Meaning of sequence_length in learn to simulate

Thank you very much for providing excellent datasets in Learning to Simulate.
However, I have some questions regarding the variables in the metadata. Why is the value of `sequence_length` smaller by 1 compared to the actual dimension of the variables in the data? For example, in the waterramps dataset, the `sequence_length` is 600, while `position.shape[1]` is 601.