Video Captioning on CaReBench Action
40.6RecallCapRL++
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CapRL++Backbone=Qwen3-VL-4B2026.06 | 40.6 | 70.5 | 51.5 | |
| Qwen3-VL-235B-A22BVariant=Instruct2026.06 | 26.6 | 81.4 | 40.1 | |
| Qwen3-VL-8BVariant=Instruct2026.06 | 26.1 | 82.1 | 39.6 | |
| Qwen3-VL-4BType=baseline2026.06 | 23.9 | 81.5 | 36.9 |