Spatio-temporal Question Answering on OC-NaVQA 1.0 (test)
73.1Descriptive Acc (Short)UQ-DAAAM
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| UQ-DAAAMReasoning model=GPT-5-mini, Refinement budget=B = 2, Uncertainty threshold=top 20-th percentile2026.06 | 73.1 | 77.8 | 78.6 | 76.1 | 15.46 | 25.35 | 58.94 | 37.84 | 1.948 | 0.91 | 1.275 | 1.589 | |
| DAAAM-QRReasoning model=GPT-5-mini, Refinement strategy=View-quality score qij, Refinement budget=B = 22026.06 | 71.5 | 76 | 74.8 | 73.8 | 17.83 | 29.47 | 57.26 | 40.14 | 2.284 | 0.905 | 1.408 | 1.734 | |
| DAAAM-ESReasoning model=GPT-5-mini, UQ Signal=evidence score, Refinement budget=B = 22026.06 | 71.4 | 76.9 | 74.8 | 74.3 | 19.28 | 34.61 | 55.87 | 39.69 | 2.703 | 0.903 | 1.341 | 1.683 | |
| DAAAM-EntReasoning model=GPT-5-mini, UQ Signal=entropy, Refinement budget=B = 22026.06 | 70.3 | 76.3 | 73.4 | 73.3 | 20.13 | 36.82 | 55.29 | 40.33 | 2.891 | 0.907 | 1.368 | 1.718 | |
| DAAAM-RRReasoning model=GPT-5-mini, Refinement strategy=Uniform random, Refinement budget=B = 22026.06 | 68.3 | 75.1 | 69.4 | 71.6 | 21.52 | 41.19 | 53.34 | 41.78 | 3.748 | 0.918 | 1.428 | 1.769 | |
| DAAAM-LNReasoning model=GPT-5-mini, UQ Signal=next-token log likelihood, Refinement budget=B = 22026.06 | 68.2 | 75.6 | 69.4 | 71.4 | 21.74 | 41.45 | 53.31 | 41.55 | 3.623 | 0.91 | 1.43 | 1.776 | |
| DAAAM-NCReasoning model=GPT-5-mini, UQ Signal=neighborhood consistency, Refinement budget=B = 22026.06 | 68.1 | 75.5 | 69.2 | 71.3 | 21.88 | 41.53 | 53.39 | 41.62 | 3.648 | 0.912 | 1.434 | 1.783 | |
| DAAAMReasoning model=GPT-5-mini, Refinement budget=N/A2026.06 | 67.9 | 75.3 | 69 | 71.1 | 22.07 | 41.69 | 53.47 | 41.75 | 3.669 | 0.914 | 1.438 | 1.792 |