Open AccessJournal Article
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Colin Raffel,Noam Shazeer,Adam Roberts,Katherine Lee,Sharan Narang,Michael Matena,Yanqi Zhou,Wei Li,Peter J. Liu +8 more
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.
read more
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.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
•Posted Content
Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers
Yi Tay,Mostafa Dehghani,Jinfeng Rao,William Fedus,Samira Abnar,Hyung Won Chung,Sharan Narang,Dani Yogatama,Ashish Vaswani,Donald Metzler +9 more
TL;DR: In this paper, the authors present scaling insights from pretraining and finetuning Transformers, showing that aside from only the model size, model shape matters for downstream fine-tuning, scaling protocols operate differently at different compute regions, and widely adopted T5-base and T5large sizes are Pareto-inefficient.
43
Length-Adaptive Transformer: Train Once with Length Drop, Use Anytime with Search
Gyuwan Kim,Kyunghyun Cho +1 more
- 04 May 2021
TL;DR: In this paper, the authors propose a Length-Adaptive Transformer (LAT) for NLP, which is a structural variant of dropout, which stochastically determines the length of a sequence at each layer and then uses a multi-objective evolutionary search to find a length configuration that maximizes the accuracy and minimizes the computational complexity under any given computational budget.
•Posted Content
Improving and Simplifying Pattern Exploiting Training
TL;DR: ADAPET as discussed by the authors modifies PET's objective to provide denser supervision during fine-tuning, which outperforms PET on SuperGLUE without any task-specific unlabeled data.
42
Damage detection using in-domain and cross-domain transfer learning
TL;DR: A combination of in-domain and cross-domain transfer learning strategies for damage detection in bridges and visual explanations of predictive models to enable algorithmic transparency and provide insights to experts about the intrinsic decision-logic of typically black-box deep models are provided.
Empirical Evaluation of Pre-trained Transformers for Human-Level NLP: The Role of Sample Size and Dimensionality.
Adithya V Ganesan,Matthew Matero,Aravind Reddy Ravula,Huy Vu,H. Andrew Schwartz +4 more
- 01 Jun 2021
TL;DR: A systematic study on the role of dimension reduction methods as well as the dimensionality of embedding vectors and sample sizes as a function of predictive performance finds that fine-tuning large models with a limited amount of data pose a significant difficulty which can be overcome with a pre-trained dimension reduction regime.