# Addressing the Memory Bottleneck in AI Model Training

arXiv: https://arxiv.org/abs/2003.08732 · HTML: https://mlsanity.com/paper/2003.08732/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 11 Mar 2020.

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

- [Few-Bit Backward: Quantized Gradients of Activation Functions for Memory Footprint Reduction](https://mlsanity.com/paper/2202.00441/) (2022)
- [IP-UNet: Intensity Projection UNet Architecture for 3D Medical Volume Segmentation](https://mlsanity.com/paper/2308.12761/) (2023)
- [Biases in Edge Language Models: Detection, Analysis, and Mitigation](https://mlsanity.com/paper/2502.11349/) (2025)
- [PRAISE: A Reproducible Workflow for Non-Programmer Scientists to Implement Published AI Models](https://mlsanity.com/paper/2610.16980/) (2026)

## Abstract (arXiv)

Using medical imaging as case-study, we demonstrate how Intel-optimized TensorFlow on an x86-based server equipped with 2nd Generation Intel Xeon Scalable Processors with large system memory allows for the training of memory-intensive AI/deep-learning models in a scale-up server configuration. We believe our work represents the first training of a deep neural network having large memory footprint (~ 1 TB) on a single-node server. We recommend this configuration to scientists and researchers who wish to develop large, state-of-the-art AI models but are currently limited by memory.