Gokce Keskin

Thanks, that works, my bad. I have one more question. Assume I have a 100 sequences. My batch size is 5, thus 20 batches in total. If the resetRNNState has...

Thanks, that makes sense. Though this bears another question now :) Assume the training data is a long sequence [1,2,3,4,5,....,100]. You first split this sequence into semi-redundant sequences of length...

I can modify the code - but I just want to make sure I get the "state preservation" concept right. Here is my understanding, please let me know if that...