Question about recurrent network input format.

[barnesew]

I'd like to request some clarification as to the input format of net.activate(some_array) from the following code snippet:

from tf_neat.recurrent_net import RecurrentNet

net = RecurrentNet.create(genome, config, bs)
outputs = net.activate(some_array)

My config file specifies 84 input, 0 hidden, and 7 output nodes. I've tried the following input formats with no success:

• Passing a list of numpy arrays with size (84,)
  • Yields: ValueError: Argument must be a dense tensor: [array([0., 0., ... 0., 0.])] - got shape [1, 84], but wanted [1].
• Passing a numpy array with size (84,)
  • Yields: ValueError: Shape must be rank 2 but is rank 1 for 'matmul' (op: 'MatMul') with input shapes: [7,84], [84].
• Passing a numpy array with size (1, 84)
  • Yields: AttributeError: 'Tensor' object has no attribute 'numpy'

All the example code has worked for me and I've set breakpoints in the example to compare my input to that of the examples. Any help is appreciated! I'm very interested in this project.