Multimodal Reasoning on PuzzleVQA
81.95AccuracyCORA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CORAModel=Qwen2-VL-7B2026.06 | 81.95 | 5.03 | |
| BaselineModel=Qwen2-VL-7B2026.06 | 77 | 30.21 | |
| CORAModel=Qwen2.5-VL-7B2026.06 | 76.6 | 3.37 | |
| BaselineModel=Qwen2.5-VL-7B2026.06 | 76.1 | 16.29 | |
| BaselineModel=Qwen2.5-VL-3B2026.06 | 65.45 | 32.71 | |
| CORAModel=Qwen2.5-VL-3B2026.06 | 64.35 | 24.06 | |
| CORAModel=Qwen2-VL-2B2026.06 | 63.35 | 99.9 | |
| BaselineModel=Qwen2-VL-2B2026.06 | 51.1 | 100 |