Rishabh Tiwari

Berkeley AI Research Lab

rishabhtiwari at berkeley dot edu

I am a Ph.D. student in Berkeley AI Research (BAIR) at EECS, UC Berkeley, advised by Prof. Kurt Keutzer, where I focus on efficient deep learning. Previously, I was a Pre-Doctoral Researcher at Google Deepmind where I worked with Dr. Pradeep Shenoy on developing algorithms to improve group-robustness of networks. My recent work involves mitigating simplicity bias without using bias labels, network architecture optimization by developing novel network pruning algorithm.

I am also a founding member and senior researcher at Transmute AI Labs, a non profit research lab, where I guide undergrad students to pursue research.

For updated details, please see my Google Scholar / LinkedIn pages.

selected publications

The Art of Scaling Reinforcement Learning Compute for LLMs

Devvrit Khatri, Lovish Madaan, Rishabh Tiwari, Rachit Bansal, and 3 more authors

International Conference on Learning Representations (ICLR), 2026 Oral

Bib PDF arXiv OpenReview

@inproceedings{khatri2025scalerl,
  title = {The Art of Scaling Reinforcement Learning Compute for LLMs},
  author = {Khatri, Devvrit and Madaan, Lovish and Tiwari, Rishabh and Bansal, Rachit and Duvvuri, Sai Surya and Zaheer, Manzil and Dhillon, Inderjit S and Brandfonbrener, David and Agarwal, Rishabh},
  booktitle = {International Conference on Learning Representations},
  year = {2026},
  url = {https://openreview.net/forum?id=FMjeC9Msws},
}

QuantSpec: Self-Speculative Decoding with Hierarchical Quantized KV Cache

Rishabh Tiwari, Haocheng Xi, Aditya Tomar, and 7 more authors

Under Review, 2025

Bib PDF

@article{tiwari2025quantspec,
  title = {QuantSpec: Self-Speculative Decoding with Hierarchical Quantized KV Cache},
  author = {Tiwari, Rishabh and Xi, Haocheng and Tomar, Aditya and Hooper, Coleman and Kim, Sehoon and Horton, Maxwell and Najibi, Mahyar and Mahoney, Michael W and Keutzer, Kurt and Gholami, Amir},
  journal = {Under Review},
  year = {2025},
}

Using Early Readouts to Mediate Featural Bias in Distillation

Rishabh Tiwari, Durga Sivasubramanian, Anmol Mekala, and 2 more authors

WACV, 2024

Bib PDF

@article{tiwari2023adversarial,
  title = {Using Early Readouts to Mediate Featural Bias in Distillation},
  author = {Tiwari, Rishabh and Sivasubramanian, Durga and Mekala, Anmol and Ramakrishnan, Ganesh and Shenoy, Pradeep},
  journal = {WACV},
  year = {2024},
}

Overcoming Simplicity Bias in Deep Networks using a Feature Sieve

Rishabh Tiwari, and Pradeep Shenoy

ICML, 2023

Bib PDF

@article{tiwari2023adversariam,
  title = {Overcoming Simplicity Bias in Deep Networks using a Feature Sieve},
  author = {Tiwari, Rishabh and Shenoy, Pradeep},
  journal = {ICML},
  year = {2023},
}

GCR: Gradient Coreset Based Replay Buffer Selection For Continual Learning

Rishabh Tiwari, Krishnateja Killamsetty, Rishabh Iyer, and 1 more author

In CVPR, 2022

Bib PDF

@inproceedings{tiwari2021gcr,
  title = {GCR: Gradient Coreset Based Replay Buffer Selection For Continual Learning},
  author = {Tiwari, Rishabh and Killamsetty, Krishnateja and Iyer, Rishabh and Shenoy, Pradeep},
  booktitle = {CVPR},
  year = {2022},
}

Chipnet: Budget-aware pruning with heaviside continuous approximations

Rishabh Tiwari, Udbhav Bamba, Arnav Chavan, and 1 more author

ICLR, 2021

Bib PDF

@article{tiwari2021chipnet,
  title = {Chipnet: Budget-aware pruning with heaviside continuous approximations},
  author = {Tiwari, Rishabh and Bamba, Udbhav and Chavan, Arnav and Gupta, Deepak K},
  journal = {ICLR},
  year = {2021},
}