Beyond the Black Box: Interpretability of LLMs in Finance

3 ashater 1 5/23/2025, 5:12:28 PM papers.ssrn.com ↗

Comments (1)

ashater · 10h ago
Our paper introduces AI explainability methods, mechanistic interpretation, and novel Finance-specific use cases. Using Sparse Autoencoders, we zoom into LLM internals and highlight Finance-related features. We provide examples of using interpretability methods to enhance sentiment scoring, detect model bias, and improve trading applications.