Video Question Answering on VideoMME w/o sub. 1.0 (test)
65Overall AccMiMo-VL-7B + TCS
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MiMo-VL-7B + TCS#Frames=322026.01 | 65 | 76.3 | 65.9 | 52.7 | |
| MiMo-VL-7B#Frames=322026.01 | 62.4 | 75.2 | 60.7 | 51.3 | |
| Qwen2-VL-7B + AKS#Frames=322026.01 | 59.9 | — | — | — | |
| Qwen2-VL-7B + Q-Frame#Frames=4+8+322026.01 | 58.3 | 69.4 | 57.1 | 48.3 | |
| Qwen2-VL-7B + TCS#Frames=322026.01 | 58 | 69.2 | 57.4 | 47.9 | |
| Qwen2-VL-7B#Frames=322026.01 | 57.6 | — | — | — | |
| Video-XL-7B#Frames=1282026.01 | 55.5 | 64 | 53.2 | 49.2 | |
| LongVILA-8B#Frames=1282026.01 | 49.2 | 60.2 | 48.2 | 38.8 |