A Bag of Tricks for Scaling CPU-based Deep FFMs to more than 300m Predictions per Second
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914871186030592 |
|---|---|
| author | Škrlj, Blaž Ben-Shalom, Benjamin Gašperšič, Grega Schwartz, Adi Hoseisi, Ramzi Ziporin, Naama Kopič, Davorin Tori, Andraž |
| author_facet | Škrlj, Blaž Ben-Shalom, Benjamin Gašperšič, Grega Schwartz, Adi Hoseisi, Ramzi Ziporin, Naama Kopič, Davorin Tori, Andraž |
| contents | Field-aware Factorization Machines (FFMs) have emerged as a powerful model for click-through rate prediction, particularly excelling in capturing complex feature interactions. In this work, we present an in-depth analysis of our in-house, Rust-based Deep FFM implementation, and detail its deployment on a CPU-only, multi-data-center scale. We overview key optimizations devised for both training and inference, demonstrated by previously unpublished benchmark results in efficient model search and online training. Further, we detail an in-house weight quantization that resulted in more than an order of magnitude reduction in bandwidth footprint related to weight transfers across data-centres. We disclose the engine and associated techniques under an open-source license to contribute to the broader machine learning community. This paper showcases one of the first successful CPU-only deployments of Deep FFMs at such scale, marking a significant stride in practical, low-footprint click-through rate prediction methodologies. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2407_10115 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A Bag of Tricks for Scaling CPU-based Deep FFMs to more than 300m Predictions per Second Škrlj, Blaž Ben-Shalom, Benjamin Gašperšič, Grega Schwartz, Adi Hoseisi, Ramzi Ziporin, Naama Kopič, Davorin Tori, Andraž Machine Learning Artificial Intelligence Information Retrieval Field-aware Factorization Machines (FFMs) have emerged as a powerful model for click-through rate prediction, particularly excelling in capturing complex feature interactions. In this work, we present an in-depth analysis of our in-house, Rust-based Deep FFM implementation, and detail its deployment on a CPU-only, multi-data-center scale. We overview key optimizations devised for both training and inference, demonstrated by previously unpublished benchmark results in efficient model search and online training. Further, we detail an in-house weight quantization that resulted in more than an order of magnitude reduction in bandwidth footprint related to weight transfers across data-centres. We disclose the engine and associated techniques under an open-source license to contribute to the broader machine learning community. This paper showcases one of the first successful CPU-only deployments of Deep FFMs at such scale, marking a significant stride in practical, low-footprint click-through rate prediction methodologies. |
| title | A Bag of Tricks for Scaling CPU-based Deep FFMs to more than 300m Predictions per Second |
| topic | Machine Learning Artificial Intelligence Information Retrieval |
| url | https://arxiv.org/abs/2407.10115 |