# The trade-offs of model size in large recommendation models : A 10000 × compressed criteo-tb DLRM model (100 GB parameters to mere 10MB)

arXiv: https://arxiv.org/abs/2207.10731 · HTML: https://mlsanity.com/paper/2207.10731/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 30 references that are in the map.
- First posted on arXiv 21 Jul 2022.

## Builds on (most cited first, 20 of 20)

- [DeepFM: A Factorization-Machine based Neural Network for CTR Prediction](https://mlsanity.com/paper/1703.04247/) (2017)
- [Deep & Cross Network for Ad Click Predictions](https://mlsanity.com/paper/1708.05123/) (2017)
- [Deep Learning Recommendation Model for Personalization and Recommendation Systems](https://mlsanity.com/paper/1906.00091/) (2019)
- [AutoInt: Automatic Feature Interaction Learning via Self-Attentive Neural Networks](https://mlsanity.com/paper/1810.11921/) (2018)
- [Sketching as a Tool for Numerical Linear Algebra](https://mlsanity.com/paper/1411.4357/) (2014)
- [xDeepFM: Combining Explicit and Implicit Feature Interactions for Recommender Systems](https://mlsanity.com/paper/1803.05170/) (2018)
- [Compressing Neural Networks with the Hashing Trick](https://mlsanity.com/paper/1504.04788/) (2015)
- [FiBiNET: Combining Feature Importance and Bilinear feature Interaction for Click-Through Rate Prediction](https://mlsanity.com/paper/1905.09433/) (2019)
- [The Architectural Implications of Facebook's DNN-based Personalized Recommendation](https://mlsanity.com/paper/1906.03109/) (2019)
- [Compositional Embeddings Using Complementary Partitions for Memory-Efficient Recommendation Systems](https://mlsanity.com/paper/1909.02107/) (2019)
- [TT-Rec: Tensor Train Compression for Deep Learning Recommendation Models](https://mlsanity.com/paper/2101.11714/) (2021)
- [AutoEmb: Automated Embedding Dimensionality Search in Streaming Recommendations](https://mlsanity.com/paper/2002.11252/) (2020)
- [Learnable Embedding Sizes for Recommender Systems](https://mlsanity.com/paper/2101.07577/) (2021)
- [Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systems](https://mlsanity.com/paper/1909.11810/) (2019)
- [Neural Input Search for Large Scale Recommendation Models](https://mlsanity.com/paper/1907.04471/) (2019)
- [Deep Learning Training in Facebook Data Centers: Design of Scale-up and Scale-out Systems](https://mlsanity.com/paper/2003.09518/) (2020)
- [Deep Learning Inference in Facebook Data Centers: Characterization, Performance Optimizations and Hardware Implications](https://mlsanity.com/paper/1811.09886/) (2018)
- [Optimizing Deep Learning Recommender Systems' Training On CPU Cluster Architectures](https://mlsanity.com/paper/2005.04680/) (2020)
- [Differentiable Neural Input Search for Recommender Systems](https://mlsanity.com/paper/2006.04466/) (2020)
- [Random Offset Block Embedding Array (ROBE) for CriteoTB Benchmark MLPerf DLRM Model : 1000× Compression and 3.1× Faster Inference](https://mlsanity.com/paper/2108.02191/) (2021)

## Built on by (most cited first, 2 of 2)

- [Review of compressed embedding layers and their applications for recommender systems](https://mlsanity.com/paper/2306.13724/) (2023)
- [Is Sliding Window All You Need? An Open Framework for Long-Sequence Recommendation](https://mlsanity.com/paper/2604.12372/) (2026)

## Abstract (arXiv)

Embedding tables dominate industrial-scale recommendation model sizes, using up to terabytes of memory. A popular and the largest publicly available machine learning MLPerf benchmark on recommendation data is a Deep Learning Recommendation Model (DLRM) trained on a terabyte of click-through data. It contains 100GB of embedding memory (25+Billion parameters). DLRMs, due to their sheer size and the associated volume of data, face difficulty in training, deploying for inference, and memory bottlenecks due to large embedding tables. This paper analyzes and extensively evaluates a generic parameter sharing setup (PSS) for compressing DLRM models. We show theoretical upper bounds on the learnable memory requirements for achieving (1 ± ε) approximations to the embedding table. Our bounds indicate exponentially fewer parameters suffice for good accuracy. To this end, we demonstrate a PSS DLRM reaching 10000× compression on criteo-tb without losing quality. Such a compression, however, comes with a caveat. It requires 4.5 × more iterations to reach the same saturation quality. The paper argues that this tradeoff needs more investigations as it might be significantly favorable. Leveraging the small size of the compressed model, we show a 4.3× improvement in training latency leading to similar overall training times. Thus, in the tradeoff between system advantage of a small DLRM model vs. slower convergence, we show that scales are tipped towards having a smaller DLRM model, leading to faster inference, easier deployment, and similar training times.