Hallucination Assessment on KPM-HA 1.0 (test)
2.71GPT-Hall ScoreLLaVA-OneVision
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LLaVA-OneVisionUtilized-LLM=Qwen2, Size=7B2026.02 | 2.71 | 57.9 | |
| MiniCPM-V2.6Utilized-LLM=Qwen2, Size=8B2026.02 | 2.55 | 54.1 | |
| InternVideo-2.5Utilized-LLM=InternLM2.5, Size=8B2026.02 | 2.48 | 57.3 | |
| VideoLLaMA3Utilized-LLM=Qwen2.5, Size=7B2026.02 | 2.45 | 56.2 | |
| Qwen2-VLUtilized-LLM=Qwen2, Size=7B2026.02 | 2.44 | 56.3 | |
| Qwen2.5-VLUtilized-LLM=Qwen2.5, Size=7B2026.02 | 2.41 | 58.5 | |
| Gemini-2.5ProUtilized-LLM=Closed-source, Size=N/A2026.02 | 2.37 | 59.5 | |
| GPT-4.1Utilized-LLM=Closed-source, Size=N/A2026.02 | 2.32 | 60.3 | |
| KPM (w/o MoPE)Utilized-LLM=Qwen2.5, Size=7B2026.02 | 2.03 | 60.7 | |
| KPMUtilized-LLM=Qwen2.5, Size=7B2026.02 | 1.92 | 61.9 | |
| Tarsier2-RecapUtilized-LLM=Qwen2, Size=7B2026.02 | 1.87 | 62.1 |