CVE-2025-46153

low-risk

Published 2025-09-25

PyTorch before 3.7.0 has a bernoulli_p decompose function in decompositions.py even though it lacks full consistency with the eager CPU implementation, negatively affecting nn.Dropout1d, nn.Dropout2d, and nn.Dropout3d for fallback_random=True.

Do I need to act?

0.07% chance of exploitation

EPSS score — low exploit probability

Not on CISA KEV list

No confirmed active exploitation reported to CISA

Patch status unknown

Check vendor advisories for fix availability and mitigation guidance

CVSS 5.3/10 Medium

NETWORK / LOW complexity

Affected Products (1)

Pytorch

Affected Vendors

Linuxfoundation

References (5)

Third Party Advisory https://gist.github.com/shaoyuyoung/4bcefba4004f8271e64b5185c95a248a

Third Party Advisory https://gist.github.com/shaoyuyoung/e636f2e7a306105b7e96809e2b85c28a

Product https://github.com/pytorch/pytorch/compare/v2.6.0...v2.7.0

Issue Tracking https://github.com/pytorch/pytorch/issues/142853

Issue Tracking https://github.com/pytorch/pytorch/pull/143460

/ 100

low-risk

Severity 21/34 · High

Exploitability 0/34 · Minimal

Exposure 5/34 · Minimal