paper_name stringlengths 4 421 | paper_url stringlengths 21 200 | paper_authors listlengths 0 125 | paper_abstract stringlengths 0 43.4k | paper_code stringlengths 1 139 | conf stringlengths 6 18 |
|---|---|---|---|---|---|
Special Section on Integrated Sensing and Communication Transceivers: Addressing Clutter, Interference, and Reconfigurability | https://doi.org/10.1109/JPROC.2026.3700246 | [
"Moeness G. Amin"
] | # | PROCIEEE2026 | |
Digital Phased Arrays: A Round Peg in a Square Hole | https://doi.org/10.1109/JPROC.2026.3672706 | [
"Benjamin R. Epstein"
] | Today’s state-of-the-art radio and microwave frequency phased arrays reflect an evolution in the steering of radio wave energy “beams” in a manner similar to the directing of radio energy to and from dish antennas. Phased arrays gained traction during WWII because of their ability to electronically steer transmitted an... | # | PROCIEEE2026 |
Next-Generation MIMO Transceivers for Integrated Sensing and Communications: Unique Security Vulnerabilities and Solutions | https://doi.org/10.1109/JPROC.2026.3697798 | [
"Kawon Han",
"Christos Masouros",
"Taneli Riihonen",
"Moeness G. Amin"
] | Integrated sensing and communications (ISAC), which are recognized as a key enabler for sixth generation (6G), have brought new opportunities for intelligent, sustainable, and connected wireless networks. Multiple-input–multiple-output (MIMO) transceiver technology lies at the core of this paradigm, providing the degre... | # | PROCIEEE2026 |
Clutter-Aware Integrated Sensing and Communication: Models, Methods, and Future Directions | https://doi.org/10.1109/JPROC.2026.3675476 | [
"Rang Liu",
"Peishi Li",
"Ming Li",
"A. Lee Swindlehurst"
] | Integrated sensing and communication (ISAC) can substantially improve spectral, hardware, and energy efficiency by unifying radar sensing and data communications. In wideband and scattering-rich environments, clutter often dominates weak target reflections and becomes a fundamental bottleneck for reliable sensing. Prac... | # | PROCIEEE2026 |
Reconfigurable Integrated Sensing and Communications (RISAC): Sparse MIMO and Hybrid Beamforming in Far and Near Fields | https://doi.org/10.1109/JPROC.2026.3670886 | [
"Xiangrong Wang",
"Fulvio Gini",
"Kaiquan Cai"
] | This article provides a comprehensive investigation into reconfigurable integrated sensing and communications (RISAC), an emerging paradigm designed to maximize the performance–cost tradeoff by exploiting the inherent spatial sparsity of multiple-input multiple-output (MIMO) arrays. Deviating from conventional static I... | # | PROCIEEE2026 |
RIS-Enabled Integrated Sensing and Communications: From Theory to Practice | https://doi.org/10.1109/JPROC.2026.3678371 | [
"J. Andrew Zhang",
"Kai Wu",
"Marco Di Renzo",
"Tie Jun Cui"
] | Reconfigurable intelligent surface (RIS)-assisted integrated sensing and communication (ISAC) is emerging as a key enabler for sixth-generation (6G) wireless systems, unifying communication, sensing, and control within a reconfigurable electromagnetic (EM) environment. This article presents a comprehensive review that ... | # | PROCIEEE2026 |
How We Got "Our" System of Units | https://doi.org/10.1109/JPROC.2026.3702728 | [
"Harold Kirkham"
] | # | PROCIEEE2026 | |
Special Issue on Earth Remote Sensing and Data Processing | https://doi.org/10.1109/JPROC.2026.3700784 | [
"Leung Tsang",
"Joel T. Johnson",
"Jiancheng Shi",
"Irena Hajnsek",
"Jeff Dozier"
] | # | PROCIEEE2026 | |
Spaceborne Synthetic Aperture Radar: Future Technologies and Mission Concepts | https://doi.org/10.1109/JPROC.2025.3621586 | [
"Alberto Moreira",
"Gerhard Krieger",
"Michelangelo Villano",
"Marwan Younis",
"Pau Prats-Iraola",
"Manfred Zink"
] | This article provides an overview of the state-of-the-art and future developments in spaceborne synthetic aperture radar (SAR). Today, we are experiencing a golden age of spaceborne SAR, with the number of satellites in orbit increasing rapidly. This article presents novel technologies and mission concepts associated w... | # | PROCIEEE2026 |
BIOMASS: ESA's P-Band SAR Mission | https://doi.org/10.1109/JPROC.2026.3687416 | [
"Klaus Scipal",
"Clément Albinet",
"Michele Caccia",
"Adriano Carbone",
"Nuno Carvalhais",
"Jérôme Chave",
"Jørgen Dall",
"Michael Fehringer",
"Antonio Leanza",
"Thuy Le Toan",
"Maktar Malik",
"Antonio Novelli",
"Philippe Paillou",
"Kostas Papathanassiou",
"Janice Patterson",
"Muriel P... | Utilizing a P-band synthetic aperture radar (SAR), the objective of BIOMASS is to deliver estimates of above-ground forest biomass, forest height (FH), and forest disturbance (FD), with unprecedented accuracy.The mission's primary scientific goal is to quantify the distribution and changes in forest biomass, thereby re... | # | PROCIEEE2026 |
High Revisit-Rate Tropical Cyclone Observations From the NASA TROPICS Satellite Constellation Mission | https://doi.org/10.1109/JPROC.2025.3582502 | [
"William J. Blackwell",
"Scott A. Braun",
"George R. Alvey",
"Robert Atlas",
"Ralf Bennartz",
"Jessica Braun",
"Kerri L. Cahoy",
"Ruiyao Chen",
"Galina Chirokova",
"Brittany Dahl",
"James Darlow",
"Mark DeMaria",
"Michael DiLiberto",
"Jason P. Dunion",
"Patrick Duran",
"Thomas J. Green... | New satellite constellations to provide high-resolution atmospheric observations from microwave (MW) sounders operating in low-Earth orbit are now coming online and are providing operationally useful data. The first of these missions, the NASA Time-Resolved Observations of Precipitation structure and storm Intensity wi... | # | PROCIEEE2026 |
Wideband Radiometry From P to S Band for Monitoring Polar Regions | https://doi.org/10.1109/JPROC.2026.3653571 | [
"Giovanni Macelloni",
"Kenneth C. Jezek",
"Marco Brogioni",
"Joel T. Johnson",
"Marion Leduc-Leballeur",
"Ghislain Picard",
"Ange Haddjeri",
"Lars Kaleschke",
"Jacqueline Boutin",
"Jean-Luc Vergely",
"Nicolas Kolodziejczyk",
"Laurent Bertino",
"Emmanuel P. Dinnat",
"Rasmus T. Tonboe",
"A... | This article reviews existing and planned contributions of spaceborne microwave radiometry from P to S band to new measurements of key geophysical variables with a particular focus on the polar regions. It summarizes the current state of spaceborne microwave radiometry to measure ice sheet thermal states, sea ice thick... | # | PROCIEEE2026 |
The In-Orbit Performance of Chinese First FengYun Rainfall Mission FY-3G | https://doi.org/10.1109/JPROC.2025.3605964 | [
"Peng Zhang",
"Jian Shang",
"Lin Chen",
"Shuze Jia",
"Honggang Yin",
"Shengli Wu",
"Wenqiang Lu",
"Hanlie Xu",
"Haofei Wang",
"Yixuan Shou",
"Guangzhen Cao",
"Sijie Chen",
"Manyun Lin",
"Aijun Zhu",
"Songyan Gu",
"Xiangang Zhao"
] | In April 2023, China launched its first precipitation measurement satellite, FengYun-3G (FY-3G), to accurately measure the spatial and vertical structures of precipitation in the middle and lower latitudes of Earth. Equipped with advanced instruments, including a dual-frequency precipitation radar (DPR), a microwave ra... | # | PROCIEEE2026 |
Spaceborne GNSS-R Bistatic Radar Remote Sensing, CYGNSS, and Future Missions | https://doi.org/10.1109/JPROC.2025.3583997 | [
"Christopher Ruf",
"Scott Gleason"
] | Global Navigation Satellite System Reflectometry (GNSS-R) is a relatively new type of radar developed for remote sensing of the Earth surface. It uses GNSS navigation signals such as those transmitted by the Global Positioning System (GPS) constellation of satellites as the radar transmitter. The radar receiver measure... | # | PROCIEEE2026 |
SDGSAT-1: A Professional Scientific Satellite for Monitoring SDG Indicators | https://doi.org/10.1109/JPROC.2025.3649854 | [
"Huadong Guo",
"Changyong Dou",
"Dong Liang",
"Nijun Jiang",
"Bihong Fu",
"Chengshan Han",
"Fansheng Chen",
"Peng Huang",
"Juanjuan Jing",
"Yu Zhang",
"Bo Cheng",
"Xiaoxue Feng",
"Yunwei Tang",
"Yonghong Hu",
"Lin Yan",
"Hao Zhang"
] | The implementation of the United Nations (UN) 2030 Agenda for Sustainable Development (2030 Agenda), with its 17 sustainable development goals (SDGs), faces challenges such as insufficient data, limited research methodologies, and uneven progress across regions. Earth observation (EO), particularly scientific satellite... | # | PROCIEEE2026 |
The Polar Radiant Energy in the Far-Infrared Experiment (PREFIRE), Broadband Thermal Spectrometry for Small Satellite Platforms | https://doi.org/10.1109/JPROC.2026.3667066 | [
"Brian J. Drouin",
"Marc C. Foote",
"Chad A. Greene",
"Brian H. Kahn",
"Sharmila Padmanabhan",
"Mary White",
"Xianglei Huang",
"Xiuhong Chen",
"Aronne Merrelli",
"Hazem Mahmoud",
"Kyle Mattingly",
"Timothy Michaels",
"Nathaniel B. Miller",
"Hamish Prince",
"Erin Wagner Hokanson",
"Nata... | For more than a decade, the remote sensing community has called for longwave spectral measurements of the Earth system to facilitate closing the radiation budget. In an effort to address this large gap in Earth-observing capability, NASA has supported the development and implementation of the Polar Radiant Energy in th... | # | PROCIEEE2026 |
AI in Satellite Remote Sensing of the Ocean | https://doi.org/10.1109/JPROC.2026.3664121 | [
"Xiaofeng Li",
"Qing Xu",
"Haoyu Wang",
"Haoyu Jiang",
"Xiaobin Yin",
"Shanshan Mu",
"Xiaolong Li",
"Hua Su",
"An Wang",
"Yi Yang",
"Yanjun Wang",
"Yibin Ren",
"Xudong Zhang",
"Yingjie Liu",
"Chong Wang"
] | Satellite remote sensing plays a fundamental role in observing oceanic processes by providing large-scale, long-term, and continuous measurements. With the increasing availability of multisource satellite data, challenges such as data gaps, complex environmental conditions, and the limitations of conventional retrieval... | # | PROCIEEE2026 |
Hybrid Deep Learning Models for Remote Sensing Image Processing | https://doi.org/10.1109/JPROC.2025.3638871 | [
"Matthieu Muller",
"Daniele Picone",
"Begüm Demir",
"Gustau Camps-Valls",
"Mauro Dalla Mura",
"Magnús Örn Úlfarsson",
"Jón Atli Benediktsson"
] | Core image processing tasks, such as super-resolution, denoising, deblurring, pansharpening, and atmospheric correction, underpin all optical remote sensing (RS) pipelines. Errors at this stage propagate through downstream applications, distorting land-cover maps, change detection, and climate records. Classical physic... | # | PROCIEEE2026 |
ProfInfer: An eBPF-based Fine-Grained LLM Inference Profiler | https://proceedings.mlsys.org/paper_files/paper/2026/hash/03dbc11a22e79cd38bea53cf518c2371-Abstract-Conference.html | [
"Bohua Zou",
"Debayan Roy",
"Dhimankumar Yogesh Airao",
"Weihao Xu",
"Binqi Sun",
"Yutao Liu",
"Haibo Chen"
] | As large language models (LLMs) move from research to production, understanding how inference engines behave in real time has become both essential and elusive. Unlike general-purpose engines such as ONNX Runtime, today’s LLM inference systems offer little operator-level visibility, leaving developers blind to where ti... | # | MLSys2026 |
SpecDiff-2: Scaling Diffusion Drafter Alignment For Faster Speculative Decoding | https://proceedings.mlsys.org/paper_files/paper/2026/hash/041dad5ed2191b44ba3ed0e00cdc3187-Abstract-Conference.html | [
"Jameson Sandler",
"Jacob Christopher",
"Tom Hartvigsen",
"Ferdinando Fioretto"
] | Speculative decoding has become the standard approach for accelerating Large Language Model (LLM) inference. It exploits a lossless draft-then-verify procedure to circumvent the latency of autoregressive decoding, achieving impressive speed-ups.
