Hello, I have another question regarding MLForecast. After training a lightGBM model, I'm feeding it new inference data that includes 5 non-consecutive data points. Surprisingly, it defines lag 1 as the value of the last previous datapoint that exists in the dataset, even though it's not yesterday's data:

the data shouldn't have gaps, we have this issue: https://github.com/Nixtla/mlforecast/issues/458 to issue a warning

got it, thank you!