Audio Hallucination Detection on CMM
95Audio-Language PAMiniCPM-o
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MiniCPM-oreported_in_cited_work=true2026.04 | 95 | 53 | 74 | 91 | 56.5 | 73.8 | 76 | |
| Qwen2.5-Omnireported_in_cited_work=true2026.04 | 92 | 78 | 85 | 95 | 56.5 | 75.8 | 81 | |
| Gemini-Flash-1.5reported_in_cited_work=true2026.04 | 88.5 | 39.5 | 64 | 79 | 36.5 | 57.8 | 76.3 | |
| DPO2026.04 | 87 | 82.5 | 84.8 | 82.5 | 72 | 77.3 | 82.9 | |
| SFT2026.04 | 86 | 78 | — | — | — | — | 82 | |
| DPO2026.04 | 86 | 80.5 | — | — | — | — | 83.3 | |
| Base model2026.04 | 85.5 | 85.5 | 85.5 | 82 | 64.5 | 73.3 | 82.5 | |
| Base model2026.04 | 85.5 | 84.5 | — | — | — | — | 85 | |
| ACPO2026.04 | 85 | 82 | — | — | — | — | 83.5 | |
| OmniDPO2026.04 | 84.5 | 81.5 | 83 | 77 | 77.5 | 77.3 | 82.4 | |
| ACPO2026.04 | 84.5 | 86.5 | 85.5 | 80.5 | 82 | 81.3 | 83.4 | |
| SFT2026.04 | 84 | 86 | 85 | 80.5 | 71.5 | 76 | 82.5 | |
| OmniDPO2026.04 | 81 | 85.5 | — | — | — | — | 83.3 |