groundlens

12/15/2025

groundlens is a Python library for detecting LLM hallucinations using embedding geometry.

Two methods, zero LLM dependencies:

• SGI (Semantic Grounding Index) — measures whether a response engages its source material. For RAG systems with context.
• DGI (Directional Grounding Index) — measures whether a response’s displacement in embedding space aligns with grounded patterns. Context-free.

pip install groundlens

from groundlens import compute_sgi

result = compute_sgi(
    question="What is the capital of France?",
    context="France is in Western Europe. Its capital is Paris.",
    response="The capital of France is Paris.",
)
# result.flagged → False

Domain-specific calibration achieves AUROC 0.90-0.99. Generic calibration: AUROC ~0.76.

GitHub repository · Documentation · PyPI