Video-to-Text Retrieval on ReversedInTime
71.3Binary AccuracyTarsier + TARA
Evaluation Results
| Method | Links | |
|---|---|---|
| Tarsier + TARAEvaluation Protocol=Zero-shot2025.12 | 71.3 | |
| ArrowRL-Qwen2.5Evaluation Protocol=Fine-tuned on RTime2025.12 | 69.6 | |
| ArrowR-Qwen2Evaluation Protocol=Fine-tuned on RTime2025.12 | 68.8 | |
| Qwen2.5VLEvaluation Protocol=Zero-shot2025.12 | 66.6 | |
| TarsierEvaluation Protocol=Zero-shot2025.12 | 65 | |
| Qwen2VLEvaluation Protocol=Zero-shot2025.12 | 62.3 | |
| UMT-NegEvaluation Protocol=Fine-tuned on RTime2025.12 | 54.2 | |
| UMTEvaluation Protocol=Fine-tuned on RTime2025.12 | 51.3 | |
| Internvideo2-1BEvaluation Protocol=Zero-shot2025.12 | 51 | |
| SingularityEvaluation Protocol=Zero-shot2025.12 | 49.9 | |
| CLIP4ClipEvaluation Protocol=Fine-tuned on RTime2025.12 | 49.8 |