Visual Question Answering on xGQA (test)
56.91Accuracy (en)Human
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| HumanEvaluation Protocol=Translate-Test, Training Data Origin=Human2024.06 | 56.91 | 51.01 | 54.1 | 52.17 | 50.58 | 54.02 | 50.84 | 49.99 | 51.82 | |
| TAGEvaluation Protocol=Translate-Test, Training Data Origin=TAG2024.06 | 56.67 | 52.65 | 55.44 | 53.56 | 53.11 | 54.83 | 53.89 | 52.65 | 53.73 | |
| MERGEEvaluation Protocol=Translate-Test, Training Data Origin=MERGE (Human + MT)2024.06 | 56.52 | 52.8 | 55.54 | 53.75 | 53.04 | 55.08 | 54.1 | 52.73 | 53.86 | |
| MTEvaluation Protocol=Translate-Test, Training Data Origin=MT2024.06 | 55.13 | 52.11 | 54.8 | 53.06 | 52.47 | 54.1 | 53.28 | 52.19 | 53.14 | |
| PaLIParameters=17B, Resolution=3782022.09 | 54.2 | 50 | 52.2 | 50.6 | 50.4 | 51.3 | 50.3 | 50.6 | — | |
| MPT2022.09 | 41.5 | 38.6 | 40.5 | 39.5 | 38.7 | 39.8 | 39.5 | 39.5 | — |