Hallucination Robustness on HallusionBench
58.4ScorePerceptionDLM-Base
Evaluation Results
| Method | Links | |
|---|---|---|
| PerceptionDLM-BaseModel Size=8B2026.06 | 58.4 | |
| PolarMemBackbone=QWEN2.5-VL-7B, Augmentation=PolarMem2026.01 | 57.8 | |
| PolarMemBackbone=InternVL-3.5-8B, Augmentation=PolarMem2026.01 | 56.3 | |
| ViLoMemBackbone=QWEN2.5-VL-32B, Augmentation=ViLoMem2026.01 | 55.1 | |
| ViLoMemBackbone=QWEN2.5-VL-7B, Augmentation=ViLoMem2026.01 | 54.9 | |
| Vanilla RAGBackbone=InternVL-3.5-8B, Augmentation=Vanilla RAG2026.01 | 54.9 | |
| InternVL-3.5-8BBackbone=InternVL-3.5-8B, Augmentation=None2026.01 | 54.7 | |
| ViLoMemBackbone=InternVL-3.5-8B, Augmentation=ViLoMem2026.01 | 53.8 | |
| QWEN2.5-VL-32BBackbone=QWEN2.5-VL-32B, Augmentation=None2026.01 | 53.5 | |
| PolarMemBackbone=QWEN2.5-VL-32B, Augmentation=PolarMem2026.01 | 52.9 | |
| QWEN2.5-VL-7BBackbone=QWEN2.5-VL-7B, Augmentation=None2026.01 | 52.7 | |
| Vanilla RAGBackbone=QWEN2.5-VL-32B, Augmentation=Vanilla RAG2026.01 | 52.7 | |
| PolarMemBackbone=LLaVA-Next-Mistral-7B, Augmentation=PolarMem2026.01 | 52 | |
| Qwen2.5-VLModel Size=7B2026.06 | 51.9 | |
| ViLoMemBackbone=LLaVA-Next-Mistral-7B, Augmentation=ViLoMem2026.01 | 51.5 | |
| LLaDA-VModel Size=8B2026.06 | 50.9 | |
| Vanilla RAGBackbone=QWEN2.5-VL-7B, Augmentation=Vanilla RAG2026.01 | 50.3 | |
| InternVL3Model Size=8B2026.06 | 49.9 | |
| ViLoMemBackbone=DeepSeek-VL2, Augmentation=ViLoMem2026.01 | 49.6 | |
| PolarMemBackbone=LLaVA-Next-Llama-3-8B, Augmentation=PolarMem2026.01 | 49 | |
| PolarMemBackbone=InternVL-3.5-4B, Augmentation=PolarMem2026.01 | 48 | |
| PolarMemBackbone=DeepSeek-VL2, Augmentation=PolarMem2026.01 | 48 | |
| ViLoMemBackbone=LLaVA-Next-Llama-3-8B, Augmentation=ViLoMem2026.01 | 46.9 | |
| DeepSeek-VL2Backbone=DeepSeek-VL2, Augmentation=None2026.01 | 46.2 | |
| LLaVA-Next-Mistral-7BBackbone=LLaVA-Next-Mistral-7B, Augmentation=None2026.01 | 45.4 | |
| ViLoMemBackbone=InternVL-3.5-4B, Augmentation=ViLoMem2026.01 | 45.2 | |
| ViLoMemBackbone=DeepSeek-VL2-Small, Augmentation=ViLoMem2026.01 | 45.1 | |
| InternVL-3.5-4BBackbone=InternVL-3.5-4B, Augmentation=None2026.01 | 44.8 | |
| Vanilla RAGBackbone=LLaVA-Next-Mistral-7B, Augmentation=Vanilla RAG2026.01 | 44.8 | |
| LLaVA-Next-Llama-3-8BBackbone=LLaVA-Next-Llama-3-8B, Augmentation=None2026.01 | 44.7 | |
| SDAR-VLModel Size=8B2026.06 | 44.4 | |
| Vanilla RAGBackbone=DeepSeek-VL2, Augmentation=Vanilla RAG2026.01 | 44 | |
| Vanilla RAGBackbone=InternVL-3.5-4B, Augmentation=Vanilla RAG2026.01 | 43.4 | |
| DeepSeek-VL2-SmallBackbone=DeepSeek-VL2-Small, Augmentation=None2026.01 | 43.4 | |
| PolarMemBackbone=DeepSeek-VL2-Small, Augmentation=PolarMem2026.01 | 42.8 | |
| Vanilla RAGBackbone=LLaVA-Next-Llama-3-8B, Augmentation=Vanilla RAG2026.01 | 42.8 | |
| Vanilla RAGBackbone=DeepSeek-VL2-Small, Augmentation=Vanilla RAG2026.01 | 40.5 |