sup3r.utilities.loss_metrics.tf_derivative# tf_derivative(x, axis=1)[source]# Custom derivative function for compatibility with tensorflow. Note Matches np.gradient by using the central difference approximation. Parameters: x (tf.Tensor) – (n_observations, spatial_1, spatial_2, temporal) axis (int) – Axis to take derivative over