Sep 23, 2019
Would it be possible to keep the neural network from giving the same answer it gave to your previous action? Whenever I play, I keep getting stuck in a loop of the game repeating "TAKEN" to whatever I say
The newest version (https://colab.research.google.com/drive/1khUaPex-gyk1wXXLuqcopiWmHmcKl4UP) actually has a tweak to fix that. The reason this happens is that in the training data people usually take several items at a time, so the network learned that TAKEN is usually followed by...TAKEN. The new version just a manual patch to keep it from repeating itself.