Reinforcement learning (RL) algorithms typically start tabula rasa, without any prior knowledge of the environment, and without any prior skills. This however often leads to low sample efficiency, requiring a large amount of interaction with the environment. This is especially true in a lifelong learning setting, in which the agent needs to continually extend its capabilities. In this paper, we examine how a pre-trained task-independent language model can make a goal-conditional RL agent more sample efficient. We do this by facilitating transfer learning between different related tasks. We experimentally demonstrate our approach on a set of object navigation tasks.

Language modeling is the task of assigning a probability distribution over sequences of words that matches the distribution of a language. Although it sounds formidable, language modeling (i.e. ELMo, BERT, GPT) is essentially just predicting words in a blank. More formally, given a context, a language model predicts the probability of a word occurring in that context. Why is this method effective?