Audio-video understanding on AV-Counting
23.1Primary ScoreQwen2.5-Omni + AVATAR
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-Omni + AVATARReinforcement Learning Strategy=AVATAR2025.08 | 23.1 | |
| AV-ReasonerReinforcement Learning Strategy=N/A2025.08 | 23 | |
| Qwen2.5-Omni + GRPOReinforcement Learning Strategy=GRPO2025.08 | 22.8 | |
| EcholnkReinforcement Learning Strategy=N/A2025.08 | 22.7 | |
| Qwen2.5-OmniReinforcement Learning Strategy=Baseline2025.08 | 22.3 | |
| Omni-R1Reinforcement Learning Strategy=N/A2025.08 | 22 | |
| HumanOmniReinforcement Learning Strategy=N/A2025.08 | 19.6 | |
| Ola-7B + AVATARReinforcement Learning Strategy=AVATAR2025.08 | 19.5 | |
| Ola-7B + GRPOReinforcement Learning Strategy=GRPO2025.08 | 18.2 | |
| Ola-7BReinforcement Learning Strategy=Baseline2025.08 | 17.4 |