Semantic Similarity Ranking on Crisscrossed Captions (test)
69.4SITSCE
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| CEArchitecture=Cross-Encoder, Fine-tuned on MSCOCO=true2024.07 | 69.4 | — | — | — | |
| ALIGNArchitecture=Dual-Encoder2024.07 | 67.6 | 72.9 | 77.2 | 72.6 | |
| CEArchitecture=Cross-Encoder2024.07 | 67.3 | — | — | — | |
| DE + CPRDArchitecture=Dual-Encoder, Fine-tuned on MSCOCO=true, Method Component=Contrastive Partial Ranking Distillation2024.07 | 66.9 | 74.4 | 76.1 | 72.5 | |
| DEArchitecture=Dual-Encoder, Fine-tuned on MSCOCO=true2024.07 | 65.1 | 73.9 | 75.7 | 71.5 | |
| DE + CPRDArchitecture=Dual-Encoder, Method Component=Contrastive Partial Ranking Distillation2024.07 | 64.3 | 74.9 | 75.6 | 71.6 | |
| DE_T2T+I2TArchitecture=Dual-Encoder2024.07 | 61.9 | 74.2 | 74.5 | 70.2 | |
| DEArchitecture=Dual-Encoder2024.07 | 61.8 | 74.1 | 75.1 | 70.3 | |
| DE_I2TArchitecture=Dual-Encoder2024.07 | 61.6 | 50.9 | 81.3 | 64.6 | |
| VSRNArchitecture=Dual-Encoder2024.07 | 60.4 | 73 | 70.1 | 67.8 | |
| VSE++Architecture=Dual-Encoder2024.07 | 55.2 | 74.4 | 73.3 | 67.6 |