Im trying to understand the role of the LSTM layer.
Henrik Larsson

To add to my confusion you are also able to do training on partially full buffers, my intuition says that you are using 1 as number of time steps and not 30 as i thought first. That would clear up my confusion. When i check the size of rnn_input it is [1,?,256] so i guess ? is 1 always?

Like what you read? Give Henrik Larsson a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.