Amruta Parulekar
1 indexed paper
Recent (6 mo)
1With code
0Influential cites
0Benchmarked
0Publications per year
126
Top categories
NLP×1AI×1Society×1HCI×1
Frequent co-authors
Research Timeline
2026
Toxic HallucinAItions: Perturbing Prompts and Tracing LLM Circuits
The paper demonstrates that increasing the toxicity of prompts significantly degrades the factual reliability of LLMs, a degradation linked to the selective amplification of perturbation-sensitive nodes within the model's internal circuits.
Highlighted terms show continued research focus across papers