Long-form Video-Language Understanding on LongVideo
62.1ScoreAdaSpark
Evaluation results
| Method | Links | |
|---|---|---|
| AdaSparkModel Size=Mid Size, Backbone=Qwen-2.5-VL-7B, SFT=true2026.04 | 62.1 | |
| FrameFusionModel Size=Mid Size, Backbone=Qwen-2.5-VL-7B, SFT=true2026.04 | 61.1 | |
| FastVModel Size=Mid Size, Backbone=Qwen-2.5-VL-7B, SFT=true2026.04 | 60.8 | |
| ToMeModel Size=Mid Size, Backbone=Qwen-2.5-VL-7B, SFT=true2026.04 | 60.8 | |
| Qwen-2.5-VL-7BModel Size=Mid Size, SFT=false2026.04 | 60.7 | |
| Qwen-2.5-VL-7B + SFTModel Size=Mid Size, SFT=true2026.04 | 59.8 | |
| VideoChat-Flash-2BModel Size=Small Size2026.04 | 58.3 | |
| MoBAModel Size=Mid Size, Backbone=Qwen-2.5-VL-7B, SFT=true2026.04 | 58.3 | |
| AdaSparkModel Size=Small Size, Backbone=Qwen-2.5-VL-3B, SFT=true2026.04 | 56.3 | |
| FrameFusionModel Size=Small Size, Backbone=Qwen-2.5-VL-3B, SFT=true2026.04 | 56 | |
| FastVModel Size=Small Size, Backbone=Qwen-2.5-VL-3B, SFT=true2026.04 | 55.9 | |
| ToMeModel Size=Small Size, Backbone=Qwen-2.5-VL-3B, SFT=true2026.04 | 55.8 | |
| Qwen-2.5-VL-3BModel Size=Small Size, SFT=false2026.04 | 54.2 | |
| Qwen-2.5-VL-3B + SFTModel Size=Small Size, SFT=true2026.04 | 53.6 | |
| InternVL2.5-2BModel Size=Small Size2026.04 | 52 | |
| MoBAModel Size=Small Size, Backbone=Qwen-2.5-VL-3B, SFT=true2026.04 | 51.2 | |
| Video-XL-7BModel Size=Mid Size2026.04 | 50.7 | |
| LongVA-7BModel Size=Mid Size2026.04 | 47.8 | |
| VideoChat2-7BModel Size=Mid Size2026.04 | 39.3 |