benglard/ConvNetPy

Problem using Trainer

Opened this issue · 0 comments

It seems like random_action_distribution isn't being used properly?

Traceback (most recent call last):
  File "main_learning2.py", line 373, in <module>
    main()
  File "main_learning2.py", line 353, in main
    action = brain.forward(simulator.map)
  File "/Users/eric/src/plithos/src/convnetpy/deepqlearn.py", line 232, in forward
    action = self.random_action()
  File "/Users/eric/src/plithos/src/convnetpy/deepqlearn.py", line 155, in random_action
    cumprob += self.random_action_distribution[k]
AttributeError: 'Brain' object has no attribute 'random_action_distribution'