There is a bug in rlcard.agents.DQNAgent python file

[blackkingback]

Good day to you all! Recently I tried to modify the DQNAgent from the source code and I noticed a bug(but I am not sure if this is a bug):

The whole DQNAgent is defined by using the functions from the torch package(as the users may notice the structure of the estimator network is a pure torch network). But near the end of the file, you can see the function:

def copy_model_parameters(sess, estimator1, estimator2)

This function however is defined by using the TensorFlow framework. If you create a DQNAgent and execute the standard rlcard example script, you will not find any bug reports(Even when you copy the parameters from the network to another network, you still will not find any bugs). But inside the source code, you can see inside a 'torch style' network there is a function that uses the TensorFlow framework and it doesn't import the TensorFlow framework in the beginning. So can you guys modify the source code file so that it will not cause any confusion?

[daochenzha]

@blackkingback Thanks for the feedback! This function is never called. We will simply remove it. This is because originally we implement everything in TensorFlow and later we adapt everything to torch.