Downloads · 30 days
0
tris-onnx-architectural-backdoor-poc
renaissancefieldlite/tris-onnx-architectural-backdoor-poc
tris-onnx-architectural-backdoor-poc is a machine learning model from renaissancefieldlite. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
This repository is a public-safe proof-of-concept for an authorized Huntr model file vulnerability submission.
Access
Public
Updated Jul 2, 2026
Repo size
428 B
Likes
0
Public
Repo makeup
Click a slice to open those files.
.py3.4 KB · 47%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Jul 2, 2026
- Updated
- Jul 2, 2026
- SHA
- b5ee1f13
Model card
From the Hugging Face model README
Tris ONNX Architectural Backdoor PoC
This repository is a public-safe proof-of-concept for an authorized Huntr model file vulnerability submission.
The ONNX model is structurally valid and passes onnx.checker.check_model().
It behaves normally for benign input and activates a parallel backdoor path only
when a trigger value is provided.
Behavior
| input_score | decision_score |
|---|---|
12.0 | 12.0 |
67.0 | 999.0 |
Class
PAIT-ONNX-200: ONNX Model Contains Architectural Backdoor
Boundary
This PoC does not execute operating system commands and is not a parser memory-corruption exploit. It demonstrates model-integrity risk through a valid ONNX graph with dormant triggered behavior.
Files
tris_onnx_architectural_backdoor_poc.onnxtris_onnx_architectural_backdoor_poc_receipt.jsonbuild_onnx_arch_backdoor_poc.py
- License
- apache-2.0
- Created
- Jul 2, 2026
- Updated
- Jul 2, 2026