Gregory Dexter

Kayhan Behdin, Qingquan Song, Aman Gupta et al.

Modern deep learning models are over-parameterized, where different optima can result in widely varying generalization performance. The Sharpness-Aware Minimization (SAM) technique modifies the fundamental loss function that steers gradient descent methods toward flatter minima, which are believed to exhibit enhanced generalization prowess. Our study delves into a specific variant of SAM known as micro-batch SAM (mSAM). This variation involves aggregating updates derived from adversarial perturbations across multiple shards (micro-batches) of a mini-batch during training. We extend a recently developed and well-studied general framework for flatness analysis to theoretically show that SAM achieves flatter minima than SGD, and mSAM achieves even flatter minima than SAM. We provide a thorough empirical evaluation of various image classification and natural language processing tasks to substantiate this theoretical advancement. We also show that contrary to previous work, mSAM can be implemented in a flexible and parallelizable manner without significantly increasing computational costs. Our implementation of mSAM yields superior generalization performance across a wide range of tasks compared to SAM, further supporting our theoretical framework.

Gregory Dexter, Rajiv Khanna, Jawad Raheel et al.

We present novel bounds for coreset construction, feature selection, and dimensionality reduction for logistic regression. All three approaches can be thought of as sketching the logistic regression inputs. On the coreset construction front, we resolve open problems from prior work and present novel bounds for the complexity of coreset construction methods. On the feature selection and dimensionality reduction front, we initiate the study of forward error bounds for logistic regression. Our bounds are tight up to constant factors and our forward error bounds can be extended to Generalized Linear Models.

Gregory Dexter, Petros Drineas, David P. Woodruff et al.

Sketching algorithms have recently proven to be a powerful approach both for designing low-space streaming algorithms as well as fast polynomial time approximation schemes (PTAS). In this work, we develop new techniques to extend the applicability of sketching-based approaches to the sparse dictionary learning and the Euclidean $k$-means clustering problems. In particular, we initiate the study of the challenging setting where the dictionary/clustering assignment for each of the $n$ input points must be output, which has surprisingly received little attention in prior work. On the fast algorithms front, we obtain a new approach for designing PTAS's for the $k$-means clustering problem, which generalizes to the first PTAS for the sparse dictionary learning problem. On the streaming algorithms front, we obtain new upper bounds and lower bounds for dictionary learning and $k$-means clustering. In particular, given a design matrix $\mathbf A\in\mathbb R^{n\times d}$ in a turnstile stream, we show an $\tilde O(nr/ε^2 + dk/ε)$ space upper bound for $r$-sparse dictionary learning of size $k$, an $\tilde O(n/ε^2 + dk/ε)$ space upper bound for $k$-means clustering, as well as an $\tilde O(n)$ space upper bound for $k$-means clustering on random order row insertion streams with a natural "bounded sensitivity" assumption. On the lower bounds side, we obtain a general $\tildeΩ(n/ε+ dk/ε)$ lower bound for $k$-means clustering, as well as an $\tildeΩ(n/ε^2)$ lower bound for algorithms which can estimate the cost of a single fixed set of candidate centers.

Gregory Dexter, Christos Boutsikas, Linkai Ma et al.

Motivated by the popularity of stochastic rounding in the context of machine learning and the training of large-scale deep neural network models, we consider stochastic nearness rounding of real matrices $\mathbf{A}$ with many more rows than columns. We provide novel theoretical evidence, supported by extensive experimental evaluation that, with high probability, the smallest singular value of a stochastically rounded matrix is well bounded away from zero -- regardless of how close $\mathbf{A}$ is to being rank deficient and even if $\mathbf{A}$ is rank-deficient. In other words, stochastic rounding \textit{implicitly regularizes} tall and skinny matrices $\mathbf{A}$ so that the rounded version has full column rank. Our proofs leverage powerful results in random matrix theory, and the idea that stochastic rounding errors do not concentrate in low-dimensional column spaces.

