Lune

Profiles2022-2026

Sham M Kakade

Harvard University

Also published as Sham Kakade

18Papers
7Last author
27Unique co-authors
0First author
0Solo

First and last author come from byline position, which some venues assign alphabetically.

Recent papers18

  1. The Potential of Second-Order Optimization for LLMs: A Study with Full Gauss-NewtonAuthor 3 of 4ICLR 2026 · 21 citations
  2. Weight Decay Improves Language Model PlasticityAuthor 4 of 4ICML 2026 · 3 citations
  3. A New Perspective on Shampoo's PreconditionerAuthor 5 of 6ICLR 2025
  4. The Role of Sparsity for Length Generalization in LLMsAuthor 4 of 5ICML 2025
  5. Scaling Laws in Linear Regression: Compute, Parameters, and DataAuthor 3 of 7NeurIPS 2024 · 57 citations
  6. Feature emergence via margin maximization: case studies in algebraic tasksAuthor 5 of 5ICLR 2024 · 35 citations
  7. Beyond Implicit Bias: The Insignificance of SGD Noise in Online LearningAuthor 5 of 6ICML 2024 · 8 citations
  8. Matching the Statistical Query Lower Bound for k-Sparse Parity Problems with Sign Stochastic Gradient DescentAuthor 4 of 4NeurIPS 2024 · 7 citations
  9. On Provable Copyright Protection for Generative ModelsAuthor 2 of 3ICML 2023 · 120 citations
  10. Finite-Sample Analysis of Learning High-Dimensional Single ReLU NeuronAuthor 6 of 6ICML 2023 · 9 citations
  11. Pareto Frontiers in Deep Feature Learning: Data, Compute, Width, and LuckAuthor 3 of 5NeurIPS 2023 · 8 citations
  12. Hidden Progress in Deep Learning: SGD Learns Parities Near the Computational LimitAuthor 4 of 6NeurIPS 2022 · 220 citations
  13. Unpacking Reward Shaping: Understanding the Benefits of Reward Engineering on Sample ComplexityAuthor 3 of 4NeurIPS 2022 · 115 citations
  14. Last Iterate Risk Bounds of SGD with Decaying Stepsize for Overparameterized Linear RegressionAuthor 5 of 5ICML 2022 · 38 citations
  15. Multi-Stage Episodic Control for Strategic Exploration in Text GamesAuthor 3 of 4ICLR 2022 · 30 citations
  16. The Power and Limitation of Pretraining-Finetuning for Linear Regression under Covariate ShiftAuthor 5 of 5NeurIPS 2022 · 29 citations
  17. Risk Bounds of Multi-Pass SGD for Least Squares in the Interpolation RegimeAuthor 5 of 5NeurIPS 2022 · 9 citations
  18. Recurrent Convolutional Neural Networks Learn Succinct Learning AlgorithmsAuthor 2 of 4NeurIPS 2022 · 2 citations
Dusk over the sea between two cliffs drawn in fine vertical lines