How to predict future data when use n_lags

[Aaron1993]

When I use NeuralProphet to predict data for the coming year, to improve the accuracy, I set n_lags = 30, n_forecastes = 365, but the results just forecasted one day of data. However, using a rolling forecast costs a lot. So, how can I make the right predict using n_lags parametric?