Yet, current speculative decoding approaches remain limited by two fun... | # | MLSys2026 |
CDLM: Consistency Diffusion Language Models for Faster Sampling | https://proceedings.mlsys.org/paper_files/paper/2026/hash/054de805fcceb78a201f5e9d53c85908-Abstract-Conference.html | [
"Minseo Kim",
"Chenfeng Xu",
"Coleman Hooper",
"Harman Singh",
"Ben Athiwaratkun",
"Ce Zhang",
"Kurt Keutzer",
"Amir Gholami"
] | Diffusion Language Models (DLMs) offer a promising parallel generation paradigm but suffer from slow inference due to numerous refinement steps and the inability to use standard KV caching. We introduce CDLM (Consistency Diffusion Language Models), a training-based acceleration method that simultaneously tackles both b... | https://github.com/SqueezeAILab/CDLM | MLSys2026 |
Unified LLM Model for Power, Performance, and Area Prediction from Hardware Code | https://proceedings.mlsys.org/paper_files/paper/2026/hash/0731f0e65559059eb9cd9d6f44ce2dd8-Abstract-Conference.html | [
"Armin Abdollahi",
"Mehdi Kamal",
"Massoud Pedram"
] | We present RocketPPA, a unified LLM-based model that predicts power, performance, and area for Verilog designs across technology nodes and optimization styles. The approach combines a large language model backbone with mixture-of-experts regression and low-rank adaptation for parameter efficiency. To improve generaliza... | # | MLSys2026 |
SuperInfer: SLO-Aware Rotary Scheduling and Memory Management for LLM Inference on Superchips | https://proceedings.mlsys.org/paper_files/paper/2026/hash/07fd64f9316f40193c6a4d87d8afa011-Abstract-Conference.html | [
"Jiahuan Yu",
"Mingtao Hu",
"Zichao Lin",
"Minjia Zhang"
] | Large Language Model (LLM) serving faces a fundamental tension between stringent latency Service Level Objectives (SLOs) and limited GPU memory capacity. When high request rates exhaust the KV cache budget, existing LLM inference systems often suffer severe head-of-line (HOL) blocking. While prior work explored PCIe-ba... | https://github.com/Supercomputing-System-AI-Lab/SuperInfer | MLSys2026 |
HELIOS : Adaptive Model And Early-Exit Selection for Efficient LLM Inference Serving | https://proceedings.mlsys.org/paper_files/paper/2026/hash/096b1019463f34eb241e87cfce8dfe16-Abstract-Conference.html | [
"Avinash Kumar",
"Shashank Nag",
"Jason Clemons",
"Lizy K. John",
"Poulami Das"
] | Early-Exit Large Language Models (EE-LLMs) enable high throughput inference by allowing tokens to exit early at intermediate layers. However, their throughput is limited by the computational and memory savings. Existing EE-LLM frameworks rely on a single model and therefore, their token generation latencies are bottlen... | # | MLSys2026 |
Scaling Up Large Language Models Serving Systems for Semantic Job Search | https://proceedings.mlsys.org/paper_files/paper/2026/hash/0a4c7cdfc0a4eb1b13bb84a9b6220c37-Abstract-Conference.html | [
"Kayhan Behdin",
"Qingquan Song",
"Sriram Vasudevan",
"Jian Sheng",
"Xiaojing Ma",
"Z Zhou",
"Chuanrui Zhu",
"Guoyao Li",
"Chanh Nguyen",
"Sayan Ghosh",
"Hejian Sang",
"Ata Fatahi",
"Sundara Raman Ramachandran",
"Xiaoqing Wang",
"Qing Lan",
"Vinay Y S",
"Qi Guo",
"Caleb Johnson",
... | Large Language Models (LLMs) have demonstrated impressive quality when applied to predictive tasks such as relevance ranking and semantic search. However, deployment of such LLMs remains prohibitively expensive for industry applications with strict latency and throughput requirements. In this work, we present lessons a... | # | MLSys2026 |
Massive-Scale Out-Of-Core UMAP on the GPU | https://proceedings.mlsys.org/paper_files/paper/2026/hash/0badcb4e95306df76a719409155e46e8-Abstract-Conference.html | [
"Jinsol Park",
"Corey J. Nolet",
"Edward Raff",
"Tim Oates",
"Akira Naruse"
] | The Uniform Manifold Approximation and Projection (UMAP) algorithm has become a widely popular technique to reduce the dimensionality of a set of vectors, both for visualization and as a pre-processing step for follow-on machine learning tasks. UMAP is often an integral part of iterative and exploratory workflows, but ... | # | MLSys2026 |
AccelOpt: A Self-Improving LLM Agentic System for AI Accelerator Kernel Optimization | https://proceedings.mlsys.org/paper_files/paper/2026/hash/0f8426558905746fc38da5e335700aec-Abstract-Conference.html | [
"Genghan Zhang",
"Shaowei Zhu",
"Anjiang Wei",
"Zhenyu Song",
"Allen Nie",
"Zhen Jia",
"Nandita Vijaykumar",
"Yida Wang",
"Kunle Olukotun"
] | We present AccelOpt, a self-improving large language model (LLM) agentic system that autonomously optimizes kernels for emerging AI acclerators, eliminating the need for expert-provided hardware-specific optimization knowledge.
AccelOpt explores the kernel optimization space through iterative generation, informed by an... | https://github.com/zhang677/AccelOpt | MLSys2026 |
BOOST: BOttleneck-Optimized Scalable Training Framework for Low-Rank Large Language Models | https://proceedings.mlsys.org/paper_files/paper/2026/hash/127e7093c38a45290524237be8eb39c5-Abstract-Conference.html | [
"Zhengyang Wang",
"Ziyue Liu",
"Ruijie Zhang",
"Avinash Maurya",
"Bogdan Nicolae",
"Paul Hovland",
"Franck Cappello",
"Zheng Zhang"
] | The scale of transformer model pre-training is constrained by the increasing computation and communication cost. Low-rank bottleneck architectures offer a promising solution to significantly reduce the training time and memory footprint with minimum impact on accuracy. Despite algorithmic efficiency, bottleneck archite... | # | MLSys2026 |
TiDAR: Think in Diffusion, Talk in Autoregression | https://proceedings.mlsys.org/paper_files/paper/2026/hash/1367d856028f65a9555b0274db09e608-Abstract-Conference.html | [
"Jingyu Liu",
"Xin Dong",
"Zhifan Ye",
"Rishabh Mehta",
"Yonggan Fu",
"Vartika Singh",
"Ce Zhang",
"Pavlo Molchanov"
] | Diffusion language models hold the promise of fast parallel generation, while autoregressive (AR) models typically excel in quality due to their causal structure aligning naturally with language modeling. This raises a fundamental question: can we achieve a synergy with high throughput, higher GPU utilization, and AR l... | # | MLSys2026 |
Rethinking DVFS for Mobile LLMs: Unified Energy-Aware Scheduling with CORE | https://proceedings.mlsys.org/paper_files/paper/2026/hash/136b9a13861308c8948cd308ccd02658-Abstract-Conference.html | [
"Zongpu Zhang",
"Pranab Dash",
"Qiang Xu",
"Y. Charlie Hu",
"Jian Li",
"Haibing Guan"
] | Despite the rapid adoption of large language models (LLMs) in mobile applications, deploying them efficiently on resource-constrained devices remains challenging due to limited compute, memory, and energy constraints. In this paper, we first evaluate the energy efficiency of state-of-the-art mobile LLM frameworks acros... | # | MLSys2026 |
ZK-APEX: ZERO-KNOWLEDGE APPROXIMATE PERSONALIZED UNLEARNING WITH EXECUTABLE PROOFS | https://proceedings.mlsys.org/paper_files/paper/2026/hash/148865706acbd18627d3fc15cc3f3b93-Abstract-Conference.html | [
"Mohammad M Maheri",
"Sunil Cotterill",
"Alex Davidson",
"Hamed Haddadi"
] | Machine unlearning removes the influence of specified data from trained models to satisfy privacy, copyright, and safety requirements (e.g., the “right to be forgotten”). In practice, providers distribute a global model to edge devices, that each locally personalize the model based on their private data.
