ExT5: Towards Extreme Multi-Task Scaling for Transfer LearningDownload PDF

29 Sept 2021, 00:35 (modified: 11 Feb 2022, 02:03)ICLR 2022 PosterReaders: Everyone
Keywords: Natural Language Processing, Transfer Learning, Multi-task Learning
Abstract: Despite the recent success of multi-task learning and transfer learning for natural language processing (NLP), few works have systematically studied the effect of scaling up the number of tasks during pre-training. Towards this goal, this paper introduces ExMix (Extreme Mixture): a massive collection of 107 supervised NLP tasks across diverse domains and task-families. Using ExMix, we study the effect of multi-task pre-training at the largest scale to date, and analyze co-training transfer amongst common families of tasks. Through this analysis, we show that manually curating an ideal set of tasks for multi-task pre-training is not straightforward, and that multi-task scaling can vastly improve models on its own. Finally, we propose ExT5: a model pre-trained using a multi-task objective of self-supervised span denoising and supervised ExMix. Via extensive experiments, we show that ExT5 outperforms strong T5 baselines on SuperGLUE, GEM, Rainbow, Closed-Book QA tasks, and several tasks outside of ExMix. ExT5 also significantly improves sample efficiency while pre-training.
One-sentence Summary: Using a suite of 107 NLP tasks, we show that massively multi-task pre-training can improve downstream performance on NLP tasks, overcoming trends of negative transfer between tasks while fine-tuning.
12 Replies