Gregory Dexter, Petros Drineas, Rajiv Khanna

We provide space complexity lower bounds for data structures that approximate logistic loss up to $ε$-relative error on a logistic regression problem with data $\mathbf{X} \in \mathbb{R}^{n \times d}$ and labels $\mathbf{y} \in \{-1,1\}^d$. The space complexity of existing coreset constructions depend on a natural complexity measure $μ_\mathbf{y}(\mathbf{X})$, first defined in (Munteanu, 2018). We give an $\tildeΩ(\frac{d}{ε^2})$ space complexity lower bound in the regime $μ_\mathbf{y}(\mathbf{X}) = O(1)$ that shows existing coresets are optimal in this regime up to lower order factors. We also prove a general $\tildeΩ(d\cdot μ_\mathbf{y}(\mathbf{X}))$ space lower bound when $ε$ is constant, showing that the dependency on $μ_\mathbf{y}(\mathbf{X})$ is not an artifact of mergeable coresets. Finally, we refute a prior conjecture that $μ_\mathbf{y}(\mathbf{X})$ is hard to compute by providing an efficient linear programming formulation, and we empirically compare our algorithm to prior approximate methods.

Aman Gupta, Rafael Celente, Abhishek Shivanna et al.

The Muon optimizer, based on matrix orthogonalization, has recently shown faster convergence and up to 2x computational efficiency over AdamW in LLM pretraining. Like AdamW, Muon is stateful, requiring storage of both model weights and accumulated gradients. While 8-bit AdamW variants mitigate this overhead using blockwise quantization, they are typically stable only under dynamic quantization - which improves stability on linear quantization for extreme values. In this paper, we introduce the 8-bit Muon optimizer using blockwise quantization, supporting both linear and dynamic schemes. We demonstrate that 8-bit Muon maintains stability under both, while delivering $\sim$74\% reduction in memory footprint compared to full-precision Muon. In extensive experiments, 8-bit Muon closely matches the performance of Muon while outperforming AdamW and 8-bit AdamW in pre-training a 1.6B model on 4B FineWeb tokens. It also shows competitive results when fine-tuning the Llama 3.2 3B model on post-training data. We also provide a theoretical perspective to help explain this robustness under quantization.

Kayhan Behdin, Ata Fatahibaarzi, Qingquan Song et al.

Large language models (LLMs) have demonstrated remarkable performance across a wide range of industrial applications, from search and recommendation systems to generative tasks. Although scaling laws indicate that larger models generally yield better generalization and performance, their substantial computational requirements often render them impractical for many real-world scenarios at scale. In this paper, we present a comprehensive set of insights for training and deploying small language models (SLMs) that deliver high performance for a variety of industry use cases. We focus on two key techniques: (1) knowledge distillation and (2) model compression via structured pruning and quantization. These approaches enable SLMs to retain much of the quality of their larger counterparts while significantly reducing training/serving costs and latency. We detail the impact of these techniques on a variety of use cases in a large professional social network platform and share deployment lessons, including hardware optimization strategies that improve speed and throughput for both predictive and reasoning-based applications in Recommendation Systems.

Gregory Dexter, Borja Ocejo, Sathiya Keerthi et al.

Stochastic Gradient Descent (SGD) stands as a cornerstone optimization algorithm with proven real-world empirical successes but relatively limited theoretical understanding. Recent research has illuminated a key factor contributing to its practical efficacy: the implicit regularization it instigates. Several studies have investigated the linear stability property of SGD in the vicinity of a stationary point as a predictive proxy for sharpness and generalization error in overparameterized neural networks (Wu et al., 2022; Jastrzebski et al., 2019; Cohen et al., 2021). In this paper, we delve deeper into the relationship between linear stability and sharpness. More specifically, we meticulously delineate the necessary and sufficient conditions for linear stability, contingent on hyperparameters of SGD and the sharpness at the optimum. Towards this end, we introduce a novel coherence measure of the loss Hessian that encapsulates pertinent geometric properties of the loss function that are relevant to the linear stability of SGD. It enables us to provide a simplified sufficient condition for identifying linear instability at an optimum. Notably, compared to previous works, our analysis relies on significantly milder assumptions and is applicable for a broader class of loss functions than known before, encompassing not only mean-squared error but also cross-entropy loss.

Gregory Dexter, Kevin Bello, Jean Honorio

Inverse Reinforcement Learning (IRL) is the problem of finding a reward function which describes observed/known expert behavior. The IRL setting is remarkably useful for automated control, in situations where the reward function is difficult to specify manually or as a means to extract agent preference. In this work, we provide a new IRL algorithm for the continuous state space setting with unknown transition dynamics by modeling the system using a basis of orthonormal functions. Moreover, we provide a proof of correctness and formal guarantees on the sample and time complexity of our algorithm. Finally, we present synthetic experiments to corroborate our theoretical guarantees.