Visual Abductive Reasoning on VAR (test)
11.35BLEU@4Human
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Human2026.01 | 11.35 | 19.36 | 36.92 | 147.79 | 40.59 | |
| AbductiveMLLM2026.01 | 6.54 | 13.41 | 27.95 | 57.04 | 36.8 | |
| Qwen2VL-7BFine-tuned=true2026.01 | 5.67 | 12.77 | 27.11 | 50.82 | 36.03 | |
| UPD-Trans2026.01 | 5.4 | 11.16 | 25.62 | 41.66 | 30.8 | |
| KN-VLM2026.01 | 4.72 | 10.74 | 24.4 | 37.2 | 33.17 | |
| REASONER2026.01 | 3.44 | 9.05 | 22.89 | 30.75 | 30.64 | |
| PDVC2026.01 | 3 | 8.54 | 20.71 | 25.14 | 27.8 | |
| Trans-XL2026.01 | 2.96 | 7.51 | 20.94 | 24.54 | 27.23 | |
| MART2026.01 | 2.86 | 7.47 | 20.87 | 24.05 | 27.77 | |
| Qwen2VL-7BFine-tuned=false2026.01 | 2.41 | 11.29 | 21.61 | 29.25 | 30.01 | |
| MFT2026.01 | 1.81 | 7.16 | 19.16 | 17.67 | 25.9 | |
| VideoChat2-7BFine-tuned=false2026.01 | 1.24 | 7.55 | 17.06 | 19.51 | 26.4 | |
| VTrans2026.01 | 0.71 | 6.92 | 19.12 | 7.11 | 22.13 | |
| GPT-4o-miniFine-tuned=false2026.01 | 0.63 | 7.38 | 13.64 | 7.3 | 12.27 |