Image-Text Retrieval on MSCOCO 1K
83.1I2T RecallFully-FT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Fully-FTArchitecture=VSE∞2023.08 | 83.1 | 71.7 | 542.7 | |
| Fully-FTParams. (M)=201.2, Mem. (G)=176.82026.04 | 83.1 | 71.7 | 542.7 | |
| Fully-FTParams. (M)=201.2, Mem. (G)=176.82026.04 | 83.1 | 71.7 | 542.7 | |
| Ours‡Params. (M)=11.8, Mem. (G)=25.42026.04 | 82.8 | 71 | 542.6 | |
| Ours♠Params. (M)=11.4, Mem. (G)=25.2, QPS=14.832026.04 | 82.2 | 69.2 | 537.9 | |
| Ours†Params. (M)=12.9, Mem. (G)=25.52026.04 | 81.9 | 69.1 | 538.7 | |
| SHERLParams. (M)=11.3, Mem. (G)=24.4, QPS=10.242026.04 | 81.8 | 69.2 | 537.5 | |
| SHERLParams. (M)=11.3, Mem. (G)=24.42026.04 | 81.8 | 69.2 | 537.5 | |
| Ours♡Params. (M)=12.4, Mem. (G)=25.1, QPS=14.712026.04 | 81.3 | 68.8 | 536.4 | |
| UniPTArchitecture=VSE∞2023.08 | 80.6 | 67.5 | 532.9 | |
| UniPTParams. (M)=12.4, Mem. (G)=24.4, QPS=8.612026.04 | 80.6 | 67.5 | 532.9 | |
| UniPTParams. (M)=12.4, Mem. (G)=24.42026.04 | 80.6 | 67.5 | 532.9 | |
| LSTArchitecture=VSE∞2023.08 | 78.2 | 64.8 | 525.8 | |
| LSTParams. (M)=9.7, Mem. (G)=24.42026.04 | 78.2 | 64.8 | 525.8 | |
| LSTParams. (M)=9.7, Mem. (G)=24.42026.04 | 78.2 | 64.8 | 525.8 | |
| Partially⇓Architecture=VSE∞2023.08 | 74.6 | 59.5 | 510.5 |