This repo contains an implementation of Deep Q-Network and Deep Recurrent Q-Network considering different models (RNN, LSTM and DNN) in a DTR scenario. Every approach can be enhanced with several exploration strategies, like deterministic epsilon-greedy, and softmax.
