publications

2026

1. SWE-fficiency: Can Language Models Optimize Real-World Repositories on Real Workloads?

   Jeffrey Jian Ma, Milad Hashemi, Amir Yazdanbakhsh, Kevin Swersky, Ofir Press, Enhui Li, Vijay Janapa Reddi, and Parthasarathy Ranganathan

   In The Fourth-Third International Conference on Machine Learning, 2026

   Abs

   Optimizing the performance of large-scale software repositories demands expertise in code reasoning and software engineering (SWE) to reduce runtime while preserving program correctness. However, most benchmarks emphasize what to fix rather than how to fix code. We introduce SWE-fficiency, a benchmark for evaluating repository-level performance optimization on real workloads. Our suite contains 498 tasks across nine widely used data-science, machine-learning, and HPC repositories (e.g., numpy, pandas, scipy): given a complete codebase and a slow workload, an agent must investigate code semantics, localize bottlenecks and relevant tests, and produce a patch that matches or exceeds expert speedup while passing the same unit tests. To enable this how-to-fix evaluation, our automated pipeline scrapes GitHub pull requests for performance-improving edits, combining keyword filtering, static analysis, coverage tooling, and execution validation to both confirm expert speedup baselines and identify relevant repository unit tests. Empirical evaluation of state-of-the-art agents reveals significant underperformance. On average, agents achieve less than 0.15x the expert speedup: agents struggle in localizing optimization opportunities, reasoning about execution across functions, and maintaining correctness in proposed edits. We release the benchmark and accompanying data pipeline to facilitate research on automated performance engineering and long-horizon software reasoning.

2. QuArch: A Benchmark for Evaluating LLM Reasoning in Computer Architecture

   Shvetank Prakash, Andrew Cheng, Arya Tschand, Mark Mazumder, Varun Gohil, Jeffrey Ma, Jason Yik, Zishen Wan, Jessica Quaye, Elisavet Lydia Alvanaki, Avinash Kumar, Chandrashis Mazumdar, Tuhin Khare, Alexander Ingare, Ikechukwu Uchendu, Radhika Ghosal, Abhishek Tyagi, Chenyu Wang, Andrea Mattia Garavagno, Sarah Gu, Alice Guo, Grace Hur, Luca Carloni, Tushar Krishna, Ankita Nayak, Amir Yazdanbakhsh, and Vijay Janapa Reddi

   2026

   Abs

   The field of computer architecture, which bridges high-level software abstractions and low-level hardware implementations, remains absent from current large language model (LLM) evaluations. To this end, we present QuArch (pronounced ’quark’), the first benchmark designed to facilitate the development and evaluation of LLM knowledge and reasoning capabilities specifically in computer architecture. QuArch provides a comprehensive collection of 2,671 expert-validated question-answer (QA) pairs covering various aspects of computer architecture, including processor design, memory systems, and interconnection networks. Our evaluation reveals that while frontier models possess domain-specific knowledge, they struggle with skills that require higher-order thinking in computer architecture. Frontier model accuracies vary widely (from 34% to 72%) on these advanced questions, highlighting persistent gaps in architectural reasoning across analysis, design, and implementation QAs. By holistically assessing fundamental skills, QuArch provides a foundation for building and measuring LLM capabilities that can accelerate innovation in computing systems. With over 140 contributors from 40 institutions, this benchmark represents a community effort to set the standard for architectural reasoning in LLM evaluation.

2025

1. QuArch: A Question-Answering Dataset for AI Agents in Computer Architecture

   Shvetank Prakash, Andrew Cheng, Jason Yik, Arya Tschand, Radhika Ghosal, Ikechukwu Uchendu, Jessica Quaye, Jeffrey Ma, Shreyas Grampurohit, Sofia Giannuzzi, Arnav Balyan, Fin Amin, Aadya Pipersenia, Yash Choudhary, Ankita Nayak, Amir Yazdanbakhsh, and Vijay Janapa Reddi

   IEEE Computer Architecture Letters, 2025

   Abs

   We introduce QuArch, a dataset of 1500 human-validated question-answer pairs designed to evaluate and enhance language models’ understanding of computer architecture. The dataset covers areas including processor design, memory systems, and performance optimization. Our analysis highlights a significant performance gap: the best closed-source model achieves 84% accuracy, while the top small open-source model reaches 72%. We observe notable struggles in memory systems, interconnection networks, and benchmarking. Fine-tuning with QuArch improves small model accuracy by up to 8%, establishing a foundation for advancing AI-driven computer architecture research. The dataset and leaderboard are at https://harvard-edge.github.io/QuArch/.

