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.
https://gist.github.com/shaoyuyoung/e636f2e7a306105b7e96809e2b85c28a
https://gist.github.com/shaoyuyoung/4bcefba4004f8271e64b5185c95a248a
https://github.com/pytorch/pytorch/pull/143460
https://github.com/pytorch/pytorch/issues/142853
https://github.com/pytorch/pytorch/compare/v2.6.0...v2.7.0
Source: Mitre, NVD
Published: 2025-09-25
Updated: 2025-10-03
Base Score: 5
Vector: CVSS2#AV:N/AC:L/Au:N/C:P/I:N/A:N
Severity: Medium
Base Score: 5.3
Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:N/A:N
EPSS: 0.00031