Preetum Nakkiran

82 papers A* 22A 5B 2Misc 1Journal 50Unranked 2

Year	Rank	Type	Title / Venue / Authors
2025	A*	`conf`	A Formal Framework for Understanding Length Generalization in Transformers. ICLR Xinting Huang, Andy Yang, Satwik Bhattamishra, Yash Raj Sarrof, Andreas Krebs, Hattie Zhou, Preetum Nakkiran, Michael Hahn
2025	J	`jnl`	Classifier-Free Guidance is a Predictor-Corrector. Trans. Mach. Learn. Res. Arwen Bradley, Preetum Nakkiran
2025	A	`conf`	Composition and Control with Distilled Energy Diffusion Models and Sequential Monte Carlo. AISTATS James Thornton, Louis Béthune, Ruixiang Zhang, Arwen Bradley, Preetum Nakkiran, Shuangfei Zhai
2025	J	`jnl`	Composition and Control with Distilled Energy Diffusion Models and Sequential Monte Carlo. CoRR James Thornton, Louis Béthune, Ruixiang Zhang, Arwen Bradley, Preetum Nakkiran, Shuangfei Zhai
2025	A*	`conf`	Mechanisms of Projective Composition of Diffusion Models. ICML Arwen Bradley, Preetum Nakkiran, David Berthelot, James Thornton, Joshua M. Susskind
2025	J	`jnl`	Mechanisms of Projective Composition of Diffusion Models. CoRR Arwen Bradley, Preetum Nakkiran, David Berthelot, James Thornton, Joshua M. Susskind
2025	A*	`conf`	Normalizing Flows are Capable Generative Models. ICML Shuangfei Zhai, Ruixiang Zhang, Preetum Nakkiran, David Berthelot, Jiatao Gu, Huangjie Zheng, Tianrong Chen, Miguel Ángel Bautista, Navdeep Jaitly, Joshua M. Susskind
2025	J	`jnl`	Step-by-Step Diffusion: An Elementary Tutorial. Found. Trends Comput. Graph. Vis. Preetum Nakkiran, Arwen Bradley, Hattie Zhou, Madhu Advani
2025	J	`jnl`	Trace Length is a Simple Uncertainty Signal in Reasoning Models. CoRR Siddartha Devic, Charlotte Peale, Arwen Bradley, Sinead Williamson, Preetum Nakkiran, Aravind Gollakota
2025	J	`jnl`	Trained on Tokens, Calibrated on Concepts: The Emergence of Semantic Calibration in LLMs. CoRR Preetum Nakkiran, Arwen Bradley, Adam Golinski, Eugène Ndiaye, Michael Kirchhof, Sinead Williamson
2024	J	`jnl`	A Formal Framework for Understanding Length Generalization in Transformers. CoRR Xinting Huang, Andy Yang, Satwik Bhattamishra, Yash Raj Sarrof, Andreas Krebs, Hattie Zhou, Preetum Nakkiran, Michael Hahn
2024	J	`jnl`	Classifier-Free Guidance is a Predictor-Corrector. CoRR Arwen Bradley, Preetum Nakkiran
2024	A*	`conf`	How JEPA Avoids Noisy Features: The Implicit Bias of Deep Linear Self Distillation Networks. NeurIPS Etai Littwin, Omid Saremi, Madhu Advani, Vimal Thilak, Preetum Nakkiran, Chen Huang, Joshua M. Susskind
2024	J	`jnl`	How JEPA Avoids Noisy Features: The Implicit Bias of Deep Linear Self Distillation Networks. CoRR Etai Littwin, Omid Saremi, Madhu Advani, Vimal Thilak, Preetum Nakkiran, Chen Huang, Joshua M. Susskind
2024	A*	`conf`	LiDAR: Sensing Linear Probing Performance in Joint Embedding SSL Architectures. ICLR Vimal Thilak, Chen Huang, Omid Saremi, Laurent Dinh, Hanlin Goh, Preetum Nakkiran, Joshua M. Susskind, Etai Littwin
2024	A	`conf`	Loss Minimization Yields Multicalibration for Large Neural Networks. ITCS Jaroslaw Blasiok, Parikshit Gopalan, Lunjia Hu, Adam Tauman Kalai, Preetum Nakkiran
2024	J	`jnl`	Normalizing Flows are Capable Generative Models. CoRR Shuangfei Zhai, Ruixiang Zhang, Preetum Nakkiran, David Berthelot, Jiatao Gu, Huangjie Zheng, Tianrong Chen, Miguel Ángel Bautista, Navdeep Jaitly, Joshua M. Susskind
