Scale has opened new frontiers in natural language processing -- but at a high cost. In response, Mixture-of-Experts (MoE) and Switch Transformers have been proposed as an energy efficient path to even larger and more capable language models. But advancing the state-of-the-art across a broad set of natural language tasks has been hindered by training instabilities and uncertain quality during fine-tuning. Our work focuses on these issues and acts as a design guide. We conclude by scaling a sparse model to 269B parameters, with a computational cost comparable to a 32B dense encoder-decoder Transformer (Stable and Transferable Mixture-of-Experts or ST-MoE-32B). For the first time, a sparse model achieves state-of-the-art performance in transfer learning, across a diverse set of tasks including reasoning (SuperGLUE, ARC Easy, ARC Challenge), summarization (XSum, CNN-DM), closed book question answering (WebQA, Natural Questions), and adversarially constructed tasks (Winogrande, ANLI R3).
Building similarity graph...
Analyzing shared references across papers
Loading...
Barret Zoph
University of Southern California
Irwan Bello
Google (United States)
Sameer Kumar
Delhi Technological University
Building similarity graph...
Analyzing shared references across papers
Loading...
Zoph et al. (Thu,) studied this question.
synapsesocial.com/papers/6a08e3f627ceb0c2a2d611fc — DOI: https://doi.org/10.48550/arxiv.2202.08906
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: