Fully Connected Network Layer does not support using direction

[DaQiZi]

I tried to use the objectives. direction () function on the full connection layer of Inception V1, but it would report an error.

InvalidArgumentErrorTraceback (most recent call last)
<ipython-input-17-d5f96d6202ac> in <module>()
      4 
      5 obj = objectives.direction("softmax2_pre_activation",np.random.randn(1008))
----> 6 img = render.render_vis(model, obj)

4 frames
/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.pyc in _do_call(self, fn, *args)
   1368           pass
   1369       message = error_interpolation.interpolate(message, self._graph)
-> 1370       raise type(e)(node_def, op, message)
   1371 
   1372   def _extend_graph(self):

InvalidArgumentError: 2 root error(s) found.
  (0) Invalid argument: Computed output size would be negative: -1 [input_size: 5, effective_filter_size: 7, stride: 1]
	 [[node import/avgpool0 (defined at /usr/local/lib/python2.7/dist-packages/lucid/modelzoo/vision_base.py:142) ]]
  (1) Invalid argument: Computed output size would be negative: -1 [input_size: 5, effective_filter_size: 7, stride: 1]
	 [[node import/avgpool0 (defined at /usr/local/lib/python2.7/dist-packages/lucid/modelzoo/vision_base.py:142) ]]
	 [[Mean/_35]]
0 successful operations.

Is this function not supported on the full connection layer?

The purpose of using direction is to invert the input image using the activation vector of the full connection layer. If the direction function can not achieve this goal, what should I do?