Lattice AI Research

Research focus

Constitutional AI & AI Ethics (2)Interpretability & Mechanistic Interp (1)Red-Teaming & Adversarial Robustness (1)Eval Frameworks & Benchmarks (1)

Frequent co-authors

Agam Goyal (2)Yian Wang (1)Yuen Chen (1)Koyel Mukherjee (1)

Papers (2)

Apr 16, 2026

Apr 16, 2026·also UIUC

CausalDetox: Causal Head Selection and Intervention for Language Model Detoxification

LLMs can be detoxified with minimal performance impact by surgically intervening on a small subset of attention heads causally linked to toxicity, identified via a novel causal inference approach.

Yian Wang, Yuen Chen, Agam Goyal +1

Constitutional AI & AI Ethics Interpretability & Mechanistic Interp Red-Teaming & Adversarial Robustness

Apr 7, 2026

Apr 7, 2026·also UW-Madison

Masking or Mitigating? Deconstructing the Impact of Query Rewriting on Retriever Biases in RAG

LLM-based query rewriting in RAG can reduce retrieval bias by over 50%, but breaks down when biases combine adversarially, revealing the limits of query-side interventions.

Agam Goyal, Koyel Mukherjee, Apoorv Saxena +3

Constitutional AI & AI Ethics Eval Frameworks & Benchmarks Recommendation & Information Retrieval

Hari Sundaram

Publication activitypapers/week, last 8 weeks

Research focus

Frequent co-authors

Papers (2)

Search

Hari Sundaram

Publication activitypapers/week, last 8 weeks

Research focus

Frequent co-authors

Papers (2)