You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

ShadowLogic Graph-Level Backdoor PoC - TensorFlow SavedModel

This is a security research proof-of-concept. Do NOT use this model for production.

Description

This TensorFlow SavedModel contains a graph-level backdoor using the ShadowLogic technique. The backdoor uses ONLY standard TensorFlow operators - no custom ops, no Lambda layers, no py_function, no ReadFile/WriteFile.

Trigger

A 2x2 red patch (R>0.9, G<0.1, B<0.1) in the top-left corner of the input image forces the model to classify the input as class 0 with logit 100.0.

ModelScan Result

ModelScan 0.8.8 scans this SavedModel with SavedModelTensorflowOpScan and reports: 0 issues found - the backdoor is completely invisible to the scanner.

Downloads last month: -

Inference Providers NEW

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