Fast Inference from Transformers via Speculative Decoding

Yaniv Leviathan, Matan Kalman, Yossi Matias
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:19274-19286, 2023.

Abstract

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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-leviathan23a, title = {Fast Inference from Transformers via Speculative Decoding}, author = {Leviathan, Yaniv and Kalman, Matan and Matias, Yossi}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {19274--19286}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/leviathan23a/leviathan23a.pdf}, url = {https://proceedings.mlr.press/v202/leviathan23a.html}, abstract = {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.} }
Endnote
%0 Conference Paper %T Fast Inference from Transformers via Speculative Decoding %A Yaniv Leviathan %A Matan Kalman %A Yossi Matias %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-leviathan23a %I PMLR %P 19274--19286 %U https://proceedings.mlr.press/v202/leviathan23a.html %V 202 %X 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.
APA
Leviathan, Y., Kalman, M. & Matias, Y.. (2023). Fast Inference from Transformers via Speculative Decoding. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:19274-19286 Available from https://proceedings.mlr.press/v202/leviathan23a.html.

Related Material