Offline reinforcement learning (RL) is a challenging task, whose objective is to learn policies from static trajectory data without interacting with the environment. Recently, offline RL has been viewed as a sequence modeling problem, where an agent generates a sequence of subsequent actions based on a set of static transition experiences. However, existing approaches that use transformers to attend to all tokens naively can overlook the dependencies between different tokens and limit long-term dependency learning. In this paper, we propose the Graph Decision Transformer (GDT), a novel offline RL approach that models the input sequence into a causal graph to capture potential dependencies between fundamentally different concepts and facilitate temporal and causal relationship learning. GDT uses a graph transformer to process the graph inputs with relation-enhanced mechanisms, and an optional sequence transformer to handle fine-grained spatial information in visual tasks. Our experiments show that GDT matches or surpasses the performance of state-of-the-art offline RL methods on image-based Atari and OpenAI Gym.

Reinforcement Learning (RL) can be considered as a sequence modeling task, i.e., given a sequence of past state-action-reward experiences, a model autoregressively predicts a sequence of future actions. Recently, Transformers have been successfully adopted to model this problem. In this work, we propose State-Action-Reward Transformer (StARformer), which explicitly models local causal relations to help improve action prediction in long sequences. A sequence of such local representations combined with state representations, is then used to make action predictions over a long time span. Our experiments show that StARformer outperforms the state-of-the-art Transformer-based method on Atari (image) and Gym (state vector) benchmarks, in both offline-RL and imitation learning settings. StARformer is also more compliant with longer sequences of inputs compared to the baseline. Our code is available at https://github.com/ Reinforcement Learning (RL) naturally comes with sequential data: an agent observes a state from the environment, takes an action, observes the next state and receives a reward from the environment.

Drive-thru is a popular sales channel in the fast food industry where consumers can make food purchases without leaving their cars. Drive-thru recommendation systems allow restaurants to display food recommendations on the digital menu board as guests are making their orders. Popular recommendation models in eCommerce scenarios rely on user attributes (such as user profiles or purchase history) to generate recommendations, while such information is hard to obtain in the drive-thru use case. Thus, in this paper, we propose a new recommendation model Transformer Cross Transformer (TxT), which exploits the guest order behavior and contextual features (such as location, time, and weather) using Transformer encoders for drive-thru recommendations. Empirical results show that our TxT model achieves superior results in Burger King's drive-thru production environment compared with existing recommendation solutions. In addition, we implement a unified system to run end-to-end big data analytics and deep learning workloads on the same cluster. We find that in practice, maintaining a single big data cluster for the entire pipeline is more efficient and cost-saving. Our recommendation system is not only beneficial for drive-thru scenarios, and it can also be generalized to other customer interaction channels.

We find that, just as a large transformer model trained on language can generate coherent text, the same exact model trained on pixel sequences can generate coherent image completions and samples. By establishing a correlation between sample quality and image classification accuracy, we show that our best generative model also contains features competitive with top convolutional nets in the unsupervised setting. Unsupervised and self-supervised learning, or learning without human-labeled data, is a longstanding challenge of machine learning. Recently, it has seen incredible success in language, as transformer models like BERT, GPT-2, RoBERTa, T5, and other variants have achieved top performance on a wide array of language tasks. However, the same broad class of models has not been successful in producing strong features for image classification.