AI Paper Atlas

Research map / Benchmarks / CHAIR

CHAIR leaderboard

17 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on CHAIR. Best CHAIR_S: 16.8 (ILVAD, ICML 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

CHAIR · CHAIR_S

15 papers, each with its best reported number for its own method; lower is better.

#CHAIR_S ↓MethodPaperTopic
116.8ILVAD
InternVL3-8B
Finding the Correct Visual Evidence Without Forgetting: Mitigating Hallucination in LVLMs via Inter-Layer Visual Attention Discrepancy
ICML 2026 · Table 1
Multimodal LLMs
217KVSmooth
MiniGPT-4 7B
KVSmooth: Mitigating Hallucination in Multi-modal Large Language Models through Key-Value Smoothing
CVPR 2026 · Table 1
Multimodal LLMs
319.3+SGRS+LocoRE
Qwen2-VL-7B
Hallucination Begins Where Saliency Drops
ICLR 2026 · Table 2
Multimodal LLMs
425.8ResDec
Qwen2.5-VL-Instruct (7B)
Residual Decoding: Mitigating Hallucinations in Large Vision-Language Models via History-Aware Residual Guidance
CVPR 2026 · Table 2
Multimodal LLMs
526.2MGAP (Ours)
LLaVA v1.5 7B
Mitigating Manifold Departure: Uncertainty-Aware Subspace Rectification for Trustworthy MLLM Decoding
ICML 2026 · Table 2
Multimodal LLMs
627.1MoSVE
Vicuna-7B
Mixture of Specialized Vision Experts: Unlocking Complementary Visual Insights for Faithful MLLM Reasoning
ECCV 2026 · Table 1
Multimodal LLMs
730.8DMAS (Ours)
LLaVAv1.5 7B
Dynamic Multimodal Activation Steering for Hallucination Mitigation in Large Vision-Language Models
ICLR 2026 · Table 3
Multimodal LLMs
835.8VTI
LLaVA-1.5
Reducing Hallucinations in Large Vision-Language Models via Latent Space Steering
ICLR 2025 · Table 2
Multimodal LLMs
942.4Fox
InstructBLIP
Dismantling Pathological Shortcuts: A Causal Framework for Faithful LVLM Decoding
ICML 2026 · Table 2
Multimodal LLMs
1043ED
LLaVA-1.5 7B
Do You Keep an Eye on What I Ask? Mitigating Multimodal Hallucination via Attention-Guided Ensemble Decoding
ICLR 2025 · Table: CHAIR results
Multimodal LLMs
5 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

CHAIR · CHAIR_I

13 papers, each with its best reported number for its own method; lower is better.

#CHAIR_I ↓MethodPaperTopic
13.3LLaVA-1.5-7B + Antidote
LLaVA-1.5-7B
Antidote: A Unified Framework for Mitigating LVLM Hallucinations in Counterfactual Presupposition and Object Perception
CVPR 2025 · Table 1
Multimodal LLMs
26.8ResDec
Qwen2.5-VL-Instruct (7B)
Residual Decoding: Mitigating Hallucinations in Large Vision-Language Models via History-Aware Residual Guidance
CVPR 2026 · Table 2
Multimodal LLMs
38.2SGRS + LocoRE
LLaVA-1.5-7B
Hallucination Begins Where Saliency Drops
ICLR 2026 · Table 1
Multimodal LLMs
48.8ACD
Qwen3-VL-4B
Beyond Logits: Coherent Hallucination Mitigation via Attention Contrastive Decoding
ICML 2026 · Table 4
Multimodal LLMs
59.42ILVAD
LLaVA-1.5-7B
Finding the Correct Visual Evidence Without Forgetting: Mitigating Hallucination in LVLMs via Inter-Layer Visual Attention Discrepancy
ICML 2026 · Table 1
Multimodal LLMs
611.1VTI
LLaVA-1.5
Reducing Hallucinations in Large Vision-Language Models via Latent Space Steering
ICLR 2025 · Table 2
Multimodal LLMs
711.2PND
InstructBLIP-7B
Breaking the Illusion: When Positive Meets Negative in Multimodal Decoding
CVPR 2026 · Table 3
Multimodal LLMs
811.4DMAS (Ours)
LLaVAv1.5 7B
Dynamic Multimodal Activation Steering for Hallucination Mitigation in Large Vision-Language Models
ICLR 2026 · Table 3
Multimodal LLMs
911.98Fox
InstructBLIP
Dismantling Pathological Shortcuts: A Causal Framework for Faithful LVLM Decoding
ICML 2026 · Table 2
Multimodal LLMs
1012.2MoSVE
Vicuna-7B
Mixture of Specialized Vision Experts: Unlocking Complementary Visual Insights for Faithful MLLM Reasoning
ECCV 2026 · Table 1
Multimodal LLMs
3 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.