Keep CALM and Explore: Language Models for Action Generation in Text-based Games
This addresses the problem of action generation for autonomous agents in text-based games, offering a significant performance boost over existing methods.
The paper tackles the challenge of enormous action spaces in text-based games by proposing CALM, a language model trained on human gameplay to generate action candidates, combined with a reinforcement learning agent for re-ranking. The method achieves a 69% relative improvement in average game score over the previous state-of-the-art on unseen games in the Jericho benchmark.
Text-based games present a unique challenge for autonomous agents to operate in natural language and handle enormous action spaces. In this paper, we propose the Contextual Action Language Model (CALM) to generate a compact set of action candidates at each game state. Our key insight is to train language models on human gameplay, where people demonstrate linguistic priors and a general game sense for promising actions conditioned on game history. We combine CALM with a reinforcement learning agent which re-ranks the generated action candidates to maximize in-game rewards. We evaluate our approach using the Jericho benchmark, on games unseen by CALM during training. Our method obtains a 69% relative improvement in average game score over the previous state-of-the-art model. Surprisingly, on half of these games, CALM is competitive with or better than other models that have access to ground truth admissible actions. Code and data are available at https://github.com/princeton-nlp/calm-textgame.