Paragraph-to-video retrieval on Condensed Movie (test)
25.5Recall@1TESTA w/o agg.
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TESTA w/o agg.#PT Data=5M, #Frame=96, GFLOPS=2383, token_aggregation=disabled2023.10 | 25.5 | 46.8 | 56 | |
| TESTA#PT Data=5M, #Frame=96, GFLOPS=13812023.10 | 24.9 | 46.5 | 55.1 | |
| TESTA w/o agg.#PT Data=5M, #Frame=32, GFLOPS=786, token_aggregation=disabled2023.10 | 23.5 | 45.4 | 54.8 | |
| TESTA#PT Data=5M, #Frame=32, GFLOPS=4202023.10 | 21.5 | 42.4 | 50.7 | |
| VINDLU#PT Data=25M, #Frame=32, GFLOPS=745, implementation=re-implementation2023.10 | 18.4 | 36.4 | 44.3 | |
| LF-VILA#PT Data=8M, #Frame=32, GFLOPS=2982023.10 | 13.6 | 32.5 | 41.8 | |
| TeachText2023.10 | 12.1 | 27.4 | 37.5 | |
| MoEE2023.10 | 1.9 | 7.8 | 13.4 |