CVE-2025-46153

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.

Database specific

{
    "cna_assigner": "mitre",
    "osv_generated_from": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2025/46xxx/CVE-2025-46153.json"
}

References

Affected packages

Git / github.com/pytorch/pytorch

Affected ranges

Type

GIT

Repo

https://github.com/pytorch/pytorch

Events

Introduced

1eba9b3aa3c43f86f4a2c807ac8e12c4a7767340

Fixed

3c7ea0e7ac5b770fccab1d42dd71497ca0596623

Database specific

Show details

{
    "cpe": "cpe:2.3:a:linuxfoundation:pytorch:*:-:*:*:*:python:*:*",
    "extracted_events": [
        {
            "introduced": "0"
        },
        {
            "fixed": "3.7.0"
        },
        {
            "introduced": "2.6.0"
        },
        {
            "fixed": "2.7.0"
        }
    ],
    "source": [
        "DESCRIPTION",
        "CPE_RANGE"
    ]
}

Database specific

source

"https://storage.googleapis.com/cve-osv-conversion/osv-output/CVE-2025-46153.json"