Music Retrieval on Custom Music Video Clips Dataset (test)
2MRMuseChat
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MuseChatBackbone=Vicuna-7B, Audio Encoder=AST Encoder2023.10 | 2 | 32.79 | 63.92 | 76.53 | 40.49 | |
| Cross-Attn Fusion2023.10 | 3 | 25.71 | 56.97 | 71.07 | 31.48 | |
| Self-Attn Fusion2023.10 | 5 | 21.4 | 50.83 | 65.29 | 28.37 | |
| MVP2023.10 | 7 | 20.71 | 48.89 | 63.14 | 20.71 | |
| Sum-Fusion2023.10 | 17 | 10.58 | 28.47 | 40.19 | 21.7 | |
| Chance2023.10 | 250 | 0.2 | 1 | 2 | 0.4 |