Svanika Naidu Doddavarapu
Taught small AI models to track what they believe, accepted at an IEEE conference
Small language models can reach a right answer for the wrong reasons. Svanika studied whether compact instruction-tuned models reason better when they explicitly track their belief state step by step, looking beyond final answers to how the model gets there.
AI & Language Models
First-author paper accepted at an IEEE conference
IEEE, 2026
Svanika Naidu Doddavarapu, YRI Fellow
A 10th grader at California High School in San Ramon interested in how AI models reason.
First author of a study on explicit belief-state tracking in compact language models, accepted at an IEEE conference.
First-author paper accepted at an IEEE conference
Studied reasoning in compact instruction-tuned language models
Evaluated belief-state tracking beyond final-answer accuracy
Devansh Deka, First-author paper accepted at an IEEE conference