2024	A*	`conf`	Smooth ECE: Principled Reliability Diagrams via Kernel Smoothing. ICLR Jaroslaw Blasiok, Preetum Nakkiran
2024	J	`jnl`	Step-by-Step Diffusion: An Elementary Tutorial. CoRR Preetum Nakkiran, Arwen Bradley, Hattie Zhou, Madhu Advani
2024	A*	`conf`	Vanishing Gradients in Reinforcement Finetuning of Language Models. ICLR Noam Razin, Hattie Zhou, Omid Saremi, Vimal Thilak, Arwen Bradley, Preetum Nakkiran, Joshua M. Susskind, Etai Littwin
2024	A*	`conf`	What Algorithms can Transformers Learn? A Study in Length Generalization. ICLR Hattie Zhou, Arwen Bradley, Etai Littwin, Noam Razin, Omid Saremi, Joshua M. Susskind, Samy Bengio, Preetum Nakkiran
2024	A*	`conf`	When is Multicalibration Post-Processing Necessary? NeurIPS Dutch Hansen, Siddartha Devic, Preetum Nakkiran, Vatsal Sharan
2024	J	`jnl`	When is Multicalibration Post-Processing Necessary? CoRR Dutch Hansen, Siddartha Devic, Preetum Nakkiran, Vatsal Sharan
2023	A*	`conf`	A Unifying Theory of Distance from Calibration. STOC Jaroslaw Blasiok, Parikshit Gopalan, Lunjia Hu, Preetum Nakkiran
2023	A*	`conf`	Deconstructing Distributions: A Pointwise Framework of Learning. ICLR Gal Kaplun, Nikhil Ghosh, Saurabh Garg, Boaz Barak, Preetum Nakkiran
2023	J	`jnl`	Empirical Limitations of the NTK for Understanding Scaling Laws in Deep Learning. Trans. Mach. Learn. Res. Nikhil Vyas, Yamini Bansal, Preetum Nakkiran
2023	J	`jnl`	LiDAR: Sensing Linear Probing Performance in Joint Embedding SSL Architectures. CoRR Vimal Thilak, Chen Huang, Omid Saremi, Laurent Dinh, Hanlin Goh, Preetum Nakkiran, Joshua M. Susskind, Etai Littwin
2023	J	`jnl`	Loss minimization yields multicalibration for large neural networks. CoRR Jaroslaw Blasiok, Parikshit Gopalan, Lunjia Hu, Adam Tauman Kalai, Preetum Nakkiran
2023	J	`jnl`	Perspectives on the State and Future of Deep Learning - 2023. CoRR Micah Goldblum, Anima Anandkumar, Richard G. Baraniuk, Tom Goldstein, Kyunghyun Cho, Zachary C. Lipton, Melanie Mitchell, Preetum Nakkiran, Max Welling, Andrew Gordon Wilson
2023	J	`jnl`	Smooth ECE: Principled Reliability Diagrams via Kernel Smoothing. CoRR Jaroslaw Blasiok, Preetum Nakkiran
2023	J	`jnl`	Vanishing Gradients in Reinforcement Finetuning of Language Models. CoRR Noam Razin, Hattie Zhou, Omid Saremi, Vimal Thilak, Arwen Bradley, Preetum Nakkiran, Joshua M. Susskind, Etai Littwin
2023	J	`jnl`	What Algorithms can Transformers Learn? A Study in Length Generalization. CoRR Hattie Zhou, Arwen Bradley, Etai Littwin, Noam Razin, Omid Saremi, Josh M. Susskind, Samy Bengio, Preetum Nakkiran
2023	A*	`conf`	When Does Optimizing a Proper Loss Yield Calibration? NeurIPS Jaroslaw Blasiok, Parikshit Gopalan, Lunjia Hu, Preetum Nakkiran
2023	J	`jnl`	When Does Optimizing a Proper Loss Yield Calibration? CoRR Jaroslaw Blasiok, Parikshit Gopalan, Lunjia Hu, Preetum Nakkiran
2022	J	`jnl`	A Unifying Theory of Distance from Calibration. CoRR Jaroslaw Blasiok, Parikshit Gopalan, Lunjia Hu, Preetum Nakkiran
2022	J	`jnl`	APE: Aligning Pretrained Encoders to Quickly Learn Aligned Multimodal Representations. CoRR Elan Rosenfeld, Preetum Nakkiran, Hadi Pouransari, Oncel Tuzel, Fartash Faghri
