AI Paper Atlas

Research map / Benchmarks / POPE

POPE leaderboard

171 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on POPE. Best accuracy: 90.8 (+ HERO, ECCV 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.

POPE · accuracy

97 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
190.8+ HERO
Qwen2.5-VL-7B
HERO: Enhancing Multimodal Faithfulness via Dynamic Entropy-Aware Reinforcement Learning
ECCV 2026 · Table 2
Multimodal LLMs
290.8SPHINX-1k
LLaMA-2-13B
"SPHINX: A Mixer of Weights, Visual Embeddings and Image Scales for Multi-modal Large Language Models"
ECCV 2024 · Table 1
Multimodal LLMs
390.1AD-Loop (Ours)
BAGEL-7B
Synergizing Understanding and Generation with Interleaved Analyzing-Drafting Thinking
ICLR 2026 · Table 1
Multimodal LLMs
490.1InternVL3-14B (MUG)
InternVL3-14B
Multi-Agent Undercover Gaming: Hallucination Removal Through Counterfactual Test for Multimodal Reasoning
AAAI 2026 · Table 1
LLM agents
589.81UniFlow-XL
Qwen2.5-7B
UniFlow: A Unified Pixel Flow Tokenizer for Visual Understanding and Generation
ICLR 2026 · Table 2
Multimodal LLMs
689.6UniGame (Ours)
Janus-Pro-7B
UniGame: Turning a Unified Multimodal Model Into Its Own Adversary
CVPR 2026 · Table 2
Cross-modal learning
789.5AIF
Qwen2.5-VL-7B
Aligning What Vision-Language Models See and Perceive with Adaptive Information Flow
CVPR 2026 · Table 4
Multimodal LLMs
889.31+ Entropy-gradient grounding (Ours)
LLaVA-1.6 Mistral 7B
Entropy-Gradient Grounding: Training-Free Evidence Retrieval in Vision-Language Models
ECCV 2026 · Table 1
Multimodal LLMs
989.3LLaVA-NeXT + VisRL Iter1
LLaVA-NeXT (Vicuna-7B-1.5, CLIP-ViT-L-14-336)
VisRL: Intention-Driven Visual Perception via Reinforced Reasoning
ICCV 2025 · Table 1
Multimodal LLMs
1089.1LIRA-2B
InternVL2-2B
LIRA: Inferring Segmentation in Large Multi-modal Models with Local Interleaved Region Assistance
ICCV 2025 · Table 1
Multimodal LLMs
87 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

POPE · score

61 papers, each with its best reported number for its own method; higher is better.

#score ↑MethodPaperTopic
193.2Finetuned (GHOST images)
Qwen2.5-VL
GHOST: Hallucination-Inducing Image Generation for Multimodal LLMs
ICLR 2026 · Mitigation table
Multimodal LLMs
289.98MoCHA
Vicuna-7B
MoCHA: Advanced Vision-Language Reasoning with MoE Connector and Hierarchical Group Attention
AAAI 2026 · Table 1
Multimodal LLMs
389.71SpatialReasoner-R1 fDPO 8B
InternVL2.5
Fine-Grained Preference Optimization Improves Spatial Reasoning in VLMs
NeurIPS 2025 · Table: General Vision-Language
Multimodal LLMs
489.1Setokim
LLaMA-2-7B
Towards Semantic Equivalence of Tokenization in Multimodal LLM
ICLR 2025 · Table: image understanding ben
Multimodal LLMs
588.8VisionLLM v2
Vicuna-7B
VisionLLM v2: An End-to-End Generalist Multimodal Large Language Model for Hundreds of Vision-Language Tasks
NeurIPS 2024 · Table 1
Multimodal LLMs
688.7HaploVL-7B-Pro
Qwen2.5-7B
HaploVL: A Single-Transformer Baseline for Multi-Modal Understanding
ICML 2025 · Main results table
Language, safety and interpretability
788.7Meteor-7B
InternLM2-7B
Meteor: Mamba-based Traversal of Rationale for Large Language and Vision Models
NeurIPS 2024 · Table 1
Multimodal LLMs
888.6VLsI-7B
Qwen2-VL-7B
VLsI: Verbalized Layers-to-Interactions from Large to Small Vision Language Models
CVPR 2025 · Table 1
Multimodal LLMs
988.6MoVA
Vicuna-7B
MoVA: Adapting Mixture of Vision Experts to Multimodal Context
NeurIPS 2024 · MLLM benchmarks table
Multimodal LLMs
1088.4Tar-1.5B
Qwen2.5-Instruct
Vision as a Dialect: Unifying Visual Understanding and Generation via Text-Aligned Representations
NeurIPS 2025 · Visual Understanding Benchmark
Human and embodied modeling
51 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

POPE · F1

18 papers, each with its best reported number for its own method; higher is better.

#F1 ↑MethodPaperTopic
192.4Gap-Init (rank 1)
BLIP-2 OPT-2.7B
When Is Rank-1 Enough? Geometry-Guided Initialization for Parameter-Efficient Fine-Tuning
ICML 2026 · Table 4
CLIP and zero-shot
287.6LLaVA-NeXT-Inst-IT
Qwen2-7B
INST-IT: Boosting Instance Understanding via Explicit Visual Prompt Instruction Tuning
NeurIPS 2025 · Main results on image benchmar
Multimodal LLMs
387.5LLaVA-FA-2B
Qwen-2.5-7B
LLaVA-FA: Learning Fourier Approximation for Compressing Large Multimodal Models
ICLR 2026 · Table 2
Quantization
487.4T-DRS
InterVL2-8B
Remember Me: Bridging the Long-Range Gap in LVLMs with Three-Step Inference-Only Decay Resilience Strategies
AAAI 2026 · Table 1
Multimodal LLMs
587.2LLaVA-MoD-2B
Qwen-2-1.5B
LLaVA-MoD: Making LLaVA Tiny via MoE-Knowledge Distillation
ICLR 2025 · Hallucination results table
Language, safety and interpretability
687SGRS + LocoRE
LLaVA-1.5-7B
Hallucination Begins Where Saliency Drops
ICLR 2026 · Table 1
Multimodal LLMs
786.5MoSVE
Vicuna-7B
Mixture of Specialized Vision Experts: Unlocking Complementary Visual Insights for Faithful MLLM Reasoning
ECCV 2026 · Table 1
Multimodal LLMs
886.4FaST
Vicuna-7B
Visual Agents as Fast and Slow Thinkers
ICLR 2025 · Main results table
Multimodal LLMs
986.4AcFormer
Vicuna-7B
Visual Anchors Are Strong Information Aggregators For Multimodal Large Language Model
NeurIPS 2024 · Main results table
Multimodal LLMs
1086.31Qwen-VL-Chat LoRA-SFT on Ours-only
Qwen-VL-Chat
CertainlyUncertain: A Benchmark and Metric for Multimodal Epistemic and Aleatoric Awareness
ICLR 2025 · Table: Results on other benchm
Visual question answering
8 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.