mithridatium 0.1.0

Framework for verifying integrity of pretrained AI models

Mithridatium 🛡️

A framework for verifying the integrity of pretrained AI models

Mithridatium is a research-driven project aimed at detecting backdoors and data poisoning in downloaded pretrained models or pipelines (e.g., from Hugging Face).
Our goal is to provide a modular, command-line tool that helps researchers and engineers trust the models they use.

---

🚀 Project Overview

Modern ML pipelines often reuse pretrained weights from online repositories.
This comes with risks:

• ❌ Backdoors — models behave normally until triggered by a specific pattern.
• ❌ Data poisoning — compromised training data leading to biased or malicious models.

Mithridatium analyzes pretrained models to flag potential compromises using multiple defenses from academic research.

---

Other Functionaly will be updated as the project goes on

Quickstart

python -m venv .venv && source .venv/bin/activate
pip install -e .
pip install pytest pytest-cov

# (A) Train demo models (fast settings)

#Clean model on 5 epochs (Increase epochs for better accuracy, but it will take longer)
python -m scripts.train_resnet18 --dataset clean  --epochs 5 --output_path models/resnet18_clean.pth

#Poisoned model on 5 epochs (Increase epochs for better accuracy, but it will take longer)
python -m scripts.train_resnet18 --dataset poison --train_poison_rate 0.1 --target_class 0 \
  --epochs 5 --output_path models/resnet18_poison.pth

# (B) Run detection
mithridatium detect --model models/resnet18_poison.pth --defense mmbd --data cifar10 --out reports/mmbd.json

# (C) See summary
cat reports/mmbd.json