However, sinc... | # | MLSys2026 |
Zero redundancy distributed learning with differential privacy | https://proceedings.mlsys.org/paper_files/paper/2026/hash/1e70ac91ad26ba5b24cf11b12a1f90fe-Abstract-Conference.html | [
"Zhiqi Bu",
"Justin Chiu",
"Ruixuan Liu",
"Sheng Zha",
"George Karypis"
] | Deep learning using large models has achieved great success in a wide range of domains. However, training these models on billions of parameters is very challenging in terms of training speed, memory cost, and communication efficiency, especially under the privacy-preserving regime with differential privacy (DP). On th... | https://github.com/awslabs/fast-differential-privacy | MLSys2026 |
CATWILD: Compiler Autotuning for TPU workloads in the Wild | https://proceedings.mlsys.org/paper_files/paper/2026/hash/2093ed77c549eda95bd6f7212b735b43-Abstract-Conference.html | [
"Ignacio Cano",
"Yu Wang",
"Mike Burrows",
"Ziqiang Feng",
"Matheus Camargo",
"Chao Wang",
"David Liu",
"Tengyu Sun",
"Alexander Wertheim",
"Arissa Wongpanich",
"Christof Angermueller",
"Hyojun Kim",
"Wenqi Cao",
"Aleksey Orekhov",
"Amit Sabne",
"Emma Sevastian",
"Mehrdad Khani",
"... | Compilers play a fundamental role at achieving peak performance for machine learning (ML) workloads. However, given the diverse nature of workloads and accelerators, compilers’ heuristics and analytical cost models can result in sub-optimal performance, and thus waste precious datacenter resources. Furthermore, the mul... | # | MLSys2026 |
GriNNder: Breaking the Memory Capacity Wall in Full-Graph GNN Training with Storage Offloading | https://proceedings.mlsys.org/paper_files/paper/2026/hash/215597a92341c78cc3e7fe628be4a195-Abstract-Conference.html | [
"Jaeyong Song",
"Seongyeon Park",
"Hongsun Jang",
"Jaewon Jung",
"Hunseong Lim",
"Junguk Hong",
"Jinho Lee"
] | Full-graph training of graph neural networks (GNNs) is widely used as it enables direct validation of algorithmic improvements by preserving complete neighborhood information.
However, it typically requires multiple GPUs or servers, incurring substantial hardware and inter-device communication costs.
While existing si... | # | MLSys2026 |
ExecuTorch - A Unified PyTorch Solution to Run ML Models On-Device | https://proceedings.mlsys.org/paper_files/paper/2026/hash/236f915dd02af4f11927f67330b21d4b-Abstract-Conference.html | [
"Mergen Nachin",
"Digant Desai",
"Stephen Jia",
"Chen Lai",
"Mengwei Liu",
"Jacob Szwejbka",
"Raziel Alvarez",
"RJ Ascani",
"Dave Bort",
"Manuel Candales",
"Andrew Caples",
"Yanan Cao",
"Zhengxu Chen",
"Soumith Chintala",
"Gregory Comer",
"Tanvir Islam",
"Songhao Jia",
"Tarun Karut... | Local execution of AI on edge devices is critical for privacy, low latency, and offline operation. However, deploying models on diverse hardware remains fragmented, often requiring model conversion or complete implementation outside the PyTorch ecosystem where the model was originally authored. We introduce ExecuTorch,... | # | MLSys2026 |
EarthSight: A Distributed Framework for Low-Latency Satellite Intelligence | https://proceedings.mlsys.org/paper_files/paper/2026/hash/26289c647c6828e862e271ca3c490486-Abstract-Conference.html | [
"Ansel Erol",
"Seungjun Lee",
"Divya Mahajan"
] | Low-latency delivery of satellite imagery is essential for time-critical applications such as disaster response, intelligence, and infrastructure monitoring. However, traditional pipelines rely on downlinking all captured images before analysis, introducing delays of hours to days due to restricted communication bandwi... | # | MLSys2026 |
XProf: An Open, Scalable, and Extensible Profiling System for the Modern ML Stack | https://proceedings.mlsys.org/paper_files/paper/2026/hash/280776fda5970060e6a5efcfd43e5d2d-Abstract-Conference.html | [
"Robert Hundt",
"Naveen Kumar",
"Jose Baiocchi Paredes",
"Scott Goodson",
"Clive Verghese",
"Prasanna Rengasamy",
"Kelvin Le",
"Jiya Zhang",
"Charles Alaras",
"Yin Zhang",
"Kan Cai",
"Jiten Thakkar",
"Sai Ganesh Bandiatmakuri",
"Yogesh SY",
"Aniruddha N. Udipi",
"Vikas Agarwal"
] | Optimizing Large Models across thousands of accelerators requires deep system expertise. To address modern machine learning (ML) optimization needs, we presentXProf, the ML profiler for the OpenXLA ecosystem.XProfdelivers actionable optimization suggestions and in-depth performance analysis, empowering ML researchers a... | https://github.com/openxla/xprof | MLSys2026 |
Breaking the Ice: Analyzing Cold Start Latency in vLLM | https://proceedings.mlsys.org/paper_files/paper/2026/hash/29416b66c2149872b9d1415a3fd2c5e0-Abstract-Conference.html | [
"Huzaifa Shaaban Kabakibo",
"Animesh Trivedi",
"Lin Wang"
] | As scalable inference services become popular, the cold start latency of an inference engine becomes important. Today, vLLM has evolved into the de-facto inference engine of choice for many inference workloads.
Although popular, due to its complexity and rapid evolution, there has not been a systematic study on the sta... | https://github.com/upb-cn/vllm-startup-profiler | MLSys2026 |
Attribution-based Sparse Activation in Large Language Models | https://proceedings.mlsys.org/paper_files/paper/2026/hash/29591f355702c3f4436991335784b503-Abstract-Conference.html | [
"Jifeng Song",
"Xiangyu Yin",
"Boyuan Yang",
"Kai Huang",
"Weichen Liu",
"Wei Gao"
] | LLM inference is computationally expensive due to the LLM's large parameter sizes. Existing techniques reduce the computing cost via model retraining, but cannot well adapt to different downstream tasks or variant input data at runtime. To avoid such retraining efforts for runtime adaptability, a better option is spars... | # | MLSys2026 |
Once-for-All Channel Mixers (HyperTinyPW): Generative Compression for TinyML | https://proceedings.mlsys.org/paper_files/paper/2026/hash/2d04d97593c8c33d415337f408ed0e1b-Abstract-Conference.html | [
"Yassien Shaalan"
] | Neural networks on microcontrollers are constrained
by kilobytes of flash/SRAM, where 1×1
pointwise (PW) mixers often dominate memory
even after INT8 quantization. We present
HYPERTINYPW, a compression-as-generation
method that replaces most stored PW weights
with generated weights: a shared micro-MLP
synthesizes PW ke... | # | MLSys2026 |
Ontology-Guided Long-Term Agent Memory for Conversational RAG | https://proceedings.mlsys.org/paper_files/paper/2026/hash/2fb4be70fc9668e9ec2c71b34fb127d4-Abstract-Conference.html | [
"Shuang Cao",
"Rui Li"
] | Retrieval-augmented generation (RAG) enables LLMs to ground responses in external knowledge, but long-term,
multi-session conversations still suffer from implicit recall failures: when current user queries lack lexical overlap
with earlier facts (e.g., preferences), standard dense retrieval and long-context prompting o... | # | MLSys2026 |
GUARD: SCALABLE STRAGGLER DETECTION AND NODE HEALTH MANAGEMENT FOR LARGE-SCALE TRAINING | https://proceedings.mlsys.org/paper_files/paper/2026/hash/339caf45a6fa281cae8adc6465343464-Abstract-Conference.html | [
"Guanliang Liu",
"Abhinandan Patni",
"congzhu lin",
"Zoe zeng",
"Jack Wittmayer",
"yinghong liu",
"josh wu",
"Anthony Ko",
"Alexander Zhipa",
"Ashvin Nihalani",
"Binxuan Huang",
"Cong cheng",
"Mi Sun",
"Vijay Rajakumar",
"Rejith George Joseph",
"Parthasarathy Govindarajen"
] | Training frontier-scale foundation models involves coordinating tens of thousands of GPUs over multi-month
runs, where even minor performance degradations can accumulate into substantial efficiency losses. Existing
health-check mechanisms, such as NCCL tests or GPU burn-in, primarily focus on functional correctness and... | # | MLSys2026 |
FlashInfer-Bench: Building the Virtuous Cycle for AI-driven LLM Systems | https://proceedings.mlsys.org/paper_files/paper/2026/hash/37e44c4b5321605735be9761f9b758fc-Abstract-Conference.html | [
"Shanli Xing",
"Yiyan Zhai",
"Alexander Jiang",
"Yixin Dong",
"Yong Wu",
"Zihao Ye",
"Charlie F. Ruan",
"Yingyi Huang",
"Yineng Zhang",
"Liangsheng Yin",
"Aksara Bayyapu",
"Luis Ceze",
"Tianqi Chen"
] | Recent advances show that large language models (LLMs) can act as autonomous agents capable of generating GPU kernels, but integrating these AI-generated kernels into real-world inference systems remains challenging. FlashInfer-Bench addresses this gap by establishing a standardized, closed-loop framework that connects... | # | MLSys2026 |
CRAFT: Fine-Grained Cost-Aware Expert Replication For Efficient Mixture-of-Experts Serving | https://proceedings.mlsys.org/paper_files/paper/2026/hash/3a7f9e485845dac27423375c934cb4db-Abstract-Conference.html | [
"Adrian Zhao",
"Zhenkun Cai",
"Zhenyu Song",
"Lingfan Yu",
"Haozheng Fan",
"Jun Wu",
"Yida Wang",
"Nandita Vijaykumar"
] | Mixture-of-Experts (MoE) has recently emerged as the mainstream architecture for efficiently scaling large language models while maintaining near-constant computational cost. Expert parallelism distributes parameters by partitioning experts across devices, but this introduces token-level load imbalance during inference... | # | MLSys2026 |
Stream2LLM: Overlap Context Streaming and Prefill for Reduced Time-to-First-Token | https://proceedings.mlsys.org/paper_files/paper/2026/hash/3b3889d313ba9476c12c2d77ea66b24f-Abstract-Conference.html | [
"Rajveer Bachkaniwala",
"Chengqi Luo",
"Richard So",
"Divya Mahajan",
"Kexin Rong"
] | Context retrieval systems for LLM inference face a critical challenge: high retrieval latency creates a fundamental tension between waiting for complete context (poor time-to-first-token) and proceeding without it (reduced quality). Streaming context incrementally--overlapping retrieval with inference--can mitigate thi... | https://github.com/rajveerb/stream2llm/tree/mlsys_artifact | MLSys2026 |
