Speech-Visual Question Answering on OK-VQA Speech-converted
29.04AccuracyUni-MoE
Evaluation Results
| Method | Links | |
|---|---|---|
| Uni-MoETraining Task=MoE-Task1-short-speech, Training Epochs=3 epoch2024.05 | 29.04 | |
| Single-Modality-Expert-Task3Evaluation Protocol=zero-shot2024.05 | 27.68 | |
| Single-Modality-Expert-Task1Evaluation Protocol=zero-shot2024.05 | 26.58 | |
| Uni-MoETraining Task=MoE-Task1-short-speech, Training Epochs=2 epoch2024.05 | 26.46 | |
| Uni-MoETraining Task=MoE-Task1, Training Epochs=2 epoch2024.05 | 19.98 | |
| Uni-MoETraining Task=MoE-Task1, Training Epochs=3 epoch2024.05 | 17.68 | |
| Single-Modality-Expert-Task6Evaluation Protocol=zero-shot2024.05 | 14.4 | |
| Single-Modality-Expert-Task4Evaluation Protocol=zero-shot2024.05 | 12.26 | |
| Uni-MoETraining Task=MoE-Task2, Training Epochs=2 epoch2024.05 | 8.08 | |
| Single-Modality-Expert-Task5Evaluation Protocol=zero-shot2024.05 | 3.42 | |
| Macaw-LLMEvaluation Protocol=zero-shot2024.05 | 1.06 | |
| X-InstructBLIPEvaluation Protocol=zero-shot2024.05 | 0 |