Difficult to understand predictions in time series

[sealovecosmos]

hello，I completely agree with your idea about Neural CDE. Maybe I'm a bit clumsy, I've seen your discussions with others before. But I still don't quite understand, sorry. We all know that this model can predict on time series without dividing the training and testing sets? Just need to interpolate and predict on a time series. But I have about 100 samples in your example of predicting helical chirality. I don't quite understand this point. In addition, what is the interpolation and prediction process when the sample has multiple time series features? thank you