What ML technique should I use for multi-class classification of time-series sequential data?

Question

I have a dataset from which I have to learn sequences. Basically, there are 9 classes. The first data point of each class begins at t=0. So, I have to train the model in such a way that it classifies an input sequence. What ML technique should I use for this purpose? LSTM, Transformer, etc? By the way, each row represents a TCP packet. Also, the time difference between each timestamp is not fixed. It is in seconds. eg: t=0, then t=0.00006, then t=0.0028, etc.