Research

cool research stuff

1. MASK-OFF: Evaluating LLMs for omission of material fact

   Antyabha Rahman, Justin Michael Moeljono

   A benchmark of 500 scenarios that measures when LLMs leave out a material fact they know is harmful, without pressure to conceal it.

   AI4GOOD Workshop @ NeurIPS 2026 PosterOpenReviewCode

2. Probing the origins of reasoning performance: Representational quality for mathematical problem-solving in RL vs SFT finetuned models

   Antyabha Rahman, Akshaj Gurugubelli, Omar Ankit, Kevin Zhu, Aishwarya Balwani

   Investigating how reinforcement learning and supervised fine-tuning shape internal representations for mathematical reasoning.

   AAAI 2026 Workshop on XAI4ScienceOpenReviewarXiv