FlexTrain: Scalable Hybrid-Parallel Training with Elastic Resource Utilization and Consistent Accuracy | https://proceedings.mlsys.org/paper_files/paper/2026/hash/3c40417b8dca30c08cc361df5b33ad7e-Abstract-Conference.html | [
"Weilin Cai",
"Diandian Gu",
"Baoquan Zhong",
"Jun Wang",
"Zhuolin Zheng",
"Gaohong Liu",
"Jiang Kaihua",
"Shuguang Wang",
"Wencong Xiao",
"Jiayi Huang"
] | Large language model (LLM) training has become a critical workload in shared GPU clusters. However, our observations reveal that these clusters suffer from significant underutilization. To address this inefficiency, various elastic training techniques have been developed to dynamically adjust GPU allocations to harness... | # | MLSys2026 |
SAKURAONE: An Open Ethernet–Based AI HPC System and Its Observed Workload Dynamics in a Single-Tenant LLM Development Environment | https://proceedings.mlsys.org/paper_files/paper/2026/hash/40b8fb4f90004405e14b1ede6ab42373-Abstract-Conference.html | [
"Fumikazu KONISHI",
"Yuuki Tsubouchi",
"Hirofumi Tsuruta"
] | SAKURAONE is a managed high performance computing (HPC) cluster developed and operated by the SAKURA Internet Research Center. It builds on the KOKARYOKU PHY bare metal GPU platform and is optimized for advanced workloads, including large language model (LLM) training. In ISC 2025 TOP500, SAKURAONE is ranked 49th by HP... | # | MLSys2026 |
PRISM: Parametrically Refactor Inference for Speculative Decoding Draft Models | https://proceedings.mlsys.org/paper_files/paper/2026/hash/414fd191b3246a19a55741b938380136-Abstract-Conference.html | [
"Xuliang Wang",
"Yuetao Chen",
"Maochan Zhen",
"Fang Liu",
"Xinzhou Zheng",
"Xingwu Liu",
"Hong Xu",
"Ming Li"
] | Large Language Models (LLMs), constrained by their auto-regressive nature, suffer from slow decoding. Speculative decoding methods have emerged as a promising solution to accelerate LLM decoding, attracting attention from both systems and AI research communities. Recently, the pursuit of better draft quality has driven... | # | MLSys2026 |
Efficient Long-Context Language Model Training by Core Attention Disaggregation | https://proceedings.mlsys.org/paper_files/paper/2026/hash/423b59ae02381f27862c21d1c41a5603-Abstract-Conference.html | [
"Yonghao Zhuang",
"Junda Chen",
"Bo Pang",
"Yi Gu",
"Yibo Zhu",
"Yimin Jiang",
"Ion Stoica",
"Hao Zhang",
"Eric P. Xing"
] | We present core attention disaggregation (CAD), a technique that improves long-context LLM training by disaggregating the core attention (CA) -- the parameter-free $\mathrm{softmax}(\mathbf{QK}^{\top})\mathbf{V}$ computation -- and schedules it on an independent pool of resources. Existing systems co-locate core attent... | # | MLSys2026 |
Demystifying the Mixture of Experts Serving Tax | https://proceedings.mlsys.org/paper_files/paper/2026/hash/42a452cbafa9dd64e9ba4aa95cc1ef21-Abstract-Conference.html | [
"Pratyush Patel",
"Dayeol Lee",
"Shintaro Iwasaki",
"Arvind Krishnamurthy"
] | Mixture-of-Experts (MoEs) enable massive model sizes but incur higher serving overheads than dense models at the same per-token compute cost. This MoE tax varies with the model architecture, inference phase, and parallelism strategy. We comprehensively study the tax for different MoE models, finding that they perform 2... | # | MLSys2026 |
Sparing Strategies to Minimize Reliability Impact On Large Training Jobs | https://proceedings.mlsys.org/paper_files/paper/2026/hash/437bc4ccafd3fc6d4289bd10940be42b-Abstract-Conference.html | [
"Kevin Quirk",
"Matthew Lennie",
"Ehsan K. Ardestani",
"Satyajeet Singh Ahuja",
"Matthew Bergeron",
"Andrew Grier",
"Zhaodong Wang",
"Mustafa Ozdal",
"Xu Zhang",
"Abhinav Triguna",
"Ying Zhang",
"Mathew Oldham",
"Chunqiang Tang"
] | Training large language models (LLMs) on Meta’s AI clusters requires running long, distributed jobs that are
vulnerable to hardware failures. To maintain high availability and efficiency, production systems use sparing, i.e.,
pre-allocating spare compute resources that can replace failed components. However, choosing t... | # | MLSys2026 |
SkipKV: Selective Skipping of KV Generation and Storage for Efficient Inference with Large Reasoning Models | https://proceedings.mlsys.org/paper_files/paper/2026/hash/45c1f6a8cbf2da59ebf2c802b4f742cd-Abstract-Conference.html | [
"Jiayi Tian",
"Seyedarmin Azizi",
"Yequan Zhao",
"Erfan Baghaei Potraghloo",
"Sean McPherson",
"Sharath Nittur Sridhar",
"Zhengyang Wang",
"Zheng Zhang",
"Massoud Pedram",
"Souvik Kundu"
] | Large reasoning models (LRMs) often incur significant key-value (KV) cache overhead, due to their linear growth with the verbose chain-of-thought (CoT) reasoning. This incurs both memory overhead and throughput bottlenecks, limiting efficient deployment. To reduce KV cache size during inference, we first investigate th... | https://github.com/TTTTTTris/SkipKV | MLSys2026 |
PROMPTS: PeRformance Optimization via Multi-Agent Planning for LLM Training and Serving | https://proceedings.mlsys.org/paper_files/paper/2026/hash/48253da5351effdfea994fd7bbff7005-Abstract-Conference.html | [
"Yuran Ding",
"Ruobing Han",
"Xiaofan Zhang",
"Xinwei Chen"
] | Optimizing large-language model (LLM) training and serving on large-scale distributed systems is a significant challenge. This difficulty stems from the rapidly evolving LLM landscape, the requirement for deep domain expertise, and the need for workload-specific optimization strategies. Existing methods rely on either ... | # | MLSys2026 |
Wave: A Symbolic Python DSL And Compiler for High-Performance Machine Learning | https://proceedings.mlsys.org/paper_files/paper/2026/hash/48c34730ff9a8574481a00ce8cb5e2cb-Abstract-Conference.html | [
"Harsh Menon",
"Oleksandr Zinenko",
"Gaurav Verma",
"Stanley Winata",
"Ivan Butygin",
"Nithin Meganathan",
"Sanket Pandit",
"William Hatch",
"Surya Jasper",
"Megan Kuo",
"Sahil Faizal",
"Ashay Rane",
"Aurore De Spirlet",
"Martin Paul Lücke"
] | Modern ML models demand ever-greater compute, prompting hardware vendors to add specialized matrix cores to their GPUs. While these units unlock high throughput, they impose intricate programming models and addressing schemes that are difficult to manage by hand. This paper introduces Wave, a Python-embedded DSL for ke... | # | MLSys2026 |
A Lightweight High-Throughput Collective-Capable NoC for Large-Scale ML Accelerators | https://proceedings.mlsys.org/paper_files/paper/2026/hash/48fecef47b19fe501d27d338b6d52582-Abstract-Conference.html | [
"Luca Colagrande",
"Lorenzo Leone",
"Chen Wu",
"Tim Fischer",
"Raphael Roth",
"Luca Benini"
] | The exponential increase in Machine Learning (ML) model size and complexity has driven unprecedented demand for high-performance acceleration systems. As technology scaling enables the integration of thousands of computing elements onto a single die, the boundary between distributed and on-chip systems has blurred, mak... | # | MLSys2026 |
Meeting SLOs, Slashing Hours: Automated Enterprise LLM Optimization with OptiKIT | https://proceedings.mlsys.org/paper_files/paper/2026/hash/4904fad153f6434a7bcf04465d4be2cc-Abstract-Conference.html | [
"Nicholas Santavas",
"Kareem Eissa",
"Patrycja Cieplicka",
"Piotr Florek",
"Matteo Nulli",
"Stefan Vasilev",
"Seyyed Hadi Hashemi",
"Antonios Gasteratos",
"Shahram Khadivi"
] | Enterprise LLM deployment faces a critical scalability challenge: organizations must optimize models systematically to scale AI initiatives within constrained compute budgets, yet the specialized expertise required for manual optimization remains a niche and scarce skillset. This challenge is particularly evident in ma... | # | MLSys2026 |
SONAR: Benchmarking Topology and Collaboration in Decentralized Learning | https://proceedings.mlsys.org/paper_files/paper/2026/hash/4abf39685e89559523a4d644036fd2b3-Abstract-Conference.html | [
"Joyce Yuan",
"Yichuan Shi",
"Abhishek Singh",
"Rishi Sharma",
"Ramesh Raskar",
"Jonas Blanc",
"Martin Jaggi"
] | Decentralized machine learning relies on peer-to-peer communication, yet the role of network topology in shaping learning dynamics remains poorly understood due to the lack of controlled, reproducible evaluation frameworks. We present \textbf{SONAR}, a modular framework for topology-aware decentralized learning that un... | # | MLSys2026 |
Virtual Machine NUMA Placement at Scale: Learning the Norm, Shielding the Tail | https://proceedings.mlsys.org/paper_files/paper/2026/hash/4e3157021c5f833bb2204081f1dda573-Abstract-Conference.html | [
"Yibo Zhao",
"Tianyuan Wu",
"Hui Xue",
"Qi Chen",
"Zhenhua Han",
"Zikai Xu",
"Yuntai Chang",
"Rui Gao",
"Steve Deng",
"Ray Jui-Hao Chiang",
"Mingxia Li",
"Yuqing Yang",
"Cheng Tan",
"Fan Yang",
"Peng Cheng",
"Yongqiang Xiong",
"Lili Qiu",
"Lidong Zhou"
] | In modern data centers, servers organize memory and CPUs into Non-Uniform Memory Access (NUMA) nodes,
where unequal memory-to-CPU proximity leads to varying memory latency. Hypervisors must carefully place
Virtual Machines (VMs) to reduce remote memory access. Poor placements can lead to significant performance
degrada... | # | MLSys2026 |
Automated Algorithm Design for Auto-Tuning Optimizers | https://proceedings.mlsys.org/paper_files/paper/2026/hash/4f31327e046913c7238d5b671f5d820e-Abstract-Conference.html | [
"Floris-Jan Willemsen",
"Niki van Stein",
"Ben van Werkhoven"
] | Automatic performance tuning (auto-tuning) is essential for optimizing high-performance applications, where vast and irregular search spaces make manual exploration infeasible.
