Text-to-Video Retrieval on ReversedInTime
71.6Binary AccuracyTarsier + TARA
Evaluation Results
| Method | Links | |
|---|---|---|
| Tarsier + TARAEvaluation Protocol=Zero-shot2025.12 | 71.6 | |
| TarsierEvaluation Protocol=Zero-shot2025.12 | 64.9 | |
| ArrowR-Qwen2Evaluation Protocol=Fine-tuned on RTime2025.12 | 57.1 | |
| Qwen2VLEvaluation Protocol=Zero-shot2025.12 | 56.3 | |
| ArrowRL-Qwen2.5Evaluation Protocol=Fine-tuned on RTime2025.12 | 55.6 | |
| UMT-NegEvaluation Protocol=Fine-tuned on RTime2025.12 | 54.5 | |
| Qwen2.5VLEvaluation Protocol=Zero-shot2025.12 | 53.4 | |
| UMTEvaluation Protocol=Fine-tuned on RTime2025.12 | 51.2 | |
| Internvideo2-1BEvaluation Protocol=Zero-shot2025.12 | 50 | |
| CLIP4ClipEvaluation Protocol=Fine-tuned on RTime2025.12 | 49.8 | |
| SingularityEvaluation Protocol=Zero-shot2025.12 | 48.7 |