Multimodal Hallucination and Real-world Evaluation on RealWorldQA
74.6AccuracyDenseMLLM-4B
Evaluation Results
| Method | Links | |
|---|---|---|
| DenseMLLM-4BModel Scale=4B2026.02 | 74.6 | |
| Qwen3-VL-4BModel Scale=4B2026.02 | 70.9 | |
| InternVL-3.5-4BModel Scale=4B2026.02 | 66.3 |
Loading the SOTA2 catalog…