Train an algorithm for labeling a time series with two variables?

Question

Machine learning

Hello. There is
a time series with binary markup
13.467.0
53.431.0
88.220.0
37.783.1
66.573.1
72.821.0
48.669.0
98.361.1
27.740.1
84.475.1
33.846.1
67.375.0
93.507.5 0 52,529,0 length of sequences marked with one is arbitrary. Reading, for example, about the implementation of LSTM on phyton, I understand that the data is fed to the algorithm in blocks of the same size. I could not figure out what to do with an infinite series and an arbitrary (at least with a limit) length marked "1" sequences.