← Publications / Mechanistic Interpretability & XAI

Faithfulness Is Not Actionability: Extrinsic Validation of Circuit Selectors Under Structured Pruning

Hem Gosalia, Pratinav Seth, Aditya Kasliwal, Vinay Kumar Sankarapu

NeurIPS 2026 Workshop2026My role: Senior authorWorkshop
Accuracy retention versus patching faithfulness for circuit selectors on Llama and Gemma under structured pruning

Cite

@inproceedings{gosalia2026faithfulness,
  title         = {{Faithfulness Is Not Actionability: Extrinsic Validation of Circuit Selectors Under Structured Pruning}},
  author        = {Hem Gosalia and Pratinav Seth and Aditya Kasliwal and Vinay Kumar Sankarapu},
  booktitle     = {NeurIPS 2026 Workshop},
  year          = {2026}
}
All →