Training with Confidence: Catching Silent Errors in Deep Learning Training with Automated Proactive Checks

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jiang, Yuxuan, Zhou, Ziming, Xu, Boyu, Liu, Beijie, Xu, Runhui, Huang, Peng
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913899303927808
author Jiang, Yuxuan
Zhou, Ziming
Xu, Boyu
Liu, Beijie
Xu, Runhui
Huang, Peng
author_facet Jiang, Yuxuan
Zhou, Ziming
Xu, Boyu
Liu, Beijie
Xu, Runhui
Huang, Peng
contents Training deep learning (DL) models is a complex process, making it prone to silent errors that are challenging to detect and diagnose. This paper presents TRAINCHECK, a framework that takes a proactive checking approach to address silent training errors. TRAINCHECK automatically infers invariants tailored for DL training. It uses these invariants to proactively detect silent errors during the training process while providing debugging help. To evaluate TRAINCHECK, we reproduce 20 real-world silent training errors with diverse root causes. TRAINCHECK successfully detects 18 errors within a single training iteration. It also uncovers 6 unknown bugs in popular training libraries that lead to silent errors.
format Preprint
id arxiv_https___arxiv_org_abs_2506_14813
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Training with Confidence: Catching Silent Errors in Deep Learning Training with Automated Proactive Checks
Jiang, Yuxuan
Zhou, Ziming
Xu, Boyu
Liu, Beijie
Xu, Runhui
Huang, Peng
Machine Learning
Artificial Intelligence
Training deep learning (DL) models is a complex process, making it prone to silent errors that are challenging to detect and diagnose. This paper presents TRAINCHECK, a framework that takes a proactive checking approach to address silent training errors. TRAINCHECK automatically infers invariants tailored for DL training. It uses these invariants to proactively detect silent errors during the training process while providing debugging help. To evaluate TRAINCHECK, we reproduce 20 real-world silent training errors with diverse root causes. TRAINCHECK successfully detects 18 errors within a single training iteration. It also uncovers 6 unknown bugs in popular training libraries that lead to silent errors.
title Training with Confidence: Catching Silent Errors in Deep Learning Training with Automated Proactive Checks
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2506.14813