Advisoriesby IbisPyTorchPYSEC-2025-202
PyTorch before 3.7.0 has a bernoulli_p decompose function in decompositions.py...
Fix: upgrade to 2.7.0 or later
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.
Affected versions
| Package | Affected | Fixed in |
|---|---|---|
| torch PyPI | >= 2.6.0, < 2.7.0 | 2.7.0 |
Details and references
- CVSS 3.1
- CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:N/A:N
- Severity from
- the CVSS score
- Also known as
- BIT-pytorch-2025-46153, CVE-2025-46153
More PyTorch advisories
All PyTorch
PyTorch: denial of serviceCVE-2025-55560fixed in 2.7.1pytorch v2.8.0 was discovered to display unexpected behaviorCVE-2025-55552fixed in 2.9.0PyTorch: denial of serviceCVE-2025-55553fixed in 2.7.1pytorch v2.8.0 was discovered to contain an integer overflow in the component torch.nan_to_num-.long()CVE-2025-55554fixed in 2.9.0PyTorch: denial of serviceCVE-2025-55557fixed in 2.7.1PyTorch: buffer overflowCVE-2025-55558fixed in 2.7.1