Ranjan Satapathy

large language model (4) causal mediation (2) sparse autoencoder (2) refusal behavior (2) chain-of-thought reasoning (1) text generation (1) interpretable machine learning (1) model interpretability (1) mechanistic interpretability (1) jailbreak attack (1) hallucination reduction (1) model interpretation (1) natural language explanation (1) activation patching (1) feature intervention (1) sustainability reporting (1) faithfulness measurement (1) causal faithfulness (1) environmental social governance (1) extractive rationalization (1)