Fast Inference from Transformers via Speculative Decoding

Yaniv Leviathan, Matan Kalman, Yossi Matias

Inference from large autoregressive models like Transformers is slow - decoding K tokens takes K serial runs of the model. In this work we introduce speculative decoding - an algorithm to sample from autoregressive models faster without any changes to the outputs, by computing several tokens in parallel. At the heart of our approach lie the observations that (1) hard language-modeling tasks often include easier subtasks that can be approximated well by more efficient models, and (2) using speculative execution and a novel sampling method, we can make exact decoding from the large models faster, by running them in parallel on the outputs of the approximation models, potentially generating several tokens concurrently, and without changing the distribution. Our method can accelerate existing off-the-shelf models without retraining or architecture changes. We demonstrate it on T5-XXL and show a 2X-3X acceleration compared to the standard T5X implementation, with identical outputs.

• Google Research
• Google Research
• Bar-Ilan University
• Carnegie Mellon University
• École Polytechnique Fédérale de Lausanne
• Harvard University
• Illinois Institute of Technology
• Massachusetts Institute of Technology
• Nanyang Technological University
• National Institute of Informatics, Tokyo
• Purdue University
• Rutgers University
• Texas A&M University
• The University of Tokyo
• University of California
• University of Maryland
• University of Michigan
• University of Southern California
• Google Research
• Google DeepMind
• Google Research
• Stanford University
• Google Research
• University of California
• University of Waterloo
• Vector Institute
• Google Research
• Princeton University