Semantic parsing using sequence-to-sequence models allows parsing of deeper representations compared to traditional word tagging based models. In spite of these advantages, widespread adoption of these models for real-time conversational use cases has been stymied by higher compute requirements and thus higher latency. In this work, we propose a non-autoregressive approach to predict semantic parse trees with an efficient seq2seq model architecture. By combining non-autoregressive prediction with convolutional neural networks, we achieve significant latency gains and parameter size reduction compared to traditional RNN models. Our novel architecture achieves up to an 81% reduction in latency on TOP dataset and retains competitive performance to non-pretrained models on three different semantic parsing datasets. Our code is available at https://github.com/facebookresearch/pytext
@article{arxiv.2104.04923,
title = {Non-Autoregressive Semantic Parsing for Compositional Task-Oriented Dialog},
author = {Arun Babu and Akshat Shrivastava and Armen Aghajanyan and Ahmed Aly and Angela Fan and Marjan Ghazvininejad},
journal= {arXiv preprint arXiv:2104.04923},
year = {2021}
}