PoWER-BERT: Accelerating BERT Inference via Progressive Word-vector Elimination
Jul 12, 2020
Speakers
Saurabh Goyal
Speaker · 0 followers
Anamitra Roy Choudhury
Speaker · 0 followers
Souhabh M. Raje
Speaker · 0 followers
Venkatesan T. Chakaravarthy
Speaker · 0 followers
Yogish Sabharwal
Speaker · 0 followers
Ashish Verma
Speaker · 0 followers
About
We develop a novel method, called PoWER-BERT, for improving the inference time of the popular BERT model, while maintaining the accuracy. It works by: a) exploiting redundancy pertaining to word-vectors (intermediate encoder outputs) and eliminating the redundant vectors. b) determining which word-vectors to eliminate by developing a strategy for measuring their significance, based on the self-attention mechanism; c) learning how many word-vectors to eliminate by augmenting the BERT model and the loss function. Experiments on the standard GLUE benchmark shows that PoWER-BERT achieves up to 4.5x reduction in inference time over BERT with < 1 loss in accuracy. We show that PoWER-BERT offers significantly better trade-off between accuracy and inference time compared to prior methods. We demonstrate that our method attains up to 6.8x reduction in inference time with < 1 loss in accuracy when applied over ALBERT, a highly compressed version of BERT.We develop a novel method, called PoWER-BERT, for improving the inference time of the popular BERT model, while maintaining the accuracy. It works by: a) exploiting redundancy pertaining to word-vectors (intermediate encoder outputs) and eliminating the redundant vectors. b) determining which word-vectors to eliminate by developing a strategy for measuring their significance, based on the self-attention mechanism; c) learning how many word-vectors to eliminate by augmenting the BERT model and t…
Categories
AI & Data Science
Category · 10.8k presentations
About ICML 2020
The International Conference on Machine Learning (ICML) is the premier gathering of professionals dedicated to the advancement of the branch of artificial intelligence known as machine learning. ICML is globally renowned for presenting and publishing cutting-edge research on all aspects of machine learning used in closely related areas like artificial intelligence, statistics and data science, as well as important application areas such as machine vision, computational biology, speech recognition, and robotics. ICML is one of the fastest growing artificial intelligence conferences in the world. Participants at ICML span a wide range of backgrounds, from academic and industrial researchers, to entrepreneurs and engineers, to graduate students and postdocs.
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker
Universal Equivariant Multilayer Perceptrons
Watch later
On Convergence-Diagnostic based Step Sizes for Stochastic Gradient Descent
Watch later
Scott Pesme, …
How Good is the Bayes Posterior in Deep Neural Networks Really?
Watch later
Growing Adaptive Multi-hyperplane Machines
Watch later
Sparse Sinkhorn Attention
Watch later
Yi Tay, …