Gmail Smart Compose: Real-Time Assisted Writing
arXiv:1906.00080
Abstract
In this paper, we present Smart Compose, a novel system for generating interactive, real-time suggestions in Gmail that assists users in writing mails by reducing repetitive typing. In the design and deployment of such a large-scale and complicated system, we faced several challenges including model selection, performance evaluation, serving and other practical issues. At the core of Smart Compose is a large-scale neural language model. We leveraged state-of-the-art machine learning techniques for language model training which enabled high-quality suggestion prediction, and constructed novel serving infrastructure for high-throughput and real-time inference. Experimental results show the effectiveness of our proposed system design and deployment approach. This system is currently being served in Gmail.
References in corpus (12)
- Auto-Encoding Variational Bayes
- Sequence to Sequence Learning with Neural Networks
- Cross-lingual Language Model Pretraining
- Exploring the Limits of Language Modeling
- Revisiting Distributed Synchronous SGD
- Rethinking the Inception Architecture for Computer Vision
- Semi-supervised Sequence Learning
- Lingvo: a Modular and Scalable Framework for Sequence-to-Sequence Modeling
- On the State of the Art of Evaluation in Neural Language Models
- Efficient Transfer Learning Schemes for Personalized Language Modeling using Recurrent Neural Network
- Lightweight Adaptive Mixture of Neural and N-gram Language Models
- Personalized Language Model for Query Auto-Completion
Cited by in corpus (8)
- Extracting Training Data from Large Language Models
- On the Effectiveness of Transfer Learning for Code Search
- Understanding Unintended Memorization in Federated Learning
- On the Social and Technical Challenges of Web Search Autosuggestion Moderation
- Spelling Correction with Denoising Transformer
- IGA : An Intent-Guided Authoring Assistant
- Examination and Extension of Strategies for Improving Personalized Language Modeling via Interpolation
- NeuraCrypt is not private