2022	J	`jnl`	Benign, Tempered, or Catastrophic: A Taxonomy of Overfitting. CoRR Neil Mallinar, James B. Simon, Amirhesam Abedsoltan, Parthe Pandit, Mikhail Belkin, Preetum Nakkiran
2022	A*	`conf`	Benign, Tempered, or Catastrophic: Toward a Refined Taxonomy of Overfitting. NeurIPS Neil Mallinar, James B. Simon, Amirhesam Abedsoltan, Parthe Pandit, Mikhail Belkin, Preetum Nakkiran
2022	J	`jnl`	Deconstructing Distributions: A Pointwise Framework of Learning. CoRR Gal Kaplun, Nikhil Ghosh, Saurabh Garg, Boaz Barak, Preetum Nakkiran
2022	J	`jnl`	General Strong Polarization. J. ACM Jaroslaw Blasiok, Venkatesan Guruswami, Preetum Nakkiran, Atri Rudra, Madhu Sudan
2022	A*	`conf`	Knowledge Distillation: Bad Models Can Be Good Role Models. NeurIPS Gal Kaplun, Eran Malach, Preetum Nakkiran, Shai Shalev-Shwartz
2022	J	`jnl`	Knowledge Distillation: Bad Models Can Be Good Role Models. CoRR Gal Kaplun, Eran Malach, Preetum Nakkiran, Shai Shalev-Shwartz
2022	J	`jnl`	Limitations of Neural Collapse for Understanding Generalization in Deep Learning. CoRR Like Hui, Mikhail Belkin, Preetum Nakkiran
2022	J	`jnl`	Limitations of the NTK for Understanding Generalization in Deep Learning. CoRR Nikhil Vyas, Yamini Bansal, Preetum Nakkiran
2022	J	`jnl`	Near-Optimal NP-Hardness of Approximating Max k-CSP Theory Comput. Pasin Manurangsi, Preetum Nakkiran, Luca Trevisan
2022	J	`jnl`	The Calibration Generalization Gap. CoRR A. Michael Carrell, Neil Mallinar, James Lucas, Preetum Nakkiran
2022	J	`jnl`	What You See is What You Get: Distributional Generalization for Algorithm Design in Deep Learning. CoRR Bogdan Kulynych, Yao-Yuan Yang, Yaodong Yu, Jaroslaw Blasiok, Preetum Nakkiran
2022	A*	`conf`	What You See is What You Get: Principled Deep Learning via Distributional Generalization. NeurIPS Bogdan Kulynych, Yao-Yuan Yang, Yaodong Yu, Jaroslaw Blasiok, Preetum Nakkiran
2021	A*	`conf`	Optimal Regularization can Mitigate Double Descent. ICLR Preetum Nakkiran, Prayaag Venkat, Sham M. Kakade, Tengyu Ma
2021	A*	`conf`	Revisiting Model Stitching to Compare Neural Representations. NeurIPS Yamini Bansal, Preetum Nakkiran, Boaz Barak
2021	J	`jnl`	Revisiting Model Stitching to Compare Neural Representations. CoRR Yamini Bansal, Preetum Nakkiran, Boaz Barak
2021	A*	`conf`	The Deep Bootstrap Framework: Good Online Learners are Good Offline Generalizers. ICLR Preetum Nakkiran, Behnam Neyshabur, Hanie Sedghi
2021	J	`jnl`	Turing-Universal Learners with Optimal Scaling Laws. CoRR Preetum Nakkiran
2020	A*	`conf`	Deep Double Descent: Where Bigger Models and More Data Hurt. ICLR Preetum Nakkiran, Gal Kaplun, Yamini Bansal, Tristan Yang, Boaz Barak, Ilya Sutskever
2020	J	`jnl`	Distributional Generalization: A New Kind of Generalization. CoRR Preetum Nakkiran, Yamini Bansal
2020	J	`jnl`	Learning Rate Annealing Can Provably Help Generalization, Even for Convex Problems. CoRR Preetum Nakkiran
2020	J	`jnl`	Optimal Regularization Can Mitigate Double Descent. CoRR Preetum Nakkiran, Prayaag Venkat, Sham M. Kakade, Tengyu Ma
2020	J	`jnl`	The Deep Bootstrap: Good Online Learners are Good Offline Generalizers. CoRR Preetum Nakkiran, Behnam Neyshabur, Hanie Sedghi
2019	J	`jnl`	Adversarial Robustness May Be at Odds With Simplicity. CoRR Preetum Nakkiran
2019	A	`conf`	Algorithmic Polarization for Hidden Markov Models. ITCS Venkatesan Guruswami, Preetum Nakkiran, Madhu Sudan
