IBM Support

Deep Learning training with Pytorch can fail with RuntimeError: Pin memory thread exited unexpectedly

Troubleshooting


Problem

If you are training a Deep Learning experiment with PyTorch, you might get a runtime failure of "Pin memory thread exited unexpectedly."

Symptom

If you are training a Deep Learning experiment with PyTorch, you might get a runtime failure of "Pin memory thread exited unexpectedly."

Cause

This is due to an issue with PyTorch, which shows up in certain versions.

Resolving The Problem

To resolve, try changing the PyTorch version to another supported version.
You can also try setting these flags:

persistent_workers=False
pin_memory=False 

Document Location

Worldwide

[{"Type":"MASTER","Line of Business":{"code":"LOB10","label":"Data and AI"},"Business Unit":{"code":"BU059","label":"IBM Software w\/o TPS"},"Product":{"code":"SSRTK2","label":"IBM Watson Machine Learning Accelerator for Cloud Pak for Data"},"ARM Category":[{"code":"a8m50000000L3PBAA0","label":"DLI-\u003EModel"}],"ARM Case Number":"","Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"2.1.0;and future releases"}]

Document Information

Modified date:
18 October 2022

UID

ibm16829523