# Randomness In Neural Network Training: Characterizing The Impact of Tooling

arXiv: https://arxiv.org/abs/2106.11872 · HTML: https://mlsanity.com/paper/2106.11872/

## Key facts

- Cited by 59 papers in the map (data through 14 Oct 2026).
- Builds on 48 of its 92 references that are in the map.
- First posted on arXiv 22 Jun 2021.

## Builds on (most cited first, 30 of 48)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [ImageNet Large Scale Visual Recognition Challenge](https://mlsanity.com/paper/1409.0575/) (2014)
- [Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift](https://mlsanity.com/paper/1502.03167/) (2015)
- [Densely Connected Convolutional Networks](https://mlsanity.com/paper/1608.06993/) (2016)
- [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](https://mlsanity.com/paper/1905.11946/) (2019)
- [Rethinking the Inception Architecture for Computer Vision](https://mlsanity.com/paper/1512.00567/) (2015)
- [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://mlsanity.com/paper/1704.04861/) (2017)
- [Deep Learning Face Attributes in the Wild](https://mlsanity.com/paper/1411.7766/) (2014)
- [Random Erasing Data Augmentation](https://mlsanity.com/paper/1708.04896/) (2017)
- [Deep Reinforcement Learning that Matters](https://mlsanity.com/paper/1709.06560/) (2017)
- [In-Datacenter Performance Analysis of a Tensor Processing Unit](https://mlsanity.com/paper/1704.04760/) (2017)
- [Underspecification Presents Challenges for Credibility in Modern Machine Learning](https://mlsanity.com/paper/2011.03395/) (2020)
- [What is being transferred in transfer learning?](https://mlsanity.com/paper/2008.11687/) (2020)
- [Don't Decay the Learning Rate, Increase the Batch Size](https://mlsanity.com/paper/1711.00489/) (2017)
- [Are GANs Created Equal? A Large-Scale Study](https://mlsanity.com/paper/1711.10337/) (2017)
- [cuDNN: Efficient Primitives for Deep Learning](https://mlsanity.com/paper/1410.0759/) (2014)
- [Hidden Stratification Causes Clinically Meaningful Failures in Machine Learning for Medical Imaging](https://mlsanity.com/paper/1909.12475/) (2019)
- [Cut, Paste and Learn: Surprisingly Easy Synthesis for Instance Detection](https://mlsanity.com/paper/1708.01642/) (2017)
- [Regularizing and Optimizing LSTM Language Models](https://mlsanity.com/paper/1708.02182/) (2017)
- [Accounting for Variance in Machine Learning Benchmarks](https://mlsanity.com/paper/2103.03098/) (2021)
- [Measuring the Effects of Data Parallelism on Neural Network Training](https://mlsanity.com/paper/1811.03600/) (2018)
- [What Do Compressed Deep Neural Networks Forget?](https://mlsanity.com/paper/1911.05248/) (2019)
- [Mesh-TensorFlow: Deep Learning for Supercomputers](https://mlsanity.com/paper/1811.02084/) (2018)
- [Characterising Bias in Compressed Models](https://mlsanity.com/paper/2010.03058/) (2020)
- [MLPerf Tiny Benchmark](https://mlsanity.com/paper/2106.07597/) (2021)
- [MLPerf Training Benchmark](https://mlsanity.com/paper/1910.01500/) (2019)
- [On the State of the Art of Evaluation in Neural Language Models](https://mlsanity.com/paper/1707.05589/) (2017)
- [Are All Layers Created Equal?](https://mlsanity.com/paper/1902.01996/) (2019)

