026)
- [Mitigating Cross-Modal Hallucinations by Correcting Modality Misrouting in Omnimodal LLMs](https://mlsanity.com/paper/2610.18141/) (2026)

## Abstract (arXiv)

Omni-modal large language models (omni LLMs) have recently achieved strong performance across audiovisual understanding tasks, yet they remain highly susceptible to cross-modal hallucinations arising from spurious correlations and dominant language priors. In this work, we propose Modality-Decoupled Direct Preference Optimization (MoD-DPO), a simple and effective framework for improving modality grounding in omni LLMs. MoD-DPO introduces modality-aware regularization terms that explicitly enforce invariance to corruptions in irrelevant modalities and sensitivity to perturbations in relevant modalities, thereby reducing unintended cross-modal interactions. To further mitigate over-reliance on textual priors, we incorporate a language-prior debiasing penalty that discourages hallucination-prone text-only responses. Extensive experiments across multiple audiovisual hallucination benchmarks demonstrate that MoD-DPO consistently improves perception accuracy and hallucination resistance, outperforming previous preference optimization baselines under similar training budgets. Our findings underscore the importance of modality-faithful alignment and demonstrate a scalable path toward more reliable and resilient multimodal foundation models.