psp-dada/LLaVA-v1.6-Vicuna-13B-SENTINEL | ICCV2025 | SENTINEL:
SENTINEL introduces an automatic, sentenceβlevel early intervention strategy to prevent and mitigate object hallucinations in multimodal large language models (MLLMs). Key advantages:
Annotationβfree: No human labeling required.
Model-agnostic: Compatible with any MLLM architecture.
Efficient: Lightweight LoRA fineβtuning.
π§ Early intervention halts hallucination propagation. We find that hallucinations of MLLMs predominantly arise in early sentences and propagate through the rest of the output. SENTINEL interrupts this chain early to maximize mitigation.
π In-domain contextual preference learning without human labels. SENTINEL constructs hallucinated/factual samples via detector cross-validation and builds context-aware preference data without relying on proprietary LLMs or manual annotations.
π‘ Context matters: rich coherence drives robustness. By prioritizing context-coherent positive samples over hallucinated ones, SENTINEL significantly boosts generalization.
β»οΈ Iterative contextual bootstrapping for diverse hallucination-free contexts. Our pipeline dynamically grows non-hallucinated contexts and expands coverage across varied scenes, improving robustness across generations.
π State-of-the-art results across benchmarks. SENTINEL achieves up to 92% reduction in hallucinations and outperforms prior SOTA methods across Object HalBench, AMBER, and HallusionBench, while maintaining or improving general task performance.
This model is a PEFT (LoRA) adapter. You first need to load the base model (llava-hf/llava-v1.6-vicuna-13b-hf) and then load this adapter on top of it.
For the details of this model, please refer to the documentation of the GitHub repo.
If you find our model/code/data/paper helpful, please consider citing our papers π and starring us βοΈοΌ
@article{peng2025mitigating,
title={Mitigating Object Hallucinations via Sentence-Level Early Intervention},
author={Peng, Shangpin and Yang, Senqiao and Jiang, Li and Tian, Zhuotao},
journal={arXiv preprint arXiv:2507.12455},
year={2025}
}
If you have any questions, comments, or suggestions, please do not hesitate to submit an issue or PR to help advance research in this area.
Base model
llava-hf/llava-v1.6-vicuna-13b-hf