# Mixed-Precision Embeddings for Large-Scale Recommendation Models

arXiv: https://arxiv.org/abs/2409.20305 · HTML: https://mlsanity.com/paper/2409.20305/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 30 Sep 2024.

## Built on by (most cited first, 4 of 4)

- [Beyond Zero Initialization: Investigating the Impact of Non-Zero Initialization on LoRA Fine-Tuning Dynamics](https://mlsanity.com/paper/2505.23194/) (2025)
- [Semantic Retrieval Augmented Contrastive Learning for Sequential Recommendation](https://mlsanity.com/paper/2503.04162/) (2025)
- [The Panaceas for Improving Low-Rank Decomposition in Communication-Efficient Federated Learning](https://mlsanity.com/paper/2505.23176/) (2025)
- [NestPipe: Large-Scale Recommendation Training on 1,500+ Accelerators via Nested Pipelining](https://mlsanity.com/paper/2604.06956/) (2026)

## Abstract (arXiv)

Embedding techniques have become essential components of large databases in the deep learning era. By encoding discrete entities, such as words, items, or graph nodes, into continuous vector spaces, embeddings facilitate more efficient storage, retrieval, and processing in large databases. Especially in the domain of recommender systems, millions of categorical features are encoded as unique embedding vectors, which facilitates the modeling of similarities and interactions among features. However, numerous embedding vectors can result in significant storage overhead. In this paper, we aim to compress the embedding table through quantization techniques. Given that features vary in importance levels, we seek to identify an appropriate precision for each feature to balance model accuracy and memory usage. To this end, we propose a novel embedding compression method, termed Mixed-Precision Embeddings (MPE). Specifically, to reduce the size of the search space, we first group features by frequency and then search precision for each feature group. MPE further learns the probability distribution over precision levels for each feature group, which can be used to identify the most suitable precision with a specially designed sampling strategy. Extensive experiments on three public datasets demonstrate that MPE significantly outperforms existing embedding compression methods. Remarkably, MPE achieves about 200x compression on the Criteo dataset without comprising the prediction accuracy.