Scalable Machine Learning Training Infrastructure for Online Ads Recommendation and Auction Scoring Modeling at Google

TL;DR

This paper presents scalable infrastructure solutions for large-scale online ads machine learning at Google, addressing input processing, embedding optimization, and error handling, resulting in significant performance and cost improvements.

Contribution

It introduces novel techniques for input generation, embedding optimization, and resource management that enhance efficiency and scalability in production ML systems.

Findings

116% performance boost in production models

18% reduction in training costs

Effective handling of large-scale data and errors

Abstract

Large-scale Ads recommendation and auction scoring models at Google scale demand immense computational resources. While specialized hardware like TPUs have improved linear algebra computations, bottlenecks persist in large-scale systems. This paper proposes solutions for three critical challenges that must be addressed for efficient end-to-end execution in a widely used production infrastructure: (1) Input Generation and Ingestion Pipeline: Efficiently transforming raw features (e.g., "search query") into numerical inputs and streaming them to TPUs; (2) Large Embedding Tables: Optimizing conversion of sparse features into dense floating-point vectors for neural network consumption; (3) Interruptions and Error Handling: Minimizing resource wastage in large-scale shared datacenters. To tackle these challenges, we propose a shared input generation technique to reduce computational load of input generation by amortizing costs across many models. Furthermore, we propose partitioning, pipelining, and RPC (Remote Procedure Call) coalescing software techniques to optimize embedding operations. To maintain efficiency at scale, we describe novel preemption notice and training hold mechanisms that minimize resource wastage, and ensure prompt error resolution. These techniques have demonstrated significant improvement in Google production, achieving a 116% performance boost and an 18% reduction in training costs across representative models.