Inkit Padhi

Publications

• Final-Model-Only Data Attribution with a Unifying View of Gradient-Based Methods

  • • Dennis Wei
    • Inkit Padhi
    • et al.
  • 2025
  • NeurIPS 2025

  Conference paper

• When in Doubt, Cascade: Towards Building Efficient and Capable Guardrails

  • • Manish Nagireddy
    • Inkit Padhi
    • et al.
  • 2025
  • AIES 2025

  Conference paper

• Granite Guardian: Comprehensive LLM Safeguarding

  • • Inkit Padhi
    • Manish Nagireddy
    • et al.
  • 2025
  • NAACL 2025

  Conference paper

• Programming Refusal with Conditional Activation Steering

  • • Bruce Lee
    • Inkit Padhi
    • et al.
  • 2025
  • ICLR 2025

  Conference paper

• Contextual Value Alignment

  • • Kush Varshney
    • Miao Liu
    • et al.
  • 2025
  • ICASSP 2025

  Conference paper

• WikiContradict: A Benchmark for Evaluating LLMs on Real-World Knowledge Conflicts from Wikipedia

  • • Yufang Hou
    • Alessandra Pascale
    • et al.
  • 2024
  • NeurIPS 2024

  Conference paper

• Final-Model-Only Data Attribution with a Unifying View of Gradient-Based Methods

  • • Dennis Wei
    • Inkit Padhi
    • et al.
  • 2024
  • NeurIPS 2024

  Workshop paper

• Value Alignment from Unstructured Text

  • • Inkit Padhi
    • Karthikeyan Natesan Ramamurthy
    • et al.
  • 2024
  • NeurIPS 2024

  Workshop paper

• Value Alignment from Unstructured Text

  • • Inkit Padhi
    • Karthikeyan Natesan Ramamurthy
    • et al.
  • 2024
  • EMNLP 2024

  Conference paper

• ComVas: Contextual Moral Values Alignment System

  • • Inkit Padhi
    • Pierre Dognin
    • et al.
  • 2024
  • IJCAI 2024

  Conference paper