Hi there,

My dataset is a list of variable length time sequences as follows

LIST = [

{(x_0_0, y_0_0), (x_0_1, y_0_1), … (x_0_5, y_0_5)}

{(x_1_0, y_1_0), (x_1_1, y_1_1), … (x_1_6, y_1_6)}

{(x_2_0, y_2_0), (x_2_1, y_2_1)}

…

]

where x is the input feature and y is the label corresponding to x

I want to train a LSTM with the above dataset.

I look through a few references in the domain of NLP, the practice is to add EOSs to pad the sequence so that they all have the same length. However, adding EOSs to my sequence does not make any sense in the domain of my application (robotics).

in this case, may I know how I can train the data in batches? (batch size > 1)

Thanks a lot in advance!