Text explainer¶
Under construction
TextExplainer is exported by XWhy, but its current explain() method raises NotImplementedError.
This explainer is intended for conventional text tasks that are distinct from prompt-response LLM explanation.
Planned documentation will cover:
- text classification and regression interfaces;
- token, word, phrase, and sentence perturbation units;
- masking and replacement strategies;
- embedding and distance selection;
- token-level visualisation;
- stability and faithfulness evaluation.
For implemented prompt-response analysis, use the LLM explainer.