O'Reilly logo
  • Afroz Mohiuddin thinks this is interesting:

batch_size

From

Cover of Learning TensorFlow

Note

I find it very weird that the initial hidden state should depend in any way over the batch_size.