2. Understanding Silent Data Corruption in LLM Training

   Jeffrey Ma, Hengzhi Pei, Leonard Lausen, and George Karypis

   2025

   Abs

   As the scale of training large language models (LLMs) increases, one emergent failure is silent data corruption (SDC), where hardware produces incorrect computations without explicit failure signals. In this work, we are the first to investigate the impact of real-world SDCs on LLM training by comparing model training between healthy production nodes and unhealthy nodes exhibiting SDCs. With the help from a cloud computing platform, we access the unhealthy nodes that were swept out from production by automated fleet management. Using deterministic execution via XLA compiler and our proposed synchronization mechanisms, we isolate and analyze the impact of SDC errors on these nodes at three levels: at each submodule computation, at a single optimizer step, and at a training period. Our results reveal that the impact of SDCs on computation varies on different unhealthy nodes. Although in most cases the perturbations from SDCs on submodule computation and gradients are relatively small, SDCs can lead models to converge to different optima with different weights and even cause spikes in the training loss. Our analysis sheds light on further understanding and mitigating the impact of SDCs.

3. A2Perf: Real-World Autonomous Agents Benchmark

   Ikechukwu Uchendu, Jason Jabbour, Korneel Van Berghe, Joel Runevic, Matthew Stewart, Jeffrey Ma, Srivatsan Krishnan, Izzeddin Gur, Austin Huang, Colton Bishop, Paige Bailey, Wenjie Jiang, Ebrahim M. Songhori, Sergio Guadarrama, Jie Tan, Jordan K. Terry, Aleksandra Faust, and Vijay Janapa Reddi

   2025

   Abs

   Autonomous agents and systems cover a number of application areas, from robotics and digital assistants to combinatorial optimization, all sharing common, unresolved research challenges. It is not sufficient for agents to merely solve a given task; they must generalize to out-of-distribution tasks, perform reliably, and use hardware resources efficiently during training and inference, among other requirements. Several methods, such as reinforcement learning and imitation learning, are commonly used to tackle these problems, each with different trade-offs. However, there is a lack of benchmarking suites that define the environments, datasets, and metrics which can be used to provide a meaningful way for the community to compare progress on applying these methods to real-world problems. We introduce A2Perf–a benchmark with three environments that closely resemble real-world domains: computer chip floorplanning, web navigation, and quadruped locomotion. A2Perf provides metrics that track task performance, generalization, system resource efficiency, and reliability, which are all critical to real-world applications. Using A2Perf, we demonstrate that web navigation agents can achieve latencies comparable to human reaction times on consumer hardware, reveal reliability trade-offs between algorithms for quadruped locomotion, and quantify the energy costs of different learning approaches for computer chip-design. In addition, we propose a data cost metric to account for the cost incurred acquiring offline data for imitation learning and hybrid algorithms, which allows us to better compare these approaches. A2Perf also contains several standard baselines, enabling apples-to-apples comparisons across methods and facilitating progress in real-world autonomy. As an open-source benchmark, A2Perf is designed to remain accessible, up-to-date, and useful to the research community over the long term.

4. When Silicon Fails Silently: Characterizing Hardware-Induced Corruption in LLM Training

   Jeffrey Ma, Hengzhi Pei, Leonard Lausen, and George Karypis

   In 2025 IEEE 31st International Symposium on On-Line Testing and Robust System Design (IOLTS), 2025

   Abs

   As the scale of training large language models (LLMs) increases, one emergent failure is silent data corruption (SDC), where hardware produces incorrect computations without explicit failure signals. In this work, we summarize the first investigation of the impact of real-world SDCs on LLM training presented in our ACL work [1]. In our investigation, we compare model training between healthy production nodes and unhealthy nodes exhibiting SDCs. With the help from a cloud computing platform, we access the unhealthy nodes that were swept out from production by automated fleet management. Using deterministic execution via XLA compiler and our proposed synchronization mechanisms, we isolate and analyze the impact of SDC errors on these nodes at three levels: at each submodule computation, at a single optimizer step, and at a training period. Our results reveal that the impact of SDCs on computation varies on different unhealthy nodes. Although in most cases the perturbations from SDCs on submodule computation and gradients are relatively small, SDCs can lead models to converge to different optima with different weights and even cause spikes in the training loss.

