Video Question Answering on Video-MME no subtitles v2 (test)
14.91Non-Linear AccuracyQwen3-VL-8B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-VL-8BPipeline=Monolithic Baselines, K (Frame Pool Size)=144, Model Size=8B2026.05 | 14.91 | 30.94 | 2,441 | |
| Qwen3-VL-4BPipeline=Monolithic Baselines, K (Frame Pool Size)=144, Model Size=4B2026.05 | 14.11 | 30.06 | 1,415 | |
| GP-4BPipeline=Single-Model GridProbe, Selector Model=Qwen3-VL-4B, QA Model=Qwen3-VL-4B, K (Frame Pool Size)=1442026.05 | 12.86 | 28.28 | 440 | |
| GP-8BPipeline=Single-Model GridProbe, Selector Model=Qwen3-VL-8B, QA Model=Qwen3-VL-8B, K (Frame Pool Size)=1442026.05 | 12.6 | 28.06 | 842 | |
| GP-2B->8BPipeline=Cross-Model GridProbe, Selector Model=Qwen3-VL-2B, QA Model=Qwen3-VL-8B, K (Frame Pool Size)=1442026.05 | 11.7 | 26.72 | 677 | |
| Uniform-Meff -> 8BPipeline=Cross-Model GridProbe (Control), Selection=Uniform, QA Model=Qwen3-VL-8B, K (Frame Pool Size)=1442026.05 | 10.83 | 26.22 | 677 | |
| GP-2B->4BPipeline=Cross-Model GridProbe, Selector Model=Qwen3-VL-2B, QA Model=Qwen3-VL-4B, K (Frame Pool Size)=1442026.05 | 10.76 | 25.22 | 399 | |
| Qwen3-VL-2BPipeline=Monolithic Baselines, K (Frame Pool Size)=144, Model Size=2B2026.05 | 9.45 | 23.16 | 820 | |
| GP-2BPipeline=Single-Model GridProbe, Selector Model=Qwen3-VL-2B, QA Model=Qwen3-VL-2B, K (Frame Pool Size)=1442026.05 | 8.39 | 21.53 | 245 |