Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2501.10546 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910789178228736 |
|---|---|
| author | Kurian, George Sardashti, Somayeh Sims, Ryan Berger, Felix Holt, Gary Li, Yang Willcock, Jeremiah Wang, Kaiyuan Quiroz, Herve Salem, Abdulrahman Grady, Julian |
| author_facet | Kurian, George Sardashti, Somayeh Sims, Ryan Berger, Felix Holt, Gary Li, Yang Willcock, Jeremiah Wang, Kaiyuan Quiroz, Herve Salem, Abdulrahman Grady, Julian |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_10546 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Scalable Machine Learning Training Infrastructure for Online Ads Recommendation and Auction Scoring Modeling at Google Kurian, George Sardashti, Somayeh Sims, Ryan Berger, Felix Holt, Gary Li, Yang Willcock, Jeremiah Wang, Kaiyuan Quiroz, Herve Salem, Abdulrahman Grady, Julian Distributed, Parallel, and Cluster Computing Artificial Intelligence Machine Learning C.0; C.4; I.2.6 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. |
| title | Scalable Machine Learning Training Infrastructure for Online Ads Recommendation and Auction Scoring Modeling at Google |
| topic | Distributed, Parallel, and Cluster Computing Artificial Intelligence Machine Learning C.0; C.4; I.2.6 |
| url | https://arxiv.org/abs/2501.10546 |