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.
{
"osv_generated_from": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2025/46xxx/CVE-2025-46153.json",
"unresolved_ranges": [
{
"extracted_events": [
{
"fixed": "3.7.0"
}
],
"source": "DESCRIPTION"
}
],
"cna_assigner": "mitre"
}