SYSTEMS RESEARCH

TT-REC: Tensor Train Compression for Deep Learning Recommendation Model Embeddings

May 19, 2021

Abstract

The memory capacity of embedding tables in deep learning recommendation models (DLRMs) is increasing dramatically from tens of GBs to TBs across the industry. Given the fast growth in DLRMs, novel solutions are urgently needed in order to enable DLRM innovations. At the same time, this must be done in a fast and efficient way without having to exponentially increase infrastructure capacity demands.
In this paper, we demonstrate the promising potential of Tensor Train decomposition for DLRMs (TT-Rec), an important yet under-investigated context. We design and implement optimized kernels (TT-EmbeddingBag) to evaluate the proposed TT-Rec design. TT-EmbeddingBag is 3x faster than the SOTA TT implementation. The performance of TT-Rec is further optimized with the batched matrix multiplication and caching strategies for embedding vector lookup operations. In addition, we present mathematically and empirically the effect of weight initialization distribution on DLRM accuracy and propose to initialize the tensor cores of TT-Rec following the sampled Gaussian distribution. We evaluate TT-Rec across three important design space dimensions---memory capacity, accuracy, and timing performance---by training MLPerf-DLRM with Criteo's Kaggle and Terabyte data sets. TT-Rec compresses the model size by 4x to 221x for Kaggle, with 0.03% to 0.3% loss of accuracy correspondingly. For Terabyte, our approach achieves 112x model size reduction which comes with no accuracy loss nor training time overhead as compared to the uncompressed baseline.

Download the Paper

AUTHORS

Written by

Chunxing Yin

Bilge Acun

Xing Liu

Carole-Jean Wu

Publisher

MLSys 2021

Research Topics

Systems Research

Related Publications

December 07, 2018

SYSTEMS RESEARCH

Rethinking floating point for deep learning | Facebook AI Research

Reducing hardware overhead of neural networks for faster or lower power inference and training is an active area of research. Uniform quantization using integer multiply-add has been thoroughly investigated, which requires learning many…

Jeff Johnson

December 07, 2018

June 22, 2015

SYSTEMS RESEARCH

NLP

Fast Convolutional Nets With fbfft: A GPU Performance Evaluation | Facebook AI Research

We examine the performance profile of Convolutional Neural Network training on the current generation of NVIDIA Graphics Processing Units. We introduce two new Fast Fourier Transform convolution implementations: one based on NVIDIA’s cuFFT…

Nicolas Vasilache, Jeff Johnson, Michael Mathieu, Soumith Chintala, Serkan Piantino, Yann LeCun

June 22, 2015

March 02, 2020

SYSTEMS RESEARCH

Federated Optimization in Heterogenous Networks | Facebook AI Research

Federated Learning is a distributed learning paradigm with two key challenges that differentiate it from traditional distributed optimization: (1) significant variability in terms of the systems characteristics on each device in the network…

Tian Li, Anit Kumar Sahu, Manzil Zaheer, Maziar Sanjabi, Ameet Talwalkar, Virginia Smith

March 02, 2020

September 01, 2020

SYSTEMS RESEARCH

ResiliNet: Failure-Resilient Inference in Distributed Neural Networks

Techniques such as Federated Learning and Split Learning aim to train distributed deep learning models without sharing private data.…

Ashkan Yousefpour, Brian Q. Nguyen, Siddartha Devic, Guanhua Wang, Aboudy Kreidieh, Hans Lobel, Alexandre M. Bayen, Jason P. Jue

September 01, 2020

Help Us Pioneer The Future of AI

We share our open source frameworks, tools, libraries, and models for everything from research exploration to large-scale production deployment.