Video Comment Generation on TikTok English
83.46Info ScoreComHeat
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| ComHeatBackbone=baichuan2-13B2024.09 | 83.46 | 66.14 | 63.53 | 71.76 | 27.55 | 30.82 | -0.55 | 77.27 | 53.89 | 20.31 | |
| ViCo-fmode=fine-tuned2024.09 | 80.35 | 60.32 | 52.12 | 63.47 | 23.78 | 29.66 | -0.78 | 73.99 | 50.41 | 18.9 | |
| ViCo-umode=unpaired2024.09 | 72.03 | 52.77 | 34.34 | 47.53 | 14.36 | 20.33 | -0.83 | 73.25 | 44.37 | 10.25 | |
| ViCo-rmode=random2024.09 | 65.17 | 46.76 | 28.33 | 38.55 | 11.33 | 15.58 | -0.79 | 72.44 | 37.82 | 8.34 | |
| GIT-22024.09 | 63.59 | 41.33 | 24.69 | 32.11 | 11.67 | 15.32 | -0.98 | 68.13 | 35.92 | 8.57 | |
| CLIP4C2024.09 | 60.31 | 37.82 | 20.39 | 25.7 | 6.31 | 10.57 | -1.22 | 61.66 | 33.68 | 4.02 | |
| UT-CLIPBackbone=CLIP, Method Identity=Unified Transformer2024.09 | 55.31 | 31.58 | 17.57 | 20.33 | 6.25 | 7.23 | -1.84 | 46.98 | 30.55 | 10.5 | |
| UT-ResNetBackbone=ResNet-50, Method Identity=Unified Transformer2024.09 | 52.35 | 27.38 | 10.56 | 12.56 | 4.63 | 1.22 | — | 55.35 | 32.57 | 4.43 |