Visual Question Answering on xGQA yes/no (test)
53.22Accuracy (en)TAG
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| TAGBackbone=LLaMA-Adapter-V2, Data Origin=TAG, Evaluation Protocol=Translate-Test2024.06 | 53.22 | 49.21 | 51.74 | 50.23 | 49.54 | 52.04 | 50.38 | 49.32 | 50.35 | |
| MERGEBackbone=LLaMA-Adapter-V2, Data Origin=MERGE, Evaluation Protocol=Translate-Test2024.06 | 53.15 | 49.52 | 51.79 | 50.25 | 49.73 | 51.64 | 50.57 | 49.42 | 50.42 | |
| HumanBackbone=LLaMA-Adapter-V2, Data Origin=Human, Evaluation Protocol=Translate-Test2024.06 | 53.03 | 47.72 | 50.4 | 48.1 | 46.6 | 50.08 | 48.01 | 46.9 | 48.28 | |
| MTBackbone=LLaMA-Adapter-V2, Data Origin=MT, Evaluation Protocol=Translate-Test2024.06 | 51.41 | 48.76 | 51.11 | 49.36 | 48.93 | 50.64 | 49.98 | 48.94 | 49.67 |