Video Question Answering on Video-MME (no subtitles) 1.0 (test)
69.1AccuracyVideo-RLM
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Video-RLMbackbone=Gemini2026.03 | 69.1 | -7.1 | 390 | |
| Qwen3.5sampling=uniform2026.03 | 50.6 | -13.2 | 232 | |
| Video-RLMbackbone=Qwen2026.03 | 49.7 | -0.7 | 403 | |
| LLM over Captionscaptions=GPT-4o, reasoning_model=Qwen3.52026.03 | 36 | -28.2 | 235 |