Visual Question Answering on IRIS-120K (Judge)
97.25AccuracyIRIS-4B
Evaluation Results
| Method | Links | |
|---|---|---|
| IRIS-4BGroup=Ours2026.07 | 97.25 | |
| IRIS-8BGroup=Ours2026.07 | 97.04 | |
| IRIS-4B-TFTGroup=Ours2026.07 | 96.51 | |
| IRIS-2BGroup=Ours2026.07 | 96.19 | |
| IRIS-4B-SceneGroup=Ours2026.07 | 93.12 | |
| Lingshu-7BGroup=Med2026.07 | 84.97 | |
| Lingshu-32BGroup=Med2026.07 | 84.76 | |
| HuluMed-14BGroup=Med2026.07 | 83.92 | |
| HuluMed-32BGroup=Med2026.07 | 83.7 | |
| HuluMed-7BGroup=Med2026.07 | 83.07 | |
| Qwen3-vl-32BGroup=Gen2026.07 | 77.57 | |
| Internvl3.5-14BGroup=Gen2026.07 | 77.35 | |
| Internvl3.5-8BGroup=Gen2026.07 | 72.59 | |
| MG1.5-4BGroup=Med2026.07 | 71.01 | |
| GPT-5-miniGroup=Gen2026.07 | 69.52 | |
| Qwen3-vl-8BGroup=Gen2026.07 | 69.52 | |
| Qwen3-vl-2BGroup=Gen2026.07 | 69.31 | |
| MG-4BGroup=Med2026.07 | 66.88 | |
| HGPT-V-34BGroup=Med2026.07 | 66.56 | |
| Qwen3-vl-4BGroup=Gen2026.07 | 62.43 | |
| HGPT-V-7BGroup=Med2026.07 | 32.7 |