| Version | Supported |
|---|---|
| 0.1.x | ✅ |
Please DO NOT create public GitHub Issues for security vulnerabilities.
Instead, report security issues to: security@prometorch.ru
We will respond within 48 hours and provide a fix timeline.
- Buffer overflow protection: Allocator has alignment overflow checks (Allocator.h)
- Integer overflow protection: numel overflow checks in TensorImpl.h
- Type safety: copy_() validates dtype match
- CUDA: No
cudaFreeon shutdown (prevents double-free, matches PyTorch pattern)