mlsanityData to 14 Oct 2026

ReLoop: "Seeing Twice and Thinking Backwards" via Closed-loop Training to Mitigate Hallucinations in Multimodal understanding

arXiv 2507.04943 ↗First posted 7 Jul 2025cs.CV cs.CLcited by 1

On the map: LLM reasoning → Hallucination detection and mitigation

also in 1 more
Abstract (arXiv)

While Multimodal Large Language Models (MLLMs) have achieved remarkable progress in open-ended visual question answering, they remain vulnerable to hallucinations. These are outputs that contradict or misrepresent input semantics, posing a critical challenge to the reliability and factual consistency. Existing methods often rely on external verification or post-hoc correction, lacking an internal mechanism to validate outputs directly during training. To bridge this gap, we propose ReLoop, a unified closed-loop training framework that encourages multimodal consistency for cross-modal understanding in MLLMs. ReLoop adopts a ring-shaped structure that integrates three complementary consistency feedback mechanisms, obliging MLLMs to "seeing twice and thinking backwards". Specifically, ReLoop employs the frozen Consistency Feedback Plugin (CFP), comprising semantic reconstruction, visual description, and an attention supervision module for attention alignment. These components collectively enforce semantic reversibility, visual consistency, and interpretable attention, enabling the model to correct its outputs during training. Extensive evaluations and analyses demonstrate the effectiveness of ReLoop in reducing hallucination rates across multiple benchmarks, establishing a robust method for hallucination mitigation in MLLMs. We will release our source code and data in the camera-ready version.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Hallucination detection and mitigation, one of 17 directions in LLM reasoning. Builds on 25 of its 38 references that are in the map. First posted on arXiv 7 Jul 2025.

Builds on

25 of 38 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025Medical Hallucinations in Foundation Models and Their Impact on Healthcare26 Feb

Other references · 24

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  3. 2023Visual Instruction Tuningcited by 7,869
  4. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  5. 2023Improved Baselines with Visual Instruction Tuningcited by 4,418
  6. 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,839
  7. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Modelscited by 1,652
  8. 2023Aligning Large Multimodal Models with Factually Augmented RLHFcited by 712
  9. 2023Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decodingcited by 688
  10. 2023HallusionBench: An Advanced Diagnostic Suite for Entangled Language Hallucination and Visual Illusion in Large Vision-Language Modelscited by 583
  11. 2023OPERA: Alleviating Hallucination in Multi-Modal Large Language Models via Over-Trust Penalty and Retrospection-Allocationcited by 478
  12. 2023AMBER: An LLM-free Multi-dimensional Benchmark for MLLMs Hallucination Evaluationcited by 300
12 more
  1. 2023Woodpecker: Hallucination Correction for Multimodal Large Language Modelscited by 226
  2. 2023Beyond Hallucinations: Enhancing LVLMs through Hallucination-Aware Direct Preference Optimizationcited by 225
  3. 2024Aligning Modalities in Vision Large Language Models via Preference Fine-tuningcited by 198
  4. 2024ReDeEP: Detecting Hallucination in Retrieval-Augmented Generation via Mechanistic Interpretabilitycited by 101
  5. 2024A Survey on Benchmarks of Multimodal Large Language Modelscited by 82
  6. 2025TACO: Enhancing Multimodal In-context Learning via Task Mapping-Guided Sequence Configurationcited by 23
  7. 2025EAGLE: Enhanced Visual Grounding Minimizes Hallucinations in Instructional Multimodal Modelscited by 8
  8. 2024Token Preference Optimization with Self-Calibrated Visual-Anchored Rewards for Hallucination Mitigationcited by 7
  9. 2024NUS-Emo at SemEval-2024 Task 3: Instruction-Tuning LLM for Multimodal Emotion-Cause Analysis in Conversationscited by 5
  10. 2025Hybrid Retrieval for Hallucination Mitigation in Large Language Models: A Comparative Analysiscited by 5
  11. 2024On the Limits of Language Generation: Trade-Offs Between Hallucination and Mode Collapsecited by 2
  12. 2025Feedback-Enhanced Hallucination-Resistant Vision-Language Model for Real-Time Scene Understandingcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026LLM Reasoning Is Latent, Not the Chain of Thoughtcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025When Images Speak Louder: Mitigating Language Bias-induced Hallucinations in VLMs through Cross-Modal Guidance
  2. 2024Combating Multimodal LLM Hallucination via Bottom-Up Holistic Reasoning
  3. 2025Reallocating Attention Across Layers to Reduce Multimodal Hallucination
  4. 2024Investigating and Mitigating the Multimodal Hallucination Snowballing in Large Vision-Language Models
  5. 2025Modality Bias in LVLMs: Analyzing and Mitigating Object Hallucination via Attention Lens
  6. 2025Attention Reallocation: Towards Zero-cost and Controllable Hallucination Mitigation of MLLMs
  7. 2026Mitigating Object Hallucinations in Vision-Language Models through Region-Aware Attention Recalibration
  8. 2026Look Carefully: Adaptive Visual Reinforcements in Multimodal Large Language Models for Hallucination Mitigation