Nikunj Saunshi

43 papers A* 16B 1Journal 23Unranked 3

Year	Rank	Type	Title / Venue / Authors
2025	A*	`conf`	Efficient stagewise pretraining via progressive subnetworks. ICLR Abhishek Panigrahi, Nikunj Saunshi, Kaifeng Lyu, Sobhan Miryoosefi, Sashank J. Reddi, Satyen Kale, Sanjiv Kumar
2025	A*	`conf`	Learning to Keep a Promise: Scaling Language Model Decoding Parallelism with Learned Asynchronous Decoding. ICML Tian Jin, Ellie Y. Cheng, Zachary Ankner, Nikunj Saunshi, Blake M. Elias, Amir Yazdanbakhsh, Jonathan Ragan-Kelley, Suvinay Subramanian, Michael Carbin
2025	J	`jnl`	Learning to Keep a Promise: Scaling Language Model Decoding Parallelism with Learned Asynchronous Decoding. CoRR Tian Jin, Ellie Y. Cheng, Zachary Ankner, Nikunj Saunshi, Blake M. Elias, Amir Yazdanbakhsh, Jonathan Ragan-Kelley, Suvinay Subramanian, Michael Carbin
2025	A*	`conf`	Reasoning with Latent Thoughts: On the Power of Looped Transformers. ICLR Nikunj Saunshi, Nishanth Dikkala, Zhiyuan Li, Sanjiv Kumar, Sashank J. Reddi
2025	J	`jnl`	Reasoning with Latent Thoughts: On the Power of Looped Transformers. CoRR Nikunj Saunshi, Nishanth Dikkala, Zhiyuan Li, Sanjiv Kumar, Sashank J. Reddi
2025	J	`jnl`	StagFormer: Time Staggering Transformer Decoding for RunningLayers In Parallel. CoRR Dylan J. Cutler, Arun Kandoor, Nishanth Dikkala, Nikunj Saunshi, Xin Wang, Rina Panigrahy
2024	J	`jnl`	A Little Help Goes a Long Way: Efficient LLM Training by Leveraging Small LMs. CoRR Ankit Singh Rawat, Veeranjaneyulu Sadhanala, Afshin Rostamizadeh, Ayan Chakrabarti, Wittawat Jitkrittum, Vladimir Feinberg, Seungyeon Kim, Hrayr Harutyunyan, Nikunj Saunshi, Zachary Nado, Rakesh Shivanna, Sashank J. Reddi, Aditya Krishna Menon, Rohan Anil, Sanjiv Kumar
2024	A*	`conf`	Can Looped Transformers Learn to Implement Multi-step Gradient Descent for In-context Learning? ICML Khashayar Gatmiry, Nikunj Saunshi, Sashank J. Reddi, Stefanie Jegelka, Sanjiv Kumar
2024	J	`jnl`	Can Looped Transformers Learn to Implement Multi-step Gradient Descent for In-context Learning? CoRR Khashayar Gatmiry, Nikunj Saunshi, Sashank J. Reddi, Stefanie Jegelka, Sanjiv Kumar
2024	J	`jnl`	Efficient Stagewise Pretraining via Progressive Subnetworks. CoRR Abhishek Panigrahi, Nikunj Saunshi, Kaifeng Lyu, Sobhan Miryoosefi, Sashank J. Reddi, Satyen Kale, Sanjiv Kumar
2024	J	`jnl`	Landscape-Aware Growing: The Power of a Little LAG. CoRR Stefani Karp, Nikunj Saunshi, Sobhan Miryoosefi, Sashank J. Reddi, Sanjiv Kumar
2024	A*	`conf`	On the Inductive Bias of Stacking Towards Improving Reasoning. NeurIPS Nikunj Saunshi, Stefani Karp, Shankar Krishnan, Sobhan Miryoosefi, Sashank Jakkam Reddi, Sanjiv Kumar
2024	J	`jnl`	On the Inductive Bias of Stacking Towards Improving Reasoning. CoRR Nikunj Saunshi, Stefani Karp, Shankar Krishnan, Sobhan Miryoosefi, Sashank J. Reddi, Sanjiv Kumar
2024	J	`jnl`	On the Role of Depth and Looping for In-Context Learning with Task Diversity. CoRR Khashayar Gatmiry, Nikunj Saunshi, Sashank J. Reddi, Stefanie Jegelka, Sanjiv Kumar
2023	—	`conf`	Reasoning in Large Language Models Through Symbolic Math Word Problems. ACL (Findings) Vedant Gaur, Nikunj Saunshi
2023	J	`jnl`	Reasoning in Large Language Models Through Symbolic Math Word Problems. CoRR Vedant Gaur, Nikunj Saunshi
2023	A*	`conf`	Task-Specific Skill Localization in Fine-tuned Language Models. ICML Abhishek Panigrahi, Nikunj Saunshi, Haoyu Zhao, Sanjeev Arora
2023	J	`jnl`	Task-Specific Skill Localization in Fine-tuned Language Models. CoRR Abhishek Panigrahi, Nikunj Saunshi, Haoyu Zhao, Sanjeev Arora
2023	A*	`conf`	Understanding Influence Functions and Datamodels via Harmonic Analysis. ICLR Nikunj Saunshi, Arushi Gupta, Mark Braverman, Sanjeev Arora
