Follow
Nikhil Ghosh
Nikhil Ghosh
Verified email at berkeley.edu
Title
Cited by
Cited by
Year
Deconstructing distributions: A pointwise framework of learning
G Kaplun, N Ghosh, S Garg, B Barak, P Nakkiran
arXiv preprint arXiv:2202.09931, 2022
192022
The Three Stages of Learning Dynamics in High-dimensional Kernel Methods
N Ghosh, S Mei, B Yu
arXiv preprint arXiv:2111.07167, 2021
152021
Landmark ordinal embedding
N Ghosh, Y Chen, Y Yue
Advances in Neural Information Processing Systems 32, 2019
102019
A Universal Trade-off Between the Model Size, Test Loss, and Training Loss of Linear Predictors
N Ghosh, M Belkin
SIAM Journal on Mathematics of Data Science 5 (4), 977-1004, 2023
52023
More is better in modern machine learning: when infinite overparameterization is optimal and overfitting is obligatory
JB Simon, D Karkada, N Ghosh, M Belkin
arXiv preprint arXiv:2311.14646, 2023
52023
LoRA+: Efficient Low Rank Adaptation of Large Models
S Hayou, N Ghosh, B Yu
arXiv preprint arXiv:2402.12354, 2024
42024
On the benefits of learning to route in mixture-of-experts models
N Dikkala, N Ghosh, R Meka, R Panigrahy, N Vyas, X Wang
The 2023 Conference on Empirical Methods in Natural Language Processing, 2023
32023
Alternating updates for efficient transformers
C Baykal, D Cutler, N Dikkala, N Ghosh, R Panigrahy, X Wang
Advances in Neural Information Processing Systems 36, 2024
22024
The Effect of SGD Batch Size on Autoencoder Learning: Sparsity, Sharpness, and Feature Learning
N Ghosh, S Frei, W Ha, B Yu
arXiv preprint arXiv:2308.03215, 2023
2023
The system can't perform the operation now. Try again later.
Articles 1–9