Table 4.
Layer | Activations | Learnable Variable | Descriptions |
---|---|---|---|
Sequence input | 2 | – | Sequence input with 2 dimensions |
LSTM | 400 | InputWeights 1600 × 2, RecurrentWeights 1600 × 400, Bias 1600 × 1 |
LSTM with 400 hidden units |
Fully Connected | 1 | Weights 1 × 400, Bias 1 × 1 | 1 fully connected layer |
Regression Output | 1 | – | Mean-squared-error with response |