Artefactual#

Artefactual is a lightweight Python package for hallucination detection and robustness in LLM responses.

Features#

  • Practical: Precomputed calibration for several model families

  • Flexible: Works with vLLM, OpenAI Chat Completions, and the OpenAI Responses API formats

  • Detailed outputs: Compute both sequence-level and token-level uncertainty scores

Uncertainty Detectors#

  • EPR (Entropy Production Rate): Token- and sequence-level entropy-based metric exposing raw and calibrated probabilities

  • WEPR (Weighted EPR): Calibrated, learned weighted combination of entropy contributions