5. When Silicon Fails Silently: Characterizing Hardware-Induced Corruption in LLM Training

   Jeffrey Ma, Hengzhi Pei, Leonard Lausen, and George Karypis

   2025

   Abs

   As the scale of training large language models (LLMs) increases, one emergent failure is silent data corruption (SDC), where hardware produces incorrect computations without explicit failure signals. In this work, we summarize the first investigation of the impact of real-world SDCs on LLM training presented in our ACL work [1]. In our investigation, we compare model training between healthy production nodes and unhealthy nodes exhibiting SDCs. With the help from a cloud computing platform, we access the unhealthy nodes that were swept out from production by automated fleet management. Using deterministic execution via XLA compiler and our proposed synchronization mechanisms, we isolate and analyze the impact of SDC errors on these nodes at three levels: at each submodule computation, at a single optimizer step, and at a training period. Our results reveal that the impact of SDCs on computation varies on different unhealthy nodes. Although in most cases the perturbations from SDCs on submodule computation and gradients are relatively small, SDCs can lead models to converge to different optima with different weights and even cause spikes in the training loss.

6. SwizzlePerf: Hardware-Aware LLMs for GPU Kernel Performance Optimization

   Arya Tschand, Kesavan Ramakrishnan, Muhammad A. Awad, Ryan Swann, Jeffrey Jian Ma, Keith Lowery, and Vijay Janapa Reddi

   In NeurIPS 2025 Workshop on Machine Learning for Systems, 2025

   Abs

   Large language models (LLMs) have shown progress in GPU kernel performance engineering using inefficient search-based methods that optimize around runtime. Any existing approach lacks a key characteristic that human performance engineers rely on for near-optimal utilization - hardware-awareness. By leveraging the workload’s specific memory access patterns, architecture specifications, filtered profiling logs, and reflections on historical performance, we can make software-level optimizations that are tailored to the underlying hardware. SwizzlePerf automatically generates spatial optimizations for GPU kernels on disaggregated architectures by giving LLMs explicit hardware-awareness. For a GEMM kernel, SwizzlePerf takes less than 5 minutes to generate the same hardware-specific optimal swizzling pattern that took expert performance engineers 2 weeks to find. On a suite of 10 diverse ML and Science kernels, SwizzlePerf can generate swizzling patterns for 9 of the kernels that achieve up to a 2.06x speedup and 70% improvement in L2 hit rate. This work is the first of many steps toward systematically creating hardware-aware LLM performance engineering agents.

2024

1. FedStaleWeight: Buffered Asynchronous Federated Learning with Fair Aggregation via Staleness Reweighting

   Jeffrey Ma, Alan Tu, Yiling Chen, and Vijay Janapa Reddi

   2024

   Abs

   Federated Learning (FL) endeavors to harness decentralized data while preserving privacy, facing challenges of performance, scalability, and collaboration. Asynchronous Federated Learning (AFL) methods have emerged as promising alternatives to their synchronous counterparts bounded by the slowest agent, yet they add additional challenges in convergence guarantees, fairness with respect to compute heterogeneity, and incorporation of staleness in aggregated updates. Specifically, AFL biases model training heavily towards agents who can produce updates faster, leaving slower agents behind, who often also have differently distributed data which is not learned by the global model. Naively upweighting introduces incentive issues, where true fast updating agents may falsely report updates at a slower speed to increase their contribution to model training. We introduce FedStaleWeight, an algorithm addressing fairness in aggregating asynchronous client updates by employing average staleness to compute fair re-weightings. FedStaleWeight reframes asynchronous federated learning aggregation as a mechanism design problem, devising a weighting strategy that incentivizes truthful compute speed reporting without favoring faster update-producing agents by upweighting agent updates based on staleness. Leveraging only observed agent update staleness, FedStaleWeight results in more equitable aggregation on a per-agent basis. We both provide theoretical convergence guarantees in the smooth, non-convex setting and empirically compare FedStaleWeight against the commonly used asynchronous FedBuff with gradient averaging, demonstrating how it achieves stronger fairness, expediting convergence to a higher global model accuracy. Finally, we provide an open-source test bench to facilitate exploration of buffered AFL aggregation strategies, fostering further research in asynchronous federated learning paradigms.

