MBIVOLT-24

Model summary

TransformerLSTM trained to predict $\dot{B}$ with past $\dot{B}$ as ground truth, after preprocessing subtracting $\dot{I}$ component.

We reach an RMSE of 0.0150 on validation.