CVE-2025-46153 Other in n/a
Inconsistent decomposition logic in PyTorch's bernoulli_p function affects dropout layers when fallback_random is enabled, potentially causing unexpected behavior.
- Exploited in the wild
- Not known
- Login needed
- Not needed
- Attackable
- Over the internet
- Victim action needed
- Not needed
- Fix
- Available
What is CVE-2025-46153?
CVE-2025-46153 is a medium-severity vulnerability in n/a, published on 25 Sep 2025. An attacker can exploit it over the internet without logging in, with no action from the victim.
Official description
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.
Weakness: CWE-1176
Which versions are affected by CVE-2025-46153?
2 open source packages ship an affected version, including pytorch and torch. It affects 1 product.
| Package | Affected | Fixed in |
|---|---|---|
pytorch Bitnami | From 2.6.0 | 2.7.0 |
torch PyPI | From 2.6.0 | 2.7.0 |
Products include Linuxfoundation Pytorch.
How do I fix CVE-2025-46153?
Upgrade pytorch to 2.7.0 or later.
Is CVE-2025-46153 being exploited?
There is no known exploitation of CVE-2025-46153 in the wild and no public exploit code.
Other vulnerabilities in n/a
- CVE-2024-48063 Deserialization in n/a
- CVE-2022-45907 PyTorch vulnerable to arbitrary code execution
- CVE-2025-32434 PyTorch: `torch.load` with `weights_only=True` leads to remote code execution
- CVE-2026-24747 PyTorch Vulnerable to Remote Code Execution via Untrusted Checkpoint Files
- CVE-2024-31583 Pytorch use-after-free vulnerability
Sources
Also known as BIT-pytorch-2025-46153, DEBIAN-CVE-2025-46153, ECHO-df59-c098-b576, GHSA-h36m-9qjr-pv93, PYSEC-2025-202, UBUNTU-CVE-2025-46153.