Journal Article•
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
TL;DR: This article introduced a unified framework that converts all text-based language problems into a text-to-text format and compared pre-training objectives, architectures, unlabeled data sets, transfer approaches, and other factors on dozens of language understanding tasks.
Abstract: Transfer learning, where a model is first pre-trained on a data-rich task before being fine-tuned on a downstream task, has emerged as a powerful technique in natural language processing (NLP). The effectiveness of transfer learning has given rise to a diversity of approaches, methodology, and practice. In this paper, we explore the landscape of transfer learning techniques for NLP by introducing a unified framework that converts all text-based language problems into a text-to-text format. Our systematic study compares pre-training objectives, architectures, unlabeled data sets, transfer approaches, and other factors on dozens of language understanding tasks. By combining the insights from our exploration with scale and our new ``Colossal Clean Crawled Corpus'', we achieve state-of-the-art results on many benchmarks covering summarization, question answering, text classification, and more. To facilitate future work on transfer learning for NLP, we release our data set, pre-trained models, and code.
Citations
More filters
26 Apr 2021
TL;DR: DistIR as mentioned in this paper is an IR for explicitly representing distributed DNN computation that can capture many popular distribution strategies, such as data, horizontal, and pipeline parallelism, and can be used to automatically search for an optimal distribution strategy.
Abstract: The rapidly growing size of deep neural network (DNN) models and datasets has given rise to a variety of distribution strategies such as data, horizontal, and pipeline parallelism. However, selecting the best set of strategies for a given model and hardware configuration is challenging because debugging and testing on clusters is expensive. In this work we propose DistIR, an IR for explicitly representing distributed DNN computation that can capture many popular distribution strategies. We build an analysis framework for DistIR programs, including a simulator and reference executor that can be used to automatically search for an optimal distribution strategy. Our unified global representation also eases development of new distribution strategies, as one can reuse the lowering to per-rank backend programs. Preliminary results using a grid search over a hybrid data/horizontal/pipeline-parallel space suggest DistIR and its simulator can aid automatic DNN distribution.
5 citations
01 Aug 2021
TL;DR: The PhotoChat dataset as mentioned in this paper contains 12k dialogues, each of which is paired with a user photo that is shared during the conversation, and the best image retrieval model achieves 10.4% recall@1 and best photo intent prediction model achieves 58.1% F1 score.
Abstract: We present a new human-human dialogue dataset - PhotoChat, the first dataset that casts light on the photo sharing behavior in online messaging. PhotoChat contains 12k dialogues, each of which is paired with a user photo that is shared during the conversation. Based on this dataset, we propose two tasks to facilitate research on image-text modeling: a photo-sharing intent prediction task that predicts whether one intends to share a photo in the next conversation turn, and a photo retrieval task that retrieves the most relevant photo according to the dialogue context. In addition, for both tasks, we provide baseline models using the state-of-the-art models and report their benchmark performances. The best image retrieval model achieves 10.4% recall@1 (out of 1000 candidates) and the best photo intent prediction model achieves 58.1% F1 score, indicating that the dataset presents interesting yet challenging real-world problems. We are releasing PhotoChat to facilitate future research work among the community.
5 citations
Posted Content•
TL;DR: This paper proposed a hybrid approach for leveraging the strengths of both extractive and generative readers, which achieved state-of-the-art results on NaturalQuestions and TriviaQA.
Abstract: To date, most of recent work under the retrieval-reader framework for open-domain QA focuses on either extractive or generative reader exclusively. In this paper, we study a hybrid approach for leveraging the strengths of both models. We apply novel techniques to enhance both extractive and generative readers built upon recent pretrained neural language models, and find that proper training methods can provide large improvement over previous state-of-the-art models. We demonstrate that a simple hybrid approach by combining answers from both readers can efficiently take advantages of extractive and generative answer inference strategies and outperforms single models as well as homogeneous ensembles. Our approach outperforms previous state-of-the-art models by 3.3 and 2.7 points in exact match on NaturalQuestions and TriviaQA respectively.
5 citations
Harbin Institute of Technology1, Rensselaer Polytechnic Institute2, Tencent3, University of Illinois at Urbana–Champaign4, Columbia University5, Allen Institute for Artificial Intelligence6, University of Pennsylvania7, Hong Kong University of Science and Technology8, Zhejiang University9, University of Colorado Boulder10
TL;DR: This article presented a new information extraction system that can automatically construct temporal event graphs from a collection of news documents from multiple sources, multiple languages (English and Spanish for their experiment), and multiple data modalities (speech, text, image and video).
Abstract: We present a new information extraction system that can automatically construct temporal event graphs from a collection of news documents from multiple sources, multiple languages (English and Spanish for our experiment), and multiple data modalities (speech, text, image and video). The system advances state-of-the-art from two aspects: (1) extending from sentence-level event extraction to cross-document cross-lingual cross-media event extraction, coreference resolution and temporal event tracking; (2) using human curated event schema library to match and enhance the extraction output. We have made the dockerlized system publicly available for research purpose at GitHub, with a demo video.
5 citations
Posted Content•
TL;DR: In this article, a pre-training methodology called Learning Inductive bias for Mathematical Reasoning (LIME) is proposed to learn inductive bias for mathematical reasoning tasks, which requires only a small fraction of the computation cost of the typical downstream task.
Abstract: While designing inductive bias in neural architectures has been widely studied, we hypothesize that transformer networks are flexible enough to learn inductive bias from suitable generic tasks. Here, we replace architecture engineering by encoding inductive bias in the form of datasets. Inspired by Peirce's view that deduction, induction, and abduction form an irreducible set of reasoning primitives, we design three synthetic tasks that are intended to require the model to have these three abilities. We specifically design these synthetic tasks in a way that they are devoid of mathematical knowledge to ensure that only the fundamental reasoning biases can be learned from these tasks. This defines a new pre-training methodology called "LIME" (Learning Inductive bias for Mathematical rEasoning). Models trained with LIME significantly outperform vanilla transformers on three very different large mathematical reasoning benchmarks. Unlike dominating the computation cost as traditional pre-training approaches, LIME requires only a small fraction of the computation cost of the typical downstream task.
5 citations