Multi-Tag Selection on CC3M
0.883PrecisionTTD
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| TTDscoring=Pixel-level similarity + Text-to-Token Diffusion2024.03 | 0.883 | 0.78 | 0.828 | 0.91 | |
| Spixel (Eq. (2))scoring=Pixel-level similarity2024.03 | 0.829 | 0.745 | 0.785 | 0.886 | |
| Qwenparameters=72B, type=Large Language Model2024.03 | 0.693 | 0.562 | 0.621 | 0.809 | |
| NLTKtype=NLP parser2024.03 | 0.598 | 0.837 | 0.698 | 0.796 | |
| Vicunaparameters=33B, type=Large Language Model2024.03 | 0.527 | 0.707 | 0.604 | 0.759 | |
| Vicunaparameters=7B, type=Large Language Model2024.03 | 0.441 | 0.71 | 0.544 | 0.709 |