Hallucination Detection on HALP
0.9349AUROCGemma3-12B
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemma3-12BRepresentation=Query Tokens (QT)2026.03 | 0.9349 | |
| Molmo-V1-7BRepresentation=Query Tokens (QT)2026.03 | 0.9193 | |
| Qwen2.5-VL-7BRepresentation=Query Tokens (QT)2026.03 | 0.915 | |
| Phi4-VL-5.6BRepresentation=Query Tokens (QT)2026.03 | 0.9033 | |
| LLaVa-Next-8BRepresentation=Query Tokens (QT)2026.03 | 0.9026 | |
| SmolVLM2-2.2BRepresentation=Query Tokens (QT)2026.03 | 0.9014 | |
| Llama-3.2-11B-VisionRepresentation=Query Tokens (QT)2026.03 | 0.8959 | |
| Llama-3.2-11B-VisionRepresentation=Average2026.03 | 0.8013 | |
| Qwen2.5-VL-7BRepresentation=Average2026.03 | 0.7902 | |
| Qwen2.5-VL-7BRepresentation=Visual Features (VF)2026.03 | 0.7873 | |
| Phi4-VL-5.6BRepresentation=Vision Tokens (VT)2026.03 | 0.7738 | |
| SmolVLM2-2.2BRepresentation=Average2026.03 | 0.7715 | |
| Llama-3.2-11B-VisionRepresentation=Visual Features (VF)2026.03 | 0.7703 | |
| Phi4-VL-5.6BRepresentation=Average2026.03 | 0.7646 | |
| Molmo-V1-7BRepresentation=Average2026.03 | 0.763 | |
| Llama-3.2-11B-VisionRepresentation=Vision Tokens (VT)2026.03 | 0.7377 | |
| Gemma3-12BRepresentation=Average2026.03 | 0.7347 | |
| SmolVLM2-2.2BRepresentation=Visual Features (VF)2026.03 | 0.7238 | |
| LLaVa-Next-8BRepresentation=Average2026.03 | 0.7135 | |
| FastVLM-7BRepresentation=Vision Tokens (VT)2026.03 | 0.7028 | |
| SmolVLM2-2.2BRepresentation=Vision Tokens (VT)2026.03 | 0.6894 | |
| Molmo-V1-7BRepresentation=Vision Tokens (VT)2026.03 | 0.6867 | |
| FastVLM-7BRepresentation=Visual Features (VF)2026.03 | 0.683 | |
| Molmo-V1-7BRepresentation=Visual Features (VF)2026.03 | 0.683 | |
| Gemma3-12BRepresentation=Visual Features (VF)2026.03 | 0.6736 | |
| Qwen2.5-VL-7BRepresentation=Vision Tokens (VT)2026.03 | 0.6683 | |
| FastVLM-7BRepresentation=Average2026.03 | 0.6665 | |
| LLaVa-Next-8BRepresentation=Vision Tokens (VT)2026.03 | 0.627 | |
| Phi4-VL-5.6BRepresentation=Visual Features (VF)2026.03 | 0.6166 | |
| FastVLM-7BRepresentation=Query Tokens (QT)2026.03 | 0.6136 | |
| LLaVa-Next-8BRepresentation=Visual Features (VF)2026.03 | 0.6108 | |
| Gemma3-12BRepresentation=Vision Tokens (VT)2026.03 | 0.5956 |