2021

1. Polymatrix Competitive Gradient Descent

   Jeffrey Ma, Alistair Letcher, Florian Schäfer, Yuanyuan Shi, and Anima Anandkumar

   Nov 2021

   Abs

   Many economic games and machine learning approaches can be cast as competitive optimization problems where multiple agents are minimizing their respective objective function, which depends on all agents’ actions. While gradient descent is a reliable basic workhorse for single-agent optimization, it often leads to oscillation in competitive optimization. In this work we propose polymatrix competitive gradient descent (PCGD) as a method for solving general sum competitive optimization involving arbitrary numbers of agents. The updates of our method are obtained as the Nash equilibria of a local polymatrix approximation with a quadratic regularization, and can be computed efficiently by solving a linear system of equations. We prove local convergence of PCGD to stable fixed points for n-player general-sum games, and show that it does not require adapting the step size to the strength of the player-interactions. We use PCGD to optimize policies in multi-agent reinforcement learning and demonstrate its advantages in Snake, Markov soccer and an electricity market game. Agents trained by PCGD outperform agents trained with simultaneous gradient descent, symplectic gradient adjustment, and extragradient in Snake and Markov soccer games and on the electricity market game, PCGD trains faster than both simultaneous gradient descent and the extragradient method.

2020

1. Diagnostic Image Quality Assessment and Classification in Medical Imaging: Opportunities and Challenges

   Jeffrey Jian Ma, Ukash Nakarmi, Cedric Yue Sik Kin, Christopher Sandino, Joseph Y. Cheng, Ali B. Syed, Peter Wei, John M. Pauly, and Shreyas Vasanawala

   Proceedings of the 2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI), May 2020

   Abs

   Magnetic Resonance Imaging (MRI) suffers from several artifacts, the most common of which are motion artifacts. These artifacts often yield images that are of non-diagnostic quality. To detect such artifacts, images are prospectively evaluated by experts for their diagnostic quality, which necessitates patient-revisits and rescans whenever non-diagnostic quality scans are encountered. This motivates the need to develop an automated framework capable of accessing medical image quality and detecting diagnostic and non-diagnostic images. In this paper, we explore several convolutional neural network-based frameworks for medical image quality assessment and investigate several challenges therein.

2. Analysis of Deep Learning models for Diagnostic Image Quality Assessment in Magnetic Resonance Imaging

   Jeffrey Jian Ma, Ukash Nakarmi, Cedric Yue Sik Kin, Joseph Y. Cheng, Christopher Sandino, Ali B. Syed, Peter Wei, John M. Pauly, and Shreyas Vasanawala

   Proceedings of the 2020 28th International Society for Magnetic Resonance in Medicine (ISMRM) Annual Meeting, Aug 2020

   Abs

3. Toward Continuous Social Phenotyping: Analyzing Gaze Patterns in an Emotion Recognition Task for Children With Autism Through Wearable Smart Glasses

   Anish Nag, Nick Haber, Catalin Voss, Serena Tamura, Jena Daniels, Jeffrey Jian Ma, Bryan Chiang, Shasta Ramachandran, Jessey Schwartz, Terry Winograd, Carl Feinstein, and Dennis P Wall

   Journal of Medical Internet Research (JMIR), Apr 2020

   Abs

   Several studies have shown that facial attention differs in children with autism. Measuring eye gaze and emotion recognition in children with autism is challenging, as standard clinical assessments must be delivered in clinical settings by a trained clinician. Wearable technologies may be able to bring eye gaze and emotion recognition into natural social interactions and settings.