Open-ended Generative VQA on IRIS-120K Generative Overall
36.14Generation Average ScoreIRIS-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| IRIS-8BGroup=Ours2026.07 | 36.14 | |
| IRIS-4BGroup=Ours2026.07 | 35.91 | |
| IRIS-4B-SceneGroup=Ours2026.07 | 33.39 | |
| IRIS-4B-TFTGroup=Ours2026.07 | 32.91 | |
| IRIS-2BGroup=Ours2026.07 | 30.01 | |
| HuluMed-32BGroup=Med2026.07 | 25.51 | |
| HuluMed-14BGroup=Med2026.07 | 25.22 | |
| HuluMed-7BGroup=Med2026.07 | 25.02 | |
| Lingshu-32BGroup=Med2026.07 | 22.78 | |
| HGPT-V-34BGroup=Med2026.07 | 22.48 | |
| Lingshu-7BGroup=Med2026.07 | 19.09 | |
| HGPT-V-7BGroup=Med2026.07 | 18.12 | |
| MG-4BGroup=Med2026.07 | 17.55 | |
| Qwen3-vl-32BGroup=Gen2026.07 | 16.49 | |
| GPT-5-miniGroup=Gen2026.07 | 16.19 | |
| Internvl3.5-14BGroup=Gen2026.07 | 15.55 | |
| Internvl3.5-8BGroup=Gen2026.07 | 13.9 | |
| MG1.5-4BGroup=Med2026.07 | 12.51 | |
| Qwen3-vl-8BGroup=Gen2026.07 | 12.09 | |
| Qwen3-vl-4BGroup=Gen2026.07 | 10.69 | |
| Qwen3-vl-2BGroup=Gen2026.07 | 10.41 |