RepDL: Bit-level Reproducible Deep Learning Training and Inference

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xie, Peichen, Zhang, Xian, Chen, Shuo
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912640569180160
author Xie, Peichen
Zhang, Xian
Chen, Shuo
author_facet Xie, Peichen
Zhang, Xian
Chen, Shuo
contents Non-determinism and non-reproducibility present significant challenges in deep learning, leading to inconsistent results across runs and platforms. These issues stem from two origins: random number generation and floating-point computation. While randomness can be controlled through deterministic configurations, floating-point inconsistencies remain largely unresolved. To address this, we introduce RepDL, an open-source library that ensures deterministic and bitwise-reproducible deep learning training and inference across diverse computing environments. RepDL achieves this by enforcing correct rounding and order invariance in floating-point computation. The source code is available at https://github.com/microsoft/RepDL .
format Preprint
id arxiv_https___arxiv_org_abs_2510_09180
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RepDL: Bit-level Reproducible Deep Learning Training and Inference
Xie, Peichen
Zhang, Xian
Chen, Shuo
Machine Learning
Software Engineering
Non-determinism and non-reproducibility present significant challenges in deep learning, leading to inconsistent results across runs and platforms. These issues stem from two origins: random number generation and floating-point computation. While randomness can be controlled through deterministic configurations, floating-point inconsistencies remain largely unresolved. To address this, we introduce RepDL, an open-source library that ensures deterministic and bitwise-reproducible deep learning training and inference across diverse computing environments. RepDL achieves this by enforcing correct rounding and order invariance in floating-point computation. The source code is available at https://github.com/microsoft/RepDL .
title RepDL: Bit-level Reproducible Deep Learning Training and Inference
topic Machine Learning
Software Engineering
url https://arxiv.org/abs/2510.09180