2019	A*	`conf`	Computational Limitations in Robust Classification and Win-Win Results. COLT Akshay Degwekar, Preetum Nakkiran, Vinod Vaikuntanathan
2019	J	`jnl`	Deep Double Descent: Where Bigger Models and More Data Hurt. CoRR Preetum Nakkiran, Gal Kaplun, Yamini Bansal, Tristan Yang, Boaz Barak, Ilya Sutskever
2019	J	`jnl`	More Data Can Hurt for Linear Regression: Sample-wise Double Descent. CoRR Preetum Nakkiran
2019	A*	`conf`	SGD on Neural Networks Learns Functions of Increasing Complexity. NeurIPS Preetum Nakkiran, Gal Kaplun, Dimitris Kalimeris, Tristan Yang, Benjamin L. Edelman, Fred Zhang, Boaz Barak
2019	J	`jnl`	SGD on Neural Networks Learns Functions of Increasing Complexity. CoRR Preetum Nakkiran, Gal Kaplun, Dimitris Kalimeris, Tristan Yang, Benjamin L. Edelman, Fred Zhang, Boaz Barak
2019	—	`conf`	Tracking the l APPROX-RANDOM Chi-Ning Chou, Zhixian Lei, Preetum Nakkiran
2018	J	`jnl`	Algorithmic Polarization for Hidden Markov Models. CoRR Venkatesan Guruswami, Preetum Nakkiran, Madhu Sudan
2018	J	`jnl`	General Strong Polarization. CoRR Jaroslaw Blasiok, Venkatesan Guruswami, Preetum Nakkiran, Atri Rudra, Madhu Sudan
2018	J	`jnl`	General Strong Polarization. Electron. Colloquium Comput. Complex. Jaroslaw Blasiok, Venkatesan Guruswami, Preetum Nakkiran, Atri Rudra, Madhu Sudan
2018	A*	`conf`	General strong polarization. STOC Jaroslaw Blasiok, Venkatesan Guruswami, Preetum Nakkiran, Atri Rudra, Madhu Sudan
2018	J	`jnl`	The Generic Holdout: Preventing False-Discoveries in Adaptive Data Science. CoRR Preetum Nakkiran, Jaroslaw Blasiok
2018	J	`jnl`	Tracking the 𝓁 CoRR Chi-Ning Chou, Zhixian Lei, Preetum Nakkiran
2017	J	`jnl`	Predicting Positive and Negative Links with Noisy Queries: Theory & Practice. CoRR Charalampos E. Tsourakakis, Michael Mitzenmacher, Jaroslaw Blasiok, Ben Lawson, Preetum Nakkiran, Vasileios Nakos
2016	—	`conf`	Near-Optimal UGC-hardness of Approximating Max k-CSP_R. APPROX-RANDOM Pasin Manurangsi, Preetum Nakkiran, Luca Trevisan
2016	B	`conf`	Optimal systematic distributed storage codes with fast encoding. ISIT Preetum Nakkiran, K. V. Rashmi, Kannan Ramchandran
2015	Misc	`conf`	Automatic gain control and multi-style training for robust small-footprint keyword spotting with deep neural networks. ICASSP Rohit Prabhavalkar, Raziel Alvarez, Carolina Parada, Preetum Nakkiran, Tara N. Sainath
2015	A	`conf`	Compressing deep neural networks using a rank-constrained topology. INTERSPEECH Preetum Nakkiran, Raziel Alvarez, Rohit Prabhavalkar, Carolina Parada
2015	A	`conf`	Having Your Cake and Eating It Too: Jointly Optimal Erasure Codes for I/O, Storage, and Network-bandwidth. FAST K. V. Rashmi, Preetum Nakkiran, Jingyan Wang, Nihar B. Shah, Kannan Ramchandran
2015	J	`jnl`	Near-Optimal UGC-hardness of Approximating Max k-CSP_R. CoRR Pasin Manurangsi, Preetum Nakkiran, Luca Trevisan
2015	J	`jnl`	Optimal Systematic Distributed Storage Codes with Fast Encoding. CoRR Preetum Nakkiran, K. V. Rashmi, Kannan Ramchandran
2014	J	`jnl`	Fundamental Limits on Communication for Oblivious Updates in Storage Networks. CoRR Preetum Nakkiran, Nihar B. Shah, K. V. Rashmi
2014	B	`conf`	Fundamental limits on communication for oblivious updates in storage networks. GLOBECOM Preetum Nakkiran, Nihar B. Shah, K. V. Rashmi