While auto-tuners traditionally rely on classical approaches such as evolutionary, annealing, or surrogate-based optimizers, designing algorith... | # | MLSys2026 |
RaidServe: High-performance Resilient Serving | https://proceedings.mlsys.org/paper_files/paper/2026/hash/507b4aacefe5325908e24f042617b741-Abstract-Conference.html | [
"Ziyi Xu",
"Zhiqiang Xie",
"Swapnil Gandhi",
"Christos Kozyrakis"
] | Tensor parallelism (TP) enables large language models (LLMs) to scale inference efficiently across multiple GPUs, but its tight coupling makes systems fragile: a single GPU failure can halt execution, trigger costly KVCache recomputation, and introduce long-term compute and memory imbalance. We present RaidServe , a fa... | # | MLSys2026 |
Toward Principled LLM Safety Testing: Solving the Jailbreak Oracle Problem | https://proceedings.mlsys.org/paper_files/paper/2026/hash/50a2e625745ab078389ccd23747fc0d8-Abstract-Conference.html | [
"Shuyi Lin",
"Anshuman Suri",
"Alina Oprea",
"Cheng Tan"
] | As large language models (LLMs) become increasingly deployed in safety-critical applications, the lack of systematic methods to assess their vulnerability to jailbreak attacks presents a critical security gap. We introduce the \emph{jailbreak oracle problem}: given a model, prompt, and decoding strategy, determine whet... | https://github.com/shuyilinn/BOA/tree/mlsys2026ae | MLSys2026 |
HetRL: Efficient Reinforcement Learning for LLMs in Heterogeneous Environments | https://proceedings.mlsys.org/paper_files/paper/2026/hash/5321b1dabcd2be188d796c21b733e8c7-Abstract-Conference.html | [
"Yongjun He",
"Shuai Zhang",
"Jiading Gai",
"Xiyuan Zhang",
"Boran Han",
"Bernie Wang",
"Huzefa Rangwala",
"George Karypis"
] | As large language models (LLMs) continue to scale and new GPUs are released even more frequently, there is an increasing demand for LLM post-training in heterogeneous environments to fully leverage underutilized mid-range or previous-generation GPUs and alleviate the shortage of homogeneous high-end GPUs within a singl... | # | MLSys2026 |
Event Tensor: A Unified Abstraction for Compiling Dynamic Megakernel | https://proceedings.mlsys.org/paper_files/paper/2026/hash/53d3f45797970d323bd8a0d379c525aa-Abstract-Conference.html | [
"Hongyi Jin",
"Bohan Hou",
"Guanjie Wang",
"Ruihang Lai",
"Jinqi Chen",
"Zihao Ye",
"Yaxing Cai",
"Yixin Dong",
"Xinhao Cheng",
"Zhihao Zhang",
"Yilong Zhao",
"Yingyi Huang",
"Lijie Yang",
"Jinchen Jiang",
"Gabriele Oliaro",
"Jianan Ji",
"Xupeng Miao",
"Vinod Grover",
"Todd C. Mo... | Modern GPU workloads, especially large language model (LLM) inference, suffer from kernel launch overheads and coarse synchronization that limit inter-kernel parallelism. Recent megakernel techniques fuse multiple operators into a single persistent kernel to eliminate launch gaps and expose inter-kernel parallelism, bu... | # | MLSys2026 |
MLCommons Chakra: Advancing Performance Benchmarking and Co-design using Standardized Execution Traces | https://proceedings.mlsys.org/paper_files/paper/2026/hash/53fe824f289060ce705ed7c01dae59d2-Abstract-Conference.html | [
"Srinivas Sridharan",
"Andy Balogh",
"Bradford M. Beckmann",
"Brian Coutinho",
"Louis Feng",
"Sheng Fu",
"Sanshan Gao",
"Mehryar Garakani",
"Taekyung Heo",
"David Kanter",
"Josh Ladd",
"Ziwei Li",
"Winston Liu",
"Changhai Man",
"Dan Mihailescu",
"Spandan More",
"Joongun Park",
"Ash... | The fast pace of artificial intelligence (AI) innovation demands an agile methodology for observation, reproduction and optimization of distributed machine learning (ML) workload behavior in production AI systems and enables efficient software-hardware (SW-HW) co-design for future systems. We present Chakra, an open an... | # | MLSys2026 |
AIRS: Scaling Live Inference in Resource Constrained Environments | https://proceedings.mlsys.org/paper_files/paper/2026/hash/547108084f0c2af39b956f8eadb75d1b-Abstract-Conference.html | [
"Nilesh Jagnik",
"Xiaohao Yang",
"Tuan Do",
"Chelsea Chen",
"Harshvardhan GM"
] | Advancements in large language models (LLMs) have made them increasingly useful for complex reasoning tasks which previously required domain experts. One such task is quality evaluation of query responses produced by a search engine. Evaluation generates metrics necessary to study the quality, impact, and usefulness of... | # | MLSys2026 |
Speculative Decoding: Performance or Illusion? | https://proceedings.mlsys.org/paper_files/paper/2026/hash/554e056fe2b6d9fd27ffcd3367ae1267-Abstract-Conference.html | [
"Xiaoxuan Liu",
"Jiaxiang Yu",
"Jongseok Park",
"Ion Stoica",
"Alvin Cheung"
] | Speculative decoding (SD) has become a popular technique to accelerate Large Language Model (LLM) inference, yet its real-world effectiveness remains unclear as prior evaluations rely on research prototypes and unrealistically small batch sizes. We present, to our knowledge, the first systematic study of SD on a produc... | # | MLSys2026 |
ProTrain: Efficient LLM Training via Automatic Memory Management | https://proceedings.mlsys.org/paper_files/paper/2026/hash/56280ad5fb53967fd55e4ba1b1cfc418-Abstract-Conference.html | [
"Hanmei Yang",
"Jin Zhou",
"Yao Fu",
"Xiaoqun Wang",
"Ramine Roane",
"Hui Guan",
"Tongping Liu"
] | Memory pressure has emerged as a dominant constraint in scaling the training of large language models (LLMs), particularly in resource-constrained environments. While modern frameworks incorporate various memory-saving techniques, they often expose low-level configuration knobs that require manual tuning and specialize... | # | MLSys2026 |
BatchLLM: Optimizing Large Batched LLM Inference with Global Prefix Sharing and Throughput-oriented Token Batching | https://proceedings.mlsys.org/paper_files/paper/2026/hash/5b7ae1758452854dee4e962207d38304-Abstract-Conference.html | [
"Zhen Zheng",
"Xin Ji",
"Taosong Fang",
"Fanghao Zhou",
"Chuanjie Liu",
"Gang Peng"
] | Large language models (LLMs) increasingly play an important role in a wide range of information processing and management tasks in industry.
Many of these tasks are performed in large batches or even offline, and the performance indicator for which is throughput.
These tasks usually show the characteristic of prefix sh... | https://github.com/microsoft/MixLLM/tree/batchllm_vllm_064 | MLSys2026 |
DisAgg: Distributed Aggregators for Efficient Secure Aggregation | https://proceedings.mlsys.org/paper_files/paper/2026/hash/5c40a52354e95a6fc701a84cdcd97bc8-Abstract-Conference.html | [
"Haaris Mehmood",
"Giorgos Tatsis",
"Dimitrios Alexopoulos",
"Karthikeyan Saravanan",
"Jie Xu",
"Anastasios Drosou",
"Mete Ozay"
] | Federated learning enables collaborative model training across distributed clients, yet vanilla FL exposes client updates to the central server. Secure‑aggregation schemes protect privacy against an honest‑but‑curious server, but existing approaches often suffer from many communication rounds, heavy public‑key operati... | # | MLSys2026 |
OSWorld-Human: Benchmarking the Efficiency of Computer-Use Agents | https://proceedings.mlsys.org/paper_files/paper/2026/hash/5edb57c05c81d04beb716ef1d542fe9e-Abstract-Conference.html | [
"Reyna Abhyankar",
"Qi Qi",
"Yiying Zhang"
] | Generative AI is being leveraged to solve a variety of computer-use tasks involving desktop applications. State-of-
the-art systems have focused solely on improving accuracy on leading benchmarks. However, these systems are
practically unusable due to extremely high end-to-end latency (e.g. tens of minutes) for tasks t... | # | MLSys2026 |
SwiftGS: Algorithm and System Co-Optimization for Fast 3D Gaussian Splatting on GPUs | https://proceedings.mlsys.org/paper_files/paper/2026/hash/5f96a21345c138da929e99871fda138e-Abstract-Conference.html | [
"Lingjun Gao",
"Zhican Wang",
"Zhiwen Mo",
"Hongxiang Fan"
] | Recent advances in 3D Gaussian Splatting (3DGS) have enabled high-quality and efficient novel view synthesis, demonstrating great potential in real-world applications such as robotic perception and digital-twin construction.
