Visual Abductive Reasoning on YouCookII (test)
6.16BLEU@4AbductiveMLLM
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| AbductiveMLLMCategory=MLLM, Mode=Fine-tuned2026.01 | 6.16 | 13.46 | 30.06 | 77.7 | 30.77 | |
| Qwen2VL-7B^FTCategory=MLLM, Mode=Fine-tuned2026.01 | 5.66 | 12.62 | 28.64 | 68.44 | 29.09 | |
| REASONERCategory=Traditional Model2026.01 | 3.54 | 9.47 | 24.62 | 32.99 | 23.19 | |
| Qwen2VL-7BCategory=MLLM, Mode=Zero-shot2026.01 | 2.46 | 8.41 | 22.1 | 35.83 | 21.83 | |
| VideoChat2-7BCategory=MLLM, Mode=Zero-shot2026.01 | 0.49 | 4.59 | 14.31 | 17.82 | 7.96 | |
| GPT-4o-miniCategory=MLLM, Mode=Zero-shot2026.01 | 0.45 | 4.22 | 12.78 | 14.15 | 9.75 |