Real Time Recurrent Learning

Hello everyone,

I’m new to PyTorch and I have trouble in understanding how to implement a RTRL algorithm for training a RNN on a time serie data with a single scalar as input and a single scalar as output for every time step. Can someone give me some help or maybe the link to some tutorial/post/ repos?

I found these which is something similar but looks complicated to implement
1 Like

Thanks a lot. I already knew the first repo but I found it quite complicated as a starting point. The second one seems great!

Well if you implement it link it here then