However, 3DGS requires processing up to millions of Gaussians in parallel, imposing significan... | # | MLSys2026 |
PARROT: Persuasion and Agreement Robustness Rating of Output Truth — A Sycophancy Robustness Benchmark for LLMs | https://proceedings.mlsys.org/paper_files/paper/2026/hash/621d0fd41c720ab252e178b77c200d90-Abstract-Conference.html | [
"Yusuf Çelebi",
"Özay Ezerceli",
"Mahmoud El Hussieni"
] | This study presents PARROT (Persuasion and Agreement Robustness Rating of Output Truth), a robustness-focused framework designed to measure the degradation in accuracy that occurs under social pressure exerted on users through authority and persuasion in large language models (LLMs) the phenomenon of sycophancy (excess... | # | MLSys2026 |
Search Your Block Floating Point Scales! | https://proceedings.mlsys.org/paper_files/paper/2026/hash/633b0e871a48d542280c3ad03928e60d-Abstract-Conference.html | [
"Tanmaey Gupta",
"Hayden Prairie",
"Xiaoxia wu",
"Reyna Abhyankar",
"Qingyang Wu",
"Austin Silveria",
"Pragaash Ponnusamy",
"Jue Wang",
"Ben Athiwaratkun",
"Leon Song",
"Tri Dao",
"Daniel Y. Fu",
"Chris De Sa"
] | Quantization has emerged as a standard technique for accelerating inference for generative models by enabling faster low-precision computations and reduced memory transfers. Recently, GPU accelerators have added first-class support for microscaling Block Floating Point (BFP) formats. Standard BFP algorithms use a fixed... | # | MLSys2026 |
Accelerating Large-Scale Reasoning Model Inference with Sparse Self-Speculative Decoding | https://proceedings.mlsys.org/paper_files/paper/2026/hash/66a026c0d17040889b50f0dfa650e5e0-Abstract-Conference.html | [
"Yilong Zhao",
"Jiaming Tang",
"Kan Zhu",
"Zihao Ye",
"Chi-Chih Chang",
"Chaofan Lin",
"Jongseok Park",
"Guangxuan Xiao",
"Mohamed S. Abdelfattah",
"Mingyu Gao",
"Baris Kasikci",
"Song Han",
"Ion Stoica"
] | Reasoning language models have demonstrated remarkable capabilities on challenging tasks by generating elaborate chain-of-thought (CoT) solutions. However, such lengthy generation shifts the inference bottleneck from compute-bound to memory-bound. To generate each token, the model applies full attention to all previous... | # | MLSys2026 |
StreamDiffusionV2: A Streaming System for Dynamic and Interactive Video Generation | https://proceedings.mlsys.org/paper_files/paper/2026/hash/698cfaf72a208aef2e78bcac55b74328-Abstract-Conference.html | [
"Tianrui Feng",
"Zhi Li",
"Shuo Yang",
"Haocheng Xi",
"Muyang Li",
"Xiuyu Li",
"Lvmin Zhang",
"Keting Yang",
"Kelly Peng",
"Song Han",
"Maneesh Agrawala",
"Kurt Keutzer",
"Akio Kodaira",
"Chenfeng Xu"
] | Generative models are reshaping the live-streaming industry by redefining how content is created, styled, and delivered. Previous image-based streaming diffusion models have powered efficient and creative live streaming products but has hit limits on temporal consistency due to the foundation of image-based designs.
Re... | # | MLSys2026 |
Pylo: Towards Accessible Learned Optimizers in PyTorch | https://proceedings.mlsys.org/paper_files/paper/2026/hash/6b1d4c03391b0aa6ddde0b807a78c950-Abstract-Conference.html | [
"Paul Janson",
"Benjamin Thérien",
"Quentin Anthony",
"Xiaolong Huang",
"Abhinav Moudgil",
"Eugene Belilovsky"
] | Learned optimizers have been an active research topic over the past decade, with increasing progress toward practical, general-purpose optimizers that can serve as drop-in replacements for widely used methods like Adam. However, recent advances such as VeLO, which was meta-trained for 4000 TPU-months, remain largely in... | https://github.com/Belilovsky-Lab/pylo | MLSys2026 |
FaaScale: Unlocking Fast LLM Scaling for Serverless Inference | https://proceedings.mlsys.org/paper_files/paper/2026/hash/6e32c247076c2c0fb381e022c02d2c78-Abstract-Conference.html | [
"Minchen Yu",
"Rui Yang",
"Chaobo Jia",
"Zhaoyuan Su",
"Sheng Yao",
"Tingfeng Lan",
"Yuchen Yang",
"Zirui Wang",
"Yue Cheng",
"Wei Wang",
"Ao Wang",
"Ruichuan Chen"
] | Serverless computing is an attractive paradigm for cloud-based large language model (LLM) inference, but scaling LLMs on demand remains a major challenge due to high data transfer cost. We present FaaScale, a serverless LLM system that enables fast and resource-efficient model scaling. The key idea is a co-design princ... | # | MLSys2026 |
FarSkip-Collective: Unhobbling Blocking Communication in Mixture of Experts Models | https://proceedings.mlsys.org/paper_files/paper/2026/hash/6feb9b30798abcfae937760d183605e1-Abstract-Conference.html | [
"Yonatan Dukler",
"Guihong Li",
"Deval Shah",
"Jiang Liu",
"Vikram Appia",
"Emad Barsoum"
] | Blocking communication presents a major hurdle in running MoEs efficiently in distributed settings. To address this, we present FarSkip-Collective which modifies the architecture of modern models to enable overlapping of their computation with communication. Our approach modifies the architecture to skip connections in... | # | MLSys2026 |
REPARO: LOSS-RESILIENT GENERATIVE CODEC FOR VIDEO CONFERENCING | https://proceedings.mlsys.org/paper_files/paper/2026/hash/703f727ec10190b2fddcf8e24f52df48-Abstract-Conference.html | [
"Tianhong Li",
"Vibhaalakshmi Sivaraman",
"Pantea Karimi",
"Lijie Fan",
"Mohammad Alizadeh",
"Dina Katabi"
] | Packet loss during video conferencing often results in poor quality and video freezing. Retransmitting lost packets is often impractical due to the need for real-time playback, and using Forward Error Correction (FEC) for packet recovery is challenging due to the unpredictable and bursty nature of Internet losses. Exce... | # | MLSys2026 |
OPKV: A High-Throughput Plugin-Driven Framework for Recallable Sparsity in Paged KV Cache Systems | https://proceedings.mlsys.org/paper_files/paper/2026/hash/71381211d0abef73ed1887b83c4547b1-Abstract-Conference.html | [
"Huazheng Lao",
"Xiaofeng Li",
"Rui Xu",
"Long Chen",
"Xia Zhu",
"Jinquan Zhang"
] | Long-context large language model (LLM) inference faces severe KV cache inflation, making GPU memory a key bottleneck. Existing recallable sparsity methods mitigate memory pressure by offloading non-critical key–value (KV) pairs to CPU memory and recalling them on demand, they are intrusive to KV cache management in th... | # | MLSys2026 |
MAC-Attention: a Match--Amend--Complete scheme for fast and accurate attention computation | https://proceedings.mlsys.org/paper_files/paper/2026/hash/7398289396de403d7d0505ed791e704a-Abstract-Conference.html | [
"Jinghan Yao",
"Sam Ade Jacobs",
"Walid Krichene",
"Masahiro Tanaka",
"Dhabaleswar Panda"
] | Long-context decoding in LLMs is IO-bound: each token re-reads an ever-growing KV cache. Prior accelerations cut bytes via compression (lowering fidelity) or selection/eviction (restricting what remains accessible), which can degrade delayed recall and long-form generation. We introduce MAC-Attention, a fidelity and ac... | # | MLSys2026 |
TokenWeave: Efficient Compute-Communication Overlap for Distributed LLM Inference | https://proceedings.mlsys.org/paper_files/paper/2026/hash/73ba81c7b25134a559c8a9c39ec1a4c3-Abstract-Conference.html | [
"Raja Gond",
"Nipun Kwatra",
"Ramachandran Ramjee"
] | Distributed inference of large language models (LLMs) using tensor parallelism can introduce communication overheads of 20% even over GPUs connected via NVLink, a high-speed GPU interconnect. Several techniques have been proposed to mitigate these overheads by decomposing computations into smaller tasks and overlapping... | https://github.com/microsoft/tokenweave | MLSys2026 |
When Enough is Enough: Rank-Aware Early Termination for Vector Search | https://proceedings.mlsys.org/paper_files/paper/2026/hash/78834433edc3291f4c6cbbd2759324db-Abstract-Conference.html | [
"Jianan Lu",
"Asaf Cidon",
"Michael Freedman"
] | Graph-based vector search underpins modern LLM applications such as retrieval-augmented generation (RAG), but its efficiency is increasingly constrained by disk I/O.
Existing systems continue searching long after discovering the higher-ranked (i.e., most valuable) results for downstream applications.
