Vision Language Model Evaluation on Sycophancy Benchmark
88.8Mean ScoreLFM2-VL
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| LFM2-VLParams=450M2026.04 | 88.8 | 7.1 | 43 | 45.9 | 22.28 | 0.09 | |
| Gemma-3Params=4B2026.04 | 86 | 11.5 | 41.4 | 45.1 | 11.09 | 1.06 | |
| Qwen2-VLParams=7B2026.04 | 82.8 | 5.8 | 29.8 | 53.7 | 10.47 | 0.01 | |
| LLaVA-1.6Params=7B2026.04 | 73.7 | 19.1 | 21.2 | 54.9 | 5.98 | 6.41 | |
| Phi-3.5-VisionParams=4.2B2026.04 | 61.9 | 35.2 | 26.5 | 40.1 | 12.01 | 18.7 | |
| MiniCPM-V-4.5Params=8B2026.04 | 56 | 22.3 | 26.8 | 35.3 | 8.45 | 22.62 |