The code for Stanford's CS230 ("Deep learning") final project. In this project I am building a reinforcement learning agent for playing text-based games in the TextWorld framework.

I explored various architectures for playing text-based games in the TextWorld framework. I have started with the architectures available in the literature (LSTM-DQN [1], DRRN [2] and SSAQN [3]) and proceeded to combine them into the hybrid model, that successfully solves many games from the First TextWorld Problems competition.

1. Language understanding for text-based games using deep reinforcement learning
2. Deep reinforcement learning with an unbounded action space
3. Using reinforcement learning to learn how to play text-based games