We present Terminu... | # | MLSys2026 |
Spira: Exploiting Voxel Data Structural Properties for Efficient Sparse Convolution in Point Cloud Networks | https://proceedings.mlsys.org/paper_files/paper/2026/hash/7972f3735e104a54715922aa416fde1b-Abstract-Conference.html | [
"Dionysios Adamopoulos",
"Anastasia Poulopoulou",
"Georgios Goumas",
"Christina Giannoula"
] | Sparse Convolution (SpC) powers 3D point cloud networks widely used in autonomous driving and augmented/virtual reality. SpC builds a kernel map that stores mappings between input voxel coordinates, output coordinates,
and weight offsets, then uses this map to compute feature vectors for output coordinates. Our work id... | https://github.com/SPIN-Research-Group/Spira | MLSys2026 |
Efficient, VRAM-Constrained xLM Inference on Clients | https://proceedings.mlsys.org/paper_files/paper/2026/hash/7cd265ae802235b8d5778a4a96ff22dd-Abstract-Conference.html | [
"Aditya Ukarande",
"Deep Shekhar",
"Marc Blackstein",
"Ram Rangan"
] | To usher in the next round of client AI innovation, there is an urgent need to enable efficient, lossless inference of high-accuracy large language models (LLMs) and vision language models (VLMs), jointly referred to as xLMs, on client systems. This means efficient support for: a) interactive as well as batch modes, b)... | # | MLSys2026 |
MTraining: Distributed Dynamic Sparse Attention for Efficient Ultra-Long Context Training | https://proceedings.mlsys.org/paper_files/paper/2026/hash/7fafdf453029d7b8674b6f3dd18112bf-Abstract-Conference.html | [
"Wenxuan Li",
"Chengruidong Zhang",
"Huiqiang Jiang",
"Yucheng Li",
"Yuqing Yang",
"Lili Qiu"
] | The adoption of long context windows has become a standard feature in Large Language Models (LLMs), as extended contexts significantly enhance their capacity for complex reasoning and broaden their applicability across diverse scenarios. Dynamic sparse attention is a promising approach for reducing the computational co... | https://github.com/microsoft/MInference/tree/main/mtraining | MLSys2026 |
TeleRAG: Efficient Retrieval-Augmented Generation Inference with Lookahead Retrieval | https://proceedings.mlsys.org/paper_files/paper/2026/hash/7fd522b89ac21009b7bbe7560a9a5add-Abstract-Conference.html | [
"Chien-Yu Lin",
"Keisuke Kamahori",
"Yiyu Liu",
"Xiaoxiang Shi",
"Madhav Kashyap",
"Yile Gu",
"Rulin Shao",
"Zihao Ye",
"Kan Zhu",
"Rohan Kadekodi",
"Stephanie Wang",
"Arvind Krishnamurthy",
"Luis Ceze",
"Baris Kasikci"
] | Retrieval-augmented generation (RAG) extends large language models (LLMs) with external data sources to enhance factual correctness and domain coverage.
Modern RAG pipelines rely on large datastores, creating a significant system challenge: achieving high throughput and low latency is difficult, especially when GPU mem... | # | MLSys2026 |
MorphServe: Efficient and Workload-Aware LLM Serving via Runtime Quantized Layer Swapping and KV Cache Resizing | https://proceedings.mlsys.org/paper_files/paper/2026/hash/8144a9d62e506af0fcdeac0e456b2710-Abstract-Conference.html | [
"Zhaoyuan Su",
"Zeyu Zhang",
"Tingfeng Lan",
"Zirui Wang",
"Haiying Shen",
"Juncheng Yang",
"Yue Cheng"
] | Efficiently serving large language models (LLMs) under dynamic and bursty workloads remains a key challenge for real-world deployment. Existing serving frameworks and static model compression techniques fail to adapt to workload fluctuations, leading to either service-level objective (SLO) violations under full-precisi... | # | MLSys2026 |
CSLE: A Reinforcement Learning Platform for Autonomous Security Management | https://proceedings.mlsys.org/paper_files/paper/2026/hash/87eaaa8605a1a472d9a9756e7500517b-Abstract-Conference.html | [
"Kim Hammar"
] | Reinforcement learning is a promising approach to autonomous and adaptive security management in networked systems. However, current reinforcement learning solutions for security management are mostly limited to simulation environments and it is unclear how they generalize to operational systems. In this paper, we addr... | # | MLSys2026 |
The OpenHands Software Agent SDK: A Composable and Extensible Foundation for Production Agents | https://proceedings.mlsys.org/paper_files/paper/2026/hash/8ae9cf363ea625161f885b798c1f1f78-Abstract-Conference.html | [
"Xingyao Wang",
"Simon Rosenberg",
"Juan Michelini",
"Calvin Smith",
"Hoang H. Tran",
"Engel Nyst",
"Rohit Malhotra",
"Xuhui Zhou",
"Valerie Chen",
"Robert Brennan",
"Graham Neubig"
] | Agents are now used widely in the process of software development, but building production-ready software engineering agents is a complex task. Deploying software agents effectively requires flexibility in implementation and experimentation, reliable and secure execution, and interfaces for users to interact with agent... | # | MLSys2026 |
Agentic Operator Generation for ML ASICs | https://proceedings.mlsys.org/paper_files/paper/2026/hash/8c54e9bfed4119c873f575d1d1e2f0a0-Abstract-Conference.html | [
"Alec Hammond",
"Aram Markosyan",
"Aman Dontula",
"Simon Mahns",
"Zacharias Fisches",
"Dmitrii Pedchenko",
"Keyur Muzumdar",
"Natacha Supper",
"Site Cao",
"Haishan Zhu",
"Mark Saroufim",
"Joe Isaacson",
"Laura Wang",
"Warren Hunt",
"Kaustubh Gondkar",
"Roman Levenstein",
"Gabriel Syn... | We present TritorX, an agentic AI system designed to generate functionally correct Triton PyTorch ATen kernels at scale for emerging accelerator platforms. TritorX integrates large language models with a custom linter, JIT compilation, and a PyTorch OpInfo-based test harness. This pipeline is compatible with both real ... | # | MLSys2026 |
MoEBlaze: Breaking the Memory Wall for Efficient MoE Training on Modern GPUs | https://proceedings.mlsys.org/paper_files/paper/2026/hash/9032e5c9ec394ce768a2fa9bdc56af6c-Abstract-Conference.html | [
"Jiyuan Zhang",
"Yining Liu",
"Siqi Yan",
"Lisen Deng",
"Jennifer Cao",
"Shuqi Yang",
"Bi Xue",
"Min Ni",
"Shen Li"
] | The pervasive “memory wall” bottleneck is significantly amplified in modern large-scale Mixture-of-Experts (MoE) architectures. MoE's inherent architectural sparsity leads to sparse arithmetic compute and also introduces substantial activation memory overheads—driven by large token routing buffers and the need to mater... | # | MLSys2026 |
Blueprint, Bootstrap, and Bridge: A Security Look at NVIDIA GPU Confidential Computing | https://proceedings.mlsys.org/paper_files/paper/2026/hash/906419cd502575b617cc489a1a696a67-Abstract-Conference.html | [
"Zhongshu Gu",
"Enriquillo Valdez",
"Salman Ahmed",
"Julian James Stephen",
"Michael Le",
"Hani Jamjoom",
"Shixuan Zhao",
"Zhiqiang Lin"
] | NVIDIA GPU Confidential Computing (GPU-CC) aims to provide secure execution for AI workloads. For end users, enabling GPU-CC is seamless and requires no modifications to existing applications. However, this ease of adoption relies on a proprietary and highly complex system that is difficult to inspect, creating challen... | # | MLSys2026 |
Matrix: Peer-to-Peer Multi-Agent Synthetic Data Generation Framework | https://proceedings.mlsys.org/paper_files/paper/2026/hash/9069a8976ff06f6443e7f4172990a580-Abstract-Conference.html | [
"Dong Wang",
"Yang Li",
"Ansong Ni",
"Ching-Feng Yeh",
"Youssef Emad",
"Xinjie Lei",
"Liam Robbins",
"Karthik Padthe",
"Hu Xu",
"Xian Li",
"Asli Celikyilmaz",
"Ramya Raghavendra",
"LIFEI HUANG",
"Carole-Jean Wu",
"Shang-Wen Li"
] | Synthetic data has become increasingly important for training large language models, especially when real data is scarce, expensive, or privacy-sensitive. Many such generation tasks require coordinated multi-agent workflows, where specialized agents collaborate to produce data that is higher quality, more diverse, and ... | # | MLSys2026 |
NEST: Network- and Memory-Aware Device Placement for Distributed Deep Learning | https://proceedings.mlsys.org/paper_files/paper/2026/hash/949c07bf8756d51642c40b541c7306a8-Abstract-Conference.html | [
"Irene Wang",
"Vishnu Varma Venkata",
"Arvind Krishnamurthy",
"Divya Mahajan"
] | The growing scale of deep learning demands distributed training frameworks that jointly reason about parallelism, memory, and network topology. Prior works often rely on heuristic or topology-agnostic search, handling communication and memory separately. Without per-device memory awareness, these methods typically ensu... | https://github.com/scai-tech/Nest | MLSys2026 |
FlexiCache: Leveraging Temporal Stability of Attention Heads for Efficient KV Cache Management | https://proceedings.mlsys.org/paper_files/paper/2026/hash/94bcb01789fccf15afe2764d8fe0f40e-Abstract-Conference.html | [
"Nazmul Takbir",
"HamidReza Alikhani Koshkak",
"Nikil Dutt",
"Sangeetha Abdu Jyothi"
] | Large Language Model (LLM) serving is increasingly constrained by the growing size of the key-value (KV) cache, which scales with both context length and generation length. Prior work shows that attention is dominated by a small subset of critical tokens, yet existing systems struggle to exploit this efficiently withou... | # | MLSys2026 |
G-HEMP: FAST MULTI-GPU PRIVATE INFERENCE FOR LARGE-SCALE GCNS WITH HOMOMORPHIC ENCRYPTION | https://proceedings.mlsys.org/paper_files/paper/2026/hash/96894468eb44631a32d7ebd56f9892c7-Abstract-Conference.html | [
"Ran Ran",
"Zhaoting Gong",
"Zhaowei Li",
"Xianting Lu",
"Jiajia Li",
"Wujie Wen"
] | Homomorphic Encryption (HE) offers a promising solution for privacy-preserving Graph Convolutional Network
(GCN) inference in untrusted cloud environments by enabling computation directly on encrypted data. This
capability is particularly valuable in domains such as recommendation systems, financial analysis, and bioin... | # | MLSys2026 |
When Machine Learning Isn’t Sure: Building Resilient ML-Based Computer Systems by Embracing Uncertainty | https://proceedings.mlsys.org/paper_files/paper/2026/hash/96aca14d6c4dcd3adf54bc2c5ad7f138-Abstract-Conference.html | [
"Varun Gohil",
"Nevena Stojkovic",
"Noman Bashir",
"Sundar Dev",
"Gaurang Upasani",
"David Lo",
"Parthasarathy Ranganathan",
"Christina Delimitrou"
] | Machine learning (ML) models are increasingly used in computer systems but often suffer from poor generalizability, leading to costly failures on out-of-distribution (OOD) data. We propose an uncertainty-aware framework that improves system resilience by quantifying prediction uncertainty at runtime and rejecting unrel... | # | MLSys2026 |
Shannonic: Efficient Entropy-Optimal Compression for ML Workloads | https://proceedings.mlsys.org/paper_files/paper/2026/hash/96f39c8de84678cb2a908cd52bfd7819-Abstract-Conference.html | [
"Kareem Ibrahim",
"Mohammadjavad Maheronnaghsh",
"Andreas Moshovos"
] | We present Shannonic, a lossless compression method for machine learning tensors that achieves near-entropy-optimal compression, minimal state footprint, and high throughput. Shannonic uses an off-line pre-processing step to partition the tensor value space into optimally selected subranges and generates encoding/deco... | # | MLSys2026 |
Optimizing Deployment Configurations for LLM Inference | https://proceedings.mlsys.org/paper_files/paper/2026/hash/97dc07f1253ab33ee514f395a82fa7cc-Abstract-Conference.html | [
"Sung Min Cho",
"Jaewon Lee",
"Chunqiang Tang",
"Yejin Lee",
"Geonhwa Jeong",
"Anca Agape",
"Scott Batura",
"Vincent Boivin",
"Stephen Chen",
"Renfei Chen",
"Sijia Chen",
"Yan Cui",
"Bradley Davis",
"Zhaoxia (Summer) Deng",
"Nick Egebo",
"Emad El-Haraty",
"Sebastien Estienne",
"Lu ... | Meta's Large Language Models (LLMs)---the Llama model family---serve nearly one billion monthly active users. Deploying these models for inference involves navigating a complex design space that spans diverse hardware options (e.g., H100, H200, MI300X), multiple parallelism strategies (tensor, pipeline, expert, context... | # | MLSys2026 |
End of preview. Expand in Data Studio
PaperVault Dataset · 论文元数据库
🔎 项目简介 · Overview
PaperVault 是一份持续自动更新的统一论文元数据库，覆盖自然语言处理、计算机视觉、机器学习、数据挖掘、数据库、语音、系统、网络、安全、理论计算机科学、人机交互、计算机图形学与多媒体等方向的顶级会议与期刊。PaperVault is a continuously-updated, unified metadata database of papers from top-tier conferences and journals across NLP, Computer Vision, Machine Learning, Data Mining, Databases, Speech, Systems, Networking, Security, Theory, HCI, Graphics, and Multimedia.
