PyTorch before 3.7.0 has a bernoullip 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 fallbackrandom=True.
{ "versions": [ { "introduced": "2.6.0" }, { "fixed": "2.7.0" } ] }
"https://storage.googleapis.com/cve-osv-conversion/osv-output/CVE-2025-46153.json"