🚀 Our new explainability method beats LIME for text models
📊 Across sentiment, toxicity, and syntax tasks:
• +34–135% higher fidelity vs LIME
• Often surpasses gradient-based methods
• Fully model-agnostic
📄 Paper: arxiv.org/pdf/2601.11746
#XAI #NLPResearch #ExplainableAI #ResponsibleAI #phdlife
Apollinaire Poli Nemkova
@nemkova.bsky.social
AI Researcher
2 Followers |
3 Following |
2 Posts |
Joined: 25.11.2025 |
1.1602