🌐 源仓库 / Source: github.com/youngfish42/PaperVault — Web UI、REST API、采集流水线、Issues / PRs 全部在那里 · Web UI, REST API, crawling pipelines and issues/PRs all live there.
🆕 最近更新 · Recent Update
- 📅 最近更新 · Last updated: 2026-07-28 (Asia/Shanghai)
- 📊 数据库规模 · Database size: 649,635 篇论文 / 120 个刊物系列 / 540,155 篇含摘要 / 41,877 篇含开源代码（649,635 papers / 120 venue series / 540,155 with abstract / 41,877 with code）
📈 数据看板 · Statistics at a Glance
下列 4 张统计图与 cache.jsonl.gz 同源同步，反映本数据集的最新状态。The four charts below are generated from the same cache.jsonl.gz and always reflect the latest state of this dataset.
📦 数据集内容 · What's in this dataset
| 路径 Path | 子集 Subset | 格式 Format | 说明 Description |
|---|---|---|---|
cache/cache.jsonl.gz |
papers（默认 / default） |
gzip-compressed JSON Lines (UTF-8) | 每行一篇论文 · One paper per line; one JSON object per line |
cache/abstract_backfill_progress.jsonl.gz |
abstract_backfill_progress |
gzip-compressed JSON Lines (UTF-8) | 摘要回填流水线的进度/断点记录，不是论文元数据；仅供工作流恢复使用 · Append-only progress log of the abstract-backfill pipeline (not paper records); used by the workflow to resume between runs |
Hugging Face 会自动为 cache.jsonl.gz 生成 Parquet 视图，也可直接用 datasets.load_dataset(...) 读取，无需手动解压。Hugging Face also exposes an auto-generated Parquet view, so datasets.load_dataset(...) works out of the box.
💡 Dataset Viewer 与
datasets.load_dataset("youngfish42/PaperVault")默认展示/加载的都是papers子集（即cache/cache.jsonl.gz）。如需查看回填进度，请在 Viewer 顶部下拉框切换到abstract_backfill_progress，或调用load_dataset("youngfish42/PaperVault", name="abstract_backfill_progress")。The Dataset Viewer anddatasets.load_dataset("youngfish42/PaperVault")both default to thepaperssubset (cache/cache.jsonl.gz). To inspect backfill progress, switch the Viewer's subset dropdown toabstract_backfill_progressor callload_dataset("youngfish42/PaperVault", name="abstract_backfill_progress").
📐 字段 Schema
| Field 字段 | Type 类型 | Notes 说明 |
|---|---|---|
paper_name |
string | 论文标题（已归一化）· Normalised paper title |
paper_authors |
list[string] | 作者列表，按原始顺序 · Author names in order |
paper_url |
string | 论文在原始平台的链接（PDF 或落地页）· Canonical URL on the venue's site (PDF or landing page) |
paper_abstract |
string | 摘要；未回填时为空字符串 · Abstract; may be empty when not yet backfilled |
paper_code |
string | 从摘要中抽取出的 GitHub 仓库 URL；"#" 是「未发现代码链接」的占位符 · GitHub repository URL extracted from the abstract; "#" is the sentinel for "no code link discovered" |
conf |
string | 会议+年份标识，如 ACL2024、NIPS2023、CVPR2025；去掉末尾四位数字即可得到会议系列。注意 NeurIPS Proceedings 沿用历史命名 NIPS{year}。Venue + year identifier (e.g. ACL2024, NIPS2023, CVPR2025). Strip the trailing 4-digit year to recover the venue series. Note that NeurIPS Proceedings entries use the historical name NIPS{year}. |
缺失字段请按空字符串处理。Treat missing fields as empty strings.
⬇️ 获取方式 · How to download
下面三种方式任选其一即可，无需克隆 GitHub 仓库。Pick any one of the three options below — no GitHub clone is required.
方式 A · Option A — huggingface_hub（推荐 / recommended）
from huggingface_hub import hf_hub_download
import gzip, json
path = hf_hub_download(
repo_id="youngfish42/PaperVault",
filename="cache/cache.jsonl.gz",
repo_type="dataset",
)
with gzip.open(path, "rt", encoding="utf-8") as f:
for line in f:
record = json.loads(line)
# 在这里处理一条记录 · do something with the record
方式 B · Option B — datasets
from datasets import load_dataset
ds = load_dataset("youngfish42/PaperVault")
print(ds[next(iter(ds))][0])
数据集只有一个默认 split（非 ML 训练集），不要传
split="train"。Single default split — do not passsplit="train".
方式 C · Option C — huggingface-cli / 直接 HTTPS · Plain HTTPS
huggingface-cli download youngfish42/PaperVault \
cache/cache.jsonl.gz --repo-type dataset --local-dir ./data
💡 文件压缩后约 120 MB（会随数据持续增长），解压后是 GB 级 JSONL 流，请按行流式读取，不要整体载入内存。The file is ~120 MB compressed (and growing) and decompresses to a multi-GB JSONL stream. Stream it line-by-line rather than loading the whole thing into memory.
🔁 更新节奏 · Update cadence
数据集由三个 GitHub Actions 工作流负责重建并推送到本 Hub 仓库 / The dataset is rebuilt and pushed to this Hub repo by three GitHub Actions workflows：
| 工作流 Workflow | 触发节奏 Schedule | 推送的内容 What it pushes |
|---|---|---|
collect_papers |
每月 15 号 16:00 UTC + 手动触发 · 15th of every month at 16:00 UTC + workflow_dispatch |
增量抓取新发现的会议/年份组合 · Incremental crawl of newly-discovered conference/year combinations |
backfill_abstracts |
每月 1 号 00:00 UTC + 手动触发 · 1st of every month at 00:00 UTC + workflow_dispatch |
为已有论文回填 paper_abstract · Adds paper_abstract for papers that were collected without one |
update_readme |
仅手动触发 (workflow_dispatch) · Manual only (workflow_dispatch) |
默认仅刷新 README 与统计；当输入参数 mode=force 时执行全量重建 · Refreshes the README and statistics by default; performs a full rebuild only when invoked with mode=force |
每次推送都使用 Hugging Face 的 parent_commit 乐观锁机制，避免并发覆盖。Each push uses Hugging Face's parent_commit optimistic-lock mechanism to avoid silently overwriting concurrent updates.
🔗 关联仓库 · Related repository
如果你需要完整的搜索 Web UI（智能搜索 + Web of Science 风格的高级查询 DSL）、REST API（/api/v1/*）、抓取 / 合并 / 摘要回填流水线源码、收录会议范围、统计仪表盘、项目截图或贡献指南，请前往 GitHub 项目仓库。If you are looking for the full search Web UI (smart search + Web-of-Science-style advanced query DSL), the REST API (/api/v1/*), the crawling / merging / abstract-backfill pipelines source code, conference coverage, statistics dashboards, screenshots or contribution guidelines, please visit the GitHub repository.
👉 github.com/youngfish42/PaperVault
📜 许可证 · License
代码以 GPL-3.0 发布；每条论文记录的著作权仍属于原作者 / 出版方，本数据集仅重新分发公开可获取的元数据与链接。Code is released under GPL-3.0; individual paper records remain the IP of their authors/publishers — this dataset only redistributes publicly available bibliographic metadata and links.
- Downloads last month
- 1,315