2022	A*	`conf`	New Definitions and Evaluations for Saliency Methods: Staying Intrinsic, Complete and Sound. NeurIPS Arushi Gupta, Nikunj Saunshi, Dingli Yu, Kaifeng Lyu, Sanjeev Arora
2022	J	`jnl`	New Definitions and Evaluations for Saliency Methods: Staying Intrinsic, Complete and Sound. CoRR Arushi Gupta, Nikunj Saunshi, Dingli Yu, Kaifeng Lyu, Sanjeev Arora
2022	A*	`conf`	On Predicting Generalization using GANs. ICLR Yi Zhang, Arushi Gupta, Nikunj Saunshi, Sanjeev Arora
2022	A*	`conf`	Understanding Contrastive Learning Requires Incorporating Inductive Biases. ICML Nikunj Saunshi, Jordan T. Ash, Surbhi Goel, Dipendra Misra, Cyril Zhang, Sanjeev Arora, Sham M. Kakade, Akshay Krishnamurthy
2022	J	`jnl`	Understanding Contrastive Learning Requires Incorporating Inductive Biases. CoRR Nikunj Saunshi, Jordan T. Ash, Surbhi Goel, Dipendra Misra, Cyril Zhang, Sanjeev Arora, Sham M. Kakade, Akshay Krishnamurthy
2022	J	`jnl`	Understanding Influence Functions and Datamodels via Harmonic Analysis. CoRR Nikunj Saunshi, Arushi Gupta, Mark Braverman, Sanjeev Arora
2021	A*	`conf`	A Mathematical Exploration of Why Language Models Help Solve Downstream Tasks. ICLR Nikunj Saunshi, Sadhika Malladi, Sanjeev Arora
2021	A*	`conf`	A Representation Learning Perspective on the Importance of Train-Validation Splitting in Meta-Learning. ICML Nikunj Saunshi, Arushi Gupta, Wei Hu
2021	J	`jnl`	A Representation Learning Perspective on the Importance of Train-Validation Splitting in Meta-Learning. CoRR Nikunj Saunshi, Arushi Gupta, Wei Hu
2021	J	`jnl`	On Predicting Generalization using GANs. CoRR Yi Zhang, Arushi Gupta, Nikunj Saunshi, Sanjeev Arora
2021	A*	`conf`	Predicting What You Already Know Helps: Provable Self-Supervised Learning. NeurIPS Jason D. Lee, Qi Lei, Nikunj Saunshi, Jiacheng Zhuo
2020	J	`jnl`	A Mathematical Exploration of Why Language Models Help Solve Downstream Tasks. CoRR Nikunj Saunshi, Sadhika Malladi, Sanjeev Arora
2020	A*	`conf`	A Sample Complexity Separation between Non-Convex and Convex Meta-Learning. ICML Nikunj Saunshi, Yi Zhang, Mikhail Khodak, Sanjeev Arora
2020	J	`jnl`	A Sample Complexity Separation between Non-Convex and Convex Meta-Learning. CoRR Nikunj Saunshi, Yi Zhang, Mikhail Khodak, Sanjeev Arora
2020	J	`jnl`	Predicting What You Already Know Helps: Provable Self-Supervised Learning. CoRR Jason D. Lee, Qi Lei, Nikunj Saunshi, Jiacheng Zhuo
2020	A*	`conf`	Provable Representation Learning for Imitation Learning via Bi-level Optimization. ICML Sanjeev Arora, Simon S. Du, Sham M. Kakade, Yuping Luo, Nikunj Saunshi
2020	J	`jnl`	Provable Representation Learning for Imitation Learning via Bi-level Optimization. CoRR Sanjeev Arora, Simon S. Du, Sham M. Kakade, Yuping Luo, Nikunj Saunshi
2019	A*	`conf`	A Theoretical Analysis of Contrastive Unsupervised Representation Learning. ICML Nikunj Saunshi, Orestis Plevrakis, Sanjeev Arora, Mikhail Khodak, Hrishikesh Khandeparkar
2019	J	`jnl`	A Theoretical Analysis of Contrastive Unsupervised Representation Learning. CoRR Sanjeev Arora, Hrishikesh Khandeparkar, Mikhail Khodak, Orestis Plevrakis, Nikunj Saunshi
2018	—	`conf`	A Compressed Sensing View of Unsupervised Text Embeddings, Bag-of-n-Grams, and LSTMs. ICLR (Poster) Sanjeev Arora, Mikhail Khodak, Nikunj Saunshi, Kiran Vodrahalli
2018	—	`conf`	A La Carte Embedding: Cheap but Effective Induction of Semantic Feature Vectors. ACL (1) Mikhail Khodak, Nikunj Saunshi, Yingyu Liang, Tengyu Ma, Brandon Stewart, Sanjeev Arora
2018	J	`jnl`	A La Carte Embedding: Cheap but Effective Induction of Semantic Feature Vectors. CoRR Mikhail Khodak, Nikunj Saunshi, Yingyu Liang, Tengyu Ma, Brandon Stewart, Sanjeev Arora
2018	B	`conf`	A Large Self-Annotated Corpus for Sarcasm. LREC Mikhail Khodak, Nikunj Saunshi, Kiran Vodrahalli
2017	J	`jnl`	A Large Self-Annotated Corpus for Sarcasm. CoRR Mikhail Khodak, Nikunj Saunshi, Kiran Vodrahalli