Table A1.
Training option parameters of the SGDM solver.
| SGDM Options | Values |
|---|---|
| Momentum | 0.9 |
| Initial Learn Rate | 0.001 |
| Learn Rate Schedule | ‘none’ |
| Learn Rate Drop Factor | 0.1 |
| Learn Rate Drop Period | 10 |
| L2 Regularization | 0.0001 |
| Gradient Threshold Method | ‘l2norm’ |
| Gradient Threshold | Inf |
| Max Epochs | 5 |
| Mini-BatchSize | 32 |
| Verbose | 0 |
| Verbose Frequency | 50 |
| Validation Frequency | 30 |
| Validation Patience | Inf |
| Shuffle | ‘every-epoch’ |
| Execution Environment | ‘auto’ |
| Sequence Length | ‘longest’ |
| Sequence Padding Value | 0 |
| Sequence Padding Direction | ‘right’ |
| Dispatch In Background | 0 |
| Reset Input Normalization | 1 |
| Batch Normalization Statistics | ‘population’ |