| # Responsible Use |
|
|
| AV-QuantBench is intended for research on multimodal temporal reasoning, robustness, and evaluation. |
|
|
| Recommended uses: |
| - Benchmarking multimodal foundation models. |
| - Studying cross-modal conflict detection. |
| - Reproducing procedural audio-visual synthesis pipelines. |
| - Research on safer AI for high-stakes monitoring scenarios. |
|
|
| Not recommended uses: |
| - Deploying generated samples as real financial or medical advice. |
| - Presenting synthetic clips as authentic monitoring data. |
| - Using the dataset to deceive human operators or downstream systems. |
|
|
| The dataset contains synthetic or de-identified signals and excludes personally identifiable information. |
|
|
