Table 2 Hyperparameters used for training SCOPE
Name | Description | Values |
|---|---|---|
Hidden dimension (Dz) | Dimension of the hidden vectors in the transformer layers. | [16, 32] |
Dropout | Probability of dropout in the transformer layers. | [0.1, 0.2] |
Rollout-window (K) | Number of future hidden used as inputs to the event prediction modules. | [0, 1] |
Number of layers | Number of transformer layers in the architecture. | 2 |
Non-linearity | Whether to use a linear or non-linear prediction function for the event prediction. | [True, False] |
