Ajay Pravin Mahale
Record snapshot: Sep 3, 2026. Coverage and affiliations may be incomplete or historical.
Selected work
5- Causally Grounded Mechanistic Interpretability for LLMs with Faithful Natural-Language ExplanationsSaved credit: Contributor · 2026Explore people connected to this work →
- Explanation Multiplicity: Circuit-Level Interpretability Evidence Does Not Survive Defensible Analytic VariationSaved credit: Contributor · 2026Explore people connected to this work →
- Glassbox — mechanistic interpretability and EU AI Act compliance toolkitSaved credit: Maintainer · 2026Explore people connected to this work →
- Causally Grounded Mechanistic Interpretability and Faithful Natural-Language ExplanationsSaved credit: Maintainer · 2026Explore people connected to this work →
- Activation Patching FrameworkSaved credit: Maintainer · 2025Explore people connected to this work →