TrustLens v0.1.2 — Now Available on PyPI

[Khanz9664]

TrustLens is now fully released and available via PyPI.

pip install trustlens

What is TrustLens?

TrustLens is a modular Python library for analyzing machine learning models beyond traditional accuracy metrics.

It focuses on model reliability, covering:

• Calibration (Brier Score, ECE)
• Failure analysis (confidence gaps, misclassifications)
• Bias detection (class imbalance, subgroup performance)
• Representation analysis (embedding separability, CKA)
• Explainability (Grad-CAM, faithfulness testing)

Why this matters

In many real-world systems, accuracy alone is misleading.

TrustLens helps answer:

• Is the model overconfident?
• Where does it fail dangerously?
• Is it fair across groups?

Quick Start

from trustlens import quick_analyze

report = quick_analyze(dataset="iris")
report.show()

Install

pip install trustlens

Contributing

We are actively looking for contributors across:

• metrics (new evaluation techniques)
• visualization improvements
• explainability integrations

Check the Issues section to get started.

---

Built and maintained by Shahid Ul Islam
GitHub: https://github.com/Khanz9664/TrustLens