Rohit Babbar

Research output

Publications

Selected publications on efficient machine learning, sparse neural networks, extreme multi-label classification and efficient LLM inference.

Google Scholar GitHub

2026

Hardware-Aware Dynamic Sparse Training for Large Output Spaces

Nasib Ullah, Jinbin Zhang, Jean Lucien Randrianantenaina, Erik Schultheis and Rohit Babbar.

43rd International Conference on Machine Learning (ICML), Seoul, South Korea.

PDF Code
2025

DynaSpec: Context-aware Dynamic Speculative Sampling for Large-Vocabulary Language Models

Jinbin Zhang, Nasib Ullah, Erik Schultheis and Rohit Babbar.

arXiv preprint.

PDF Code
2025

ELMO: Efficiency via Low-precision and Peak Memory Optimization in Large Output Spaces

Jinbin Zhang, Nasib Ullah, Erik Schultheis and Rohit Babbar.

42nd International Conference on Machine Learning (ICML), Vancouver, Canada.

PDF Code
2025

Unbiased Loss Functions for Multilabel Classification with Missing Labels

Erik Schultheis and Rohit Babbar.

Transactions of Machine Learning Research (TMLR).

PDF Code
2025

UniDEC: Unified Dual Encoder and Classifier Training for Extreme Multi-Label Classification

Siddhant Kharbanda, Devaansh Gupta, Gururaj K, Pankaj Malhotra, Cho-Jui Hsieh and Rohit Babbar.

The Web Conference, Sydney, Australia.

PDF Code
2025

How Well Calibrated are Extreme Multi-label Classifiers? An Empirical Analysis

Nasib Ullah, Erik Schultheis, Jinbin Zhang and Rohit Babbar.

31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Toronto, Canada.

PDF Code
2025

Large Language Model as a Teacher for Zero-shot Tagging at Extreme Scales

Nasib Ullah, Jinbin Zhang and Rohit Babbar.

31st International Conference on Computational Linguistics (COLING), Abu Dhabi, UAE.

PDF Code
2024

Navigating Extremes: Dynamic Sparsity in Large Output Spaces

Nasib Ullah, Erik Schultheis, Mike Lasby, Yani Ioannou and Rohit Babbar.

38th Conference on Neural Information Processing Systems (NeurIPS), Vancouver, Canada.

PDF Code
2024

Gandalf: Learning Label-label Correlations in Extreme Multi-label Classification via Label Features

Siddhant Kharbanda, Devaansh Gupta, Erik Schultheis, Atmadeep Banerjee, Cho-Jui Hsieh and Rohit Babbar.

30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Barcelona, Spain.

PDF
2024

A General Online Algorithm for Optimizing Complex Performance Metrics

Wojciech Kotlowski, Erik Schultheis, Marek Wydmuch, Rohit Babbar and Krzysztof Dembczyński.

41st International Conference on Machine Learning (ICML), Vienna, Austria.

PDF Code
2024

Consistent Algorithms for Multi-label Classification with Macro@k Metrics

Erik Schultheis, Marek Wydmuch, Wojciech Kotlowski, Rohit Babbar, Strom Borman and Krzysztof Dembczyński.

12th International Conference on Learning Representations (ICLR), Vienna, Austria.

PDF Code
2023

Generalized Test Utilities for Long-tail Performance in Extreme Multi-label Classification

Erik Schultheis, Marek Wydmuch, Wojciech Kotlowski, Rohit Babbar and Krzysztof Dembczyński.

37th Conference on Neural Information Processing Systems (NeurIPS), New Orleans, USA.

PDF Code
2023

Towards Memory-Efficient Training for Extremely Large Output Spaces: Learning with 500k Labels on a Single Commodity GPU

Erik Schultheis and Rohit Babbar.

ECML-PKDD, Turin, Italy.

PDF Code