Video Question Answering on EgoThink
76Affordance ScoreGPT-5.2
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GPT-5.22026.05 | 76 | 69 | 85 | 73 | 92 | |
| GPT-4o2026.05 | 67 | 46 | 75 | 70 | 72 | |
| Ours FT (4B)Fine-tuning=Yes, Model Scale=4B2026.05 | 65 | 47 | 84 | 84 | 89 | |
| Baseline (Qwen3-VL-4B)Fine-tuning=No, Model Scale=4B2026.05 | 52 | 40 | 78 | 79 | 86 | |
| Qwen3-VL-235BModel Scale=235B2026.05 | 47 | 46 | 86 | 89 | 96 |