Could you give the simple example in 1D case？

[damonwan1]

Cause I found in the experiments directory ：

I wonder how to use flexconv to solve the problem：

1. I have total 20000 nums 1D time series data with channel 21. For example, standard Weather dataset with 20 argument to predict OT.
2. I have different sequence length, such as [21,4] , [21,6] , [21, 11] . [21, 2]....
3. I WANT to use the flexconv to finish the work that transforming all the sequence to [21, 8].

Does this case is suitable for flexconv ?
How can I train the flexconv and finish the work above??

Hope for your response!!
Thank you very much for the good job!

[rjbruin]

@dwromero How do we deal with varying sequence lengths in 1D? Padding, I presume?

[dwromero]

Yes, although it is not necessary, it is practical to do it this way to vectorize operations across all input samples :)

(sorry for the long wait, I do not have access to d.w.romeroguzman@vu.nl anymore).