## Built on by (most cited first, 30 of 59)

- [Deep Reinforcement Learning at the Edge of the Statistical Precipice](https://mlsanity.com/paper/2108.13264/) (2021)
- [JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models](https://mlsanity.com/paper/2404.01318/) (2024)
- [Data Banzhaf: A Robust Data Valuation Framework for Machine Learning](https://mlsanity.com/paper/2205.15466/) (2022)
- [Evaluating the Social Impact of Generative AI Systems in Systems and Society](https://mlsanity.com/paper/2306.05949/) (2023)
- [Elo Uncovered: Robustness and Best Practices in Language Model Evaluation](https://mlsanity.com/paper/2311.17295/) (2023)
- [Understanding and Mitigating Numerical Sources of Nondeterminism in LLM Inference](https://mlsanity.com/paper/2506.09501/) (2025)
- [How Does Quantization Affect Multilingual LLMs?](https://mlsanity.com/paper/2407.03211/) (2024)
- [Quantifying Social Biases Using Templates is Unreliable](https://mlsanity.com/paper/2210.04337/) (2022)
- [On the Factory Floor: ML Engineering for Industrial-Scale Ads Recommendation Models](https://mlsanity.com/paper/2209.05310/) (2022)
- [Sources of Irreproducibility in Machine Learning: A Review](https://mlsanity.com/paper/2204.07610/) (2022)
- [On the Variance of Neural Network Training with respect to Test Sets and Distributions](https://mlsanity.com/paper/2304.01910/) (2023)
- [MAGIC: Near-Optimal Data Attribution for Deep Learning](https://mlsanity.com/paper/2504.16430/) (2025)
- [Reproducibility in Optimization: Theoretical Framework and Limits](https://mlsanity.com/paper/2202.04598/) (2022)
- [The Co-12 Recipe for Evaluating Interpretable Part-Prototype Image Classifiers](https://mlsanity.com/paper/2307.14517/) (2023)
- [A Survey on Stability of Learning with Limited Labelled Data and its Sensitivity to the Effects of Randomness](https://mlsanity.com/paper/2312.01082/) (2023)
- [Proof-of-Learning is Currently More Broken Than You Think](https://mlsanity.com/paper/2208.03567/) (2022)
- [Neural Network Verification is a Programming Language Challenge](https://mlsanity.com/paper/2501.05867/) (2025)
- [FAIR-Ensemble: When Fairness Naturally Emerges From Deep Ensembling](https://mlsanity.com/paper/2303.00586/) (2023)
- [What Do Machine Learning Researchers Mean by "Reproducible"?](https://mlsanity.com/paper/2412.03854/) (2024)
- [A Siren Song of Open Source Reproducibility](https://mlsanity.com/paper/2204.04372/) (2022)
- [Reproducibility of Machine Learning: Terminology, Recommendations and Open Issues](https://mlsanity.com/paper/2302.12691/) (2023)
- [Optimistic Verifiable Training by Controlling Hardware Nondeterminism](https://mlsanity.com/paper/2403.09603/) (2024)
- [On the Prediction Instability of Graph Neural Networks](https://mlsanity.com/paper/2205.10070/) (2022)
- [Machine Learning needs Better Randomness Standards: Randomised Smoothing and PRNG-based attacks](https://mlsanity.com/paper/2306.14043/) (2023)
- [Power of Explanations: Towards automatic debiasing in hate speech detection](https://mlsanity.com/paper/2209.09975/) (2022)
- [Towards Inferential Reproducibility of Machine Learning Research](https://mlsanity.com/paper/2302.04054/) (2023)
- [Leveraging Per-Instance Privacy for Machine Unlearning](https://mlsanity.com/paper/2505.18786/) (2025)
- [Instability in clinical risk stratification models using deep learning](https://mlsanity.com/paper/2211.10828/) (2022)
- [Mixtera: A Data Plane for Foundation Model Training](https://mlsanity.com/paper/2502.19790/) (2025)
- [Model Parallelism With Subnetwork Data Parallelism](https://mlsanity.com/paper/2507.09029/) (2025)

## Abstract (arXiv)

The quest for determinism in machine learning has disproportionately focused on characterizing the impact of noise introduced by algorithmic design choices. In this work, we address a less well understood and studied question: how does our choice of tooling introduce randomness to deep neural network training. We conduct large scale experiments across different types of hardware, accelerators, state of art networks, and open-source datasets, to characterize how tooling choices contribute to the level of non-determinism in a system, the impact of said non-determinism, and the cost of eliminating different sources of noise. Our findings are surprising, and suggest that the impact of non-determinism in nuanced. While top-line metrics such as top-1 accuracy are not noticeably impacted, model performance on certain parts of the data distribution is far more sensitive to the introduction of randomness. Our results suggest that deterministic tooling is critical for AI safety. However, we also find that the cost of ensuring determinism varies dramatically between neural network architectures and hardware types, e.g., with overhead up to 746%, 241%, and 196% on a spectrum of widely used GPU accelerator architectures, relative to non-deterministic training. The source code used in this paper is available at https://github.com/usyd-fsalab/NeuralNetworkRandomness.