Autoregressive language models are the currently dominant paradigm for text generation, however they have some fundamental limitations that cannot be remedied by scale—for example inherently sequential and unidirectional generation. While alternate classes of models have been explored, we have limited mathematical understanding of their fundamental power and limitations. In this paper we focus on Generative Masked Language Models (GMLMs), a non-autoregressive paradigm in which we train a model to fit conditional probabilities of the data distribution via masking, which are subsequently used as inputs to a Markov Chain to draw samples from the model. These models empirically strike a promising speed-quality trade-off as each step can be typically parallelized by decoding the entire sequence in parallel. We develop a mathematical framework for analyzing and improving such models which sheds light on questions of sample complexity and inference speed and quality. Empirically, we adapt the T5 model for iteratively-refined parallel decoding, achieving 2-3x speedup in machine translation with minimal sacrifice in quality compared with autoregressive models. We run careful ablation experiments to give recommendations on key design choices, and make fine-grained observations on the common error modes in connection with our theory. Our mathematical analyses and empirical observations characterize both potentials and limitations of this approach, and can be applied to future works on improving understanding and performance of GMLMs. We released codes for our experiments.

Document and discourse segmentation are two fundamental NLP tasks pertaining to breaking up text into constituents, which are commonly used to help downstream tasks such as information retrieval or text summarization.In this work, we propose three transformer-based architectures and provide comprehensive comparisons with previously proposed approaches on three standard datasets.We establish a new state-of-the-art, reducing in particular the error rates by a large margin in all cases.We further analyze model sizes and find that we can build models with many fewer parameters while keeping good performance, thus facilitating real-world applications.Early life and marriage: Franklin Delano Roosevelt was born on January 30, 1882, in the Hudson Valley town of Hyde Park, New York, to businessman James Roosevelt I and his second wife, Sara Ann Delano.(...) Aides began to refer to her at the time as "the president's girlfriend", and gossip linking the two romantically appeared in the newspapers.(...) Legacy: Roosevelt is widely considered to be one of the most important figures in the history of the United States, as well as one of the most influential figures of the 20th century.(...) Roosevelt has also appeared on several U.S. Postage stamps.

Automatic text segmentation, which is the task of breaking a text into topically-consistent segments, is a fundamental problem in Natural Language Processing, Document Classification and Information Retrieval. Text segmentation can significantly improve the performance of various text mining algorithms, by splitting heterogeneous documents into homogeneous fragments and thus facilitating subsequent processing. Applications range from screening of radio communication transcripts to document summarization, from automatic document classification to information visualization, from automatic filtering to security policy enforcement - all rely on, or can largely benefit from, automatic document segmentation. In this article, a novel approach for automatic text and data stream segmentation is presented and studied. The proposed automatic segmentation algorithm takes advantage of feature extraction and unusual behaviour detection algorithms developed in [4, 5]. It is entirely unsupervised and flexible to allow segmentation at different scales, such as short paragraphs and large sections. We also briefly review the most popular and important algorithms for automatic text segmentation and present detailed comparisons of our approach with several of those state-of-the-art algorithms.