Aarush Sinha

NLP Researcher

About Me

Profile

I am an undergraduate student at Vellore Institute of Technology - Chennai. My research is centered on enhancing the reasoning capabilities and overall performance of compact language models while ensuring efficiency. Additionally, I work on addressing hallucination issues in models across various modalities.
My guiding principle is "Chota SOTA Models" - creating small state-of-the-art models.

Information Retrieval NLP AI Safety Reasoning in LMs

Education

Vellore Institute of Technology-Chennai

BTech in Computer Science, 2021 - Present

Research Experience

Carnegie Mellon University

Research Student, 2025 - Present

Working on the evaluation of agents under Prof. Anand Rao.

AI Institute, University of South Carolina

Research Student, 2024 - Present

Mitigating hallucinations in Text2Video models under Prof. Amitava Das.

Indian Institute of Technology-Madras

Research Student, 2023 - Present

Building efficient and scalable dense retrievers under Prof. Nirav Bhatt.

Recent Updates

New Collaboration

Started working on Agentic Systems and Evaluation with Prof. Anand Rao at CMU.

April, 2025

New Preprint

Released Graph Masked Language Models on arXiv.

March, 2025

Paper Accepted

Our paper on Text2Video Hallucinations accepted to TrustNLP@NAACL.

February, 2025

Publications

2025

ViBe: A Text-to-Video Benchmark for Evaluating Hallucination in Large Multimodal Models

Vipula Rawte, Sarthak Jain, Aarush Sinha, Garv Kaushik, Aman Bansal, Prathiksha Rumale Vishwanath, Samyak Rajesh Jain, Aishwarya Naresh Reganti, Vinija Jain, Aman Chadha, Amit P Sheth, Amitava Das

The 5th Workshop on Trustworthy NLP @ NAACL2025

Conference

Graph Masked Language Models

Aarush Sinha, OM Kumar CU

arXiv preprint

Preprint

Predicting Liquidity-Aware Bond Yields using Causal GANs and Deep Reinforcement Learning with LLM Evaluation

Jaskaran Singh Walia, Aarush Sinha, Srinitish Srinivasan, Srihari Unnikrishnan

arXiv preprint

Preprint