1

IndicParam: Benchmark to evaluate LLMs on low-resource Indic Languages
TL;DR: Comprehensive benchmark to evaluate LLMs on low-resource Indic languages. Dataset publicly available on HuggingFace. Addresses critical gap in multilingual NLP evaluation.
Ayush Maheshwari, Kaushal Sharma, Vivek Patel, Aditya Maheshwari
PDF Cite Code Dataset
ParamBench: A Graduate-Level Benchmark for Evaluating LLM Understanding on Indic Subjects
TL;DR: Graduate-level benchmark for evaluating LLM understanding on Indic subjects. Dataset available on HuggingFace. Addresses need for rigorous multilingual evaluation.
Ayush Maheshwari, Kaushal Sharma, Vivek Patel, Aditya Maheshwari
PDF Cite Code Dataset
LexGen: Domain-aware Multilingual Lexicon Generation
TL;DR: Domain-aware multilingual lexicon generation for 6 Indian languages across 8 domains using routing-based architecture. Released benchmark with 75K+ translation pairs. Accepted at ACL Main Conference 2025.
Ayush Maheshwari, Atul Kumar Singh, Karthika NJ, Krishnakant Bhat, Preethi Jyothi, Ganesh Ramakrishnan
PDF Cite Code
ARISE: Iterative Rule Induction and Synthetic Data Generation for Text Classification
TL;DR: ARISE iteratively induces rules and generates synthetic data for text classification via bootstrapping. Outperforms complex methods like contrastive learning across diverse domains and languages. Published at NAACL 2025 Findings.
Yaswanth M, Vaibhav Singh, Ayush Maheshwari, Amrith Krishna, Ganesh Ramakrishnan
PDF Cite
DictDis: Dictionary Constrained Disambiguation for Improved NMT
TL;DR: DictDis disambiguates between multiple dictionary candidate translations in lexically constrained NMT. Achieves 2-3 BLEU point improvements across regulatory, finance, engineering, and health domains. Published at EMNLP 2024 Findings.
Ayush Maheshwari, Preethi Jyothi, Ganesh Ramakrishnan
PDF Cite Code
FAIR: Filtering of Automatically Induced Rules
TL;DR: FAIR filters automatically induced rules using submodular optimization that accounts for precision, coverage, and conflicts. Outperforms existing rule-filtering approaches with statistically significant results. Published at EACL 2024.
Divya Jyoti Bajpai, Ayush Maheshwari, Manjesh Kumar Hanawal, Ganesh Ramakrishnan
PDF Cite Code
Sāmayik: A Benchmark and Dataset for English-Sanskrit Translation
TL;DR: First comprehensive benchmark and dataset for English-Sanskrit machine translation. Addresses critical gap in classical language NLP. Published at LREC-COLING 2024.
Ayush Maheshwari, Ashim Gupta, Amrith Krishna, Atul Kumar Singh, Ganesh Ramakrishnan, G. Anil Kumar, Jitin Singla
PDF Cite Code
EIGEN: Expert-Informed Joint Learning Aggregation for High-Fidelity Information Extraction from Document Images
TL;DR: EIGEN combines expert knowledge with joint learning for high-fidelity information extraction from medical document images. Applied to healthcare domain at ML4Health (NeurIPS) 2023.
Abhishek Singh, Venkatapathy Subramanian, Ayush Maheshwari, Pradeep Narayan, Devi Prasad Shetty, Ganesh Ramakrishnan
PDF Cite Code
UDAAN - Machine Learning based Post-Editing tool for Document Translation
TL;DR: A production-ready MT post-editing tool used by 100+ translators to translate technical content into Indian languages. Won Best Paper Award at CODS-COMAD 2023. Impact: First batch of engineering books translated using UDAAN were released by the President of India.
Ayush Maheshwari, Ajay Ravindran, Venkatapathy Subramanian, Ganesh Ramakrishnan
PDF Cite Code Project
UDAAN - Machine Learning based Post-Editing tool for Document Translation
SPEAR: Semi-supervised Data Programming in Python
TL;DR: Python library for programmatic data labeling using weak supervision. Reduces manual labeling effort by combining multiple weak labeling sources (rules, heuristics, models). Open-source toolkit with 100+ GitHub stars.
Sai Abhishek, Harshad Ingole, Parth Laturia, Vineeth Dorna, Ayush Maheshwari, Rishabh Iyer, Ganesh Ramakrishnan
PDF Cite Code Dataset Project