Video Classification on MMEB Video Classification (Kinetics-700, SSv2, HMDB, UCF, Breakfast) v2 (test)
63.7Classification AccuracyTARA (Tarsier-7B)
Evaluation Results
| Method | Links | |
|---|---|---|
| TARA (Tarsier-7B)Backbone=Tarsier-7B, Zero-shot=true2025.12 | 63.7 | |
| FreeRet (Qwen2VL-7B)Backbone=Qwen2VL-7B, Zero-shot=true2025.12 | 63.2 | |
| Think-Then-Embed-7BNumber of Parameters=7B, Zero-shot=true2025.12 | 57.5 | |
| VLM2Vec-V2 (7B)Number of Parameters=7B, Zero-shot=true2025.12 | 45.9 | |
| VLM2Vec-Qwen2VL (7B)Number of Parameters=7B, Zero-shot=true2025.12 | 39.1 | |
| GME (7B)Number of Parameters=7B, Zero-shot=true2025.12 | 37.4 | |
| LamRA-Qwen2.5 (7B)Number of Parameters=7B, Zero-shot=true2025.12 | 32.9 | |
| ColPali v1.3 (3B)Number of Parameters=3B, Zero-shot=true2025.12 | 26.7 |