- hook_functions.py: Check the safety of all loaded pickle files using
fickling.always_check_safety() - context_manager.py: Halt the deserialization of a malicious pickle file with the fickling context manager
- fault_injection.py: Perform a fault injection on a PyTorch model and then analyze the result with
check_safety - inject_mobilenet.py: Override the
evalmethod of a ML model using fickling and applyfickling.is_likely_safeto the model file - inject_pytorch.py: Inject a model loaded from a PyTorch file with malicious code using the PyTorch module
- numpy_poc.py: Analyze a malicious payload passed to
numpy.load() - trace_binary.py: Decompile a payload using the tracing module
- identify_pytorch_file.py: Identify 2 PyTorch files that are different formats
example
Directory actions
More options
Directory actions
More options
example
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|
parent directory.. | ||||