Reasoning on BBH S2T 1,000 items (test)
68.5Overall ScoreEA-CoT
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| EA-CoTModel=Qwen2.5-Omni, Tokens=1,024, Modality=S2T2026.06 | 68.5 | 62.4 | 80.4 | 61.6 | 69.6 | |
| EA-CoTModel=Phi-4-MM, Tokens=1,024, Modality=S2T2026.06 | 62.7 | 54.8 | 66.4 | 54.4 | 75.2 | |
| BaselineModel=Qwen2.5-Omni, Tokens=256, Modality=S2T2026.06 | 59.9 | 73.2 | 58 | 55.6 | 52.8 | |
| BaselineModel=Phi-4-MM, Tokens=256, Modality=S2T2026.06 | 53.6 | 56.4 | 59.2 | 48 | 50.8 |