Log In Sign Up

Mixed-Precision Embedding Using a Cache

by   Jie Amy Yang, et al.

In recommendation systems, practitioners observed that increase in the number of embedding tables and their sizes often leads to significant improvement in model performances. Given this and the business importance of these models to major internet companies, embedding tables for personalization tasks have grown to terabyte scale and continue to grow at a significant rate. Meanwhile, these large-scale models are often trained with GPUs where high-performance memory is a scarce resource, thus motivating numerous work on embedding table compression during training. We propose a novel change to embedding tables using a cache memory architecture, where the majority of rows in an embedding is trained in low precision, and the most frequently or recently accessed rows cached and trained in full precision. The proposed architectural change works in conjunction with standard precision reduction and computer arithmetic techniques such as quantization and stochastic rounding. For an open source deep learning recommendation model (DLRM) running with Criteo-Kaggle dataset, we achieve 3x memory reduction with INT8 precision embedding tables and full-precision cache whose size are 5 maintaining accuracy. For an industrial scale model and dataset, we achieve even higher >7x memory reduction with INT4 precision and cache size 1 embedding tables, while maintaining accuracy, and 16 speedup by reducing GPU-to-host data transfers.


page 1

page 2

page 3

page 4


A Frequency-aware Software Cache for Large Recommendation System Embeddings

Deep learning recommendation models (DLRMs) have been widely applied in ...

iMARS: An In-Memory-Computing Architecture for Recommendation Systems

Recommendation systems (RecSys) suggest items to users by predicting the...

High-Performance Training by Exploiting Hot-Embeddings in Recommendation Systems

Recommendation models are commonly used learning models that suggest rel...

Clustering Embedding Tables, Without First Learning Them

To work with categorical features, machine learning systems employ embed...

AutoShard: Automated Embedding Table Sharding for Recommender Systems

Embedding learning is an important technique in deep recommendation mode...

Learning Compressed Embeddings for On-Device Inference

In deep learning, embeddings are widely used to represent categorical en...