trustyai-examples

Examples for the Python bindings to TrustyAI's explainability library.

Binder

You can run the example Jupyter notebooks using mybinder.org by following this link.

Jupyter notebooks

Several example Jupyter notebooks are available.

Explainers

• Counterfactual explanations
• LIME explanations
• SHAP explanations
• PDP explanations
• Time-series explainers
  • TSLime¹
  • TSSaliency¹
  • Time-series Individual Conditional Expectation (TSICE)¹
  • Full examples
    • Energy load forecasting¹
    • Engine fault detection¹
• Detoxification
  • LLM detoxification using TMaRCo²

Metrics

• Bias/fairness
  • Group fairness metrics
• Language
  • Language performance

Footnotes

1. These explainers require the extra dependencies, installed with pip install trustyai[extras]. ↩ ↩² ↩³ ↩⁴ ↩⁵

2. Detoxification requires the optional dependencies, installed with pip install trustyai[detoxify]. ↩