Visual Question Answering on FlowVQA (test)
71.3844PrecisionCROP - CLIP
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CROP - CLIPVision Encoder=CROP - CLIP2026.02 | 71.3844 | 54.3008 | 61.8101 | |
| PSEUDO - SaCLIPVision Encoder=PSEUDO - SaCLIP2026.02 | 66.0028 | 61.2537 | 64.7963 | |
| GRAN - SaCLIPVision Encoder=GRAN - SaCLIP2026.02 | 62.0144 | 60.1086 | 63.4233 | |
| PSEUDO - CLIPVision Encoder=PSEUDO - CLIP2026.02 | 55.8813 | 48.8994 | 53.6966 | |
| GRAN - CLIPVision Encoder=GRAN - CLIP2026.02 | 53.2666 | 46.7946 | 50.7943 | |
| w/o FTVision Encoder=w/o FT2026.02 | 48.3461 | 42.5924 | 49.4744 |