SurrealML unknown output normaliser SIGABRT PoC

This repository contains a benign security research PoC for a .surml wrapper that aborts the native SurrealML loader on an unknown output normaliser type.

Files:

control_same_size.surml
malicious_unknown_output_normaliser.surml
reproduce.py

Observed behavior:

control artifact:
- loads successfully
- raw_compute([1.0]) == [1.0]
malicious artifact:
- same size as control
- process terminates with SIGABRT during SurMlFile.load(..., Engine.ONNX)

Reproduction:

python3 build_poc.py
python3 reproduce.py

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support