Skip to yearly menu bar Skip to main content


Timezone: US/Pacific
Full Schedule Mon Tue Wed Thu Fri
Filter Events
Registration Desk
8:00 AM - 5:00 PM
Oral
8:30 AM - 10:00 AM
5 Events in this session
PRISM: PARAMETRICALLY RESTRUCTURED INFERENCE FOR SPECULATIVE SAMPLING DRAFT MODELS
Xuliang Wang ⋅ Yuetao Chen ⋅ Maochan Zhen ⋅ Fang LIU ⋅ Xinzhou Zheng ⋅ Xingwu Liu ⋅ Hong Xu ⋅ Ming Li
CDLM: CONSISTENCY DIFFUSION LANGUAGE MODELS FOR FASTER SAMPLING
Minseo Kim ⋅ Chenfeng Xu ⋅ Coleman Hooper ⋅ Harman Singh ⋅ Ben Athiwaratkun ⋅ Ce Zhang ⋅ Kurt Keutzer ⋅ Amir Gholami
Speculative Decoding: Performance or Illusion?
Lily Liu ⋅ Jiaxiang Yu ⋅ Jongseok Park ⋅ Alvin Cheung ⋅ Ion Stoica
SpecDiff-2: Scaling Diffusion Drafter Alignment For Faster Speculative Decoding
Jameson Sandler ⋅ Jacob K Christopher ⋅ ⋅ Ferdinando Fioretto
Accelerating Large-Scale Reasoning Model Inference with Sparse Self-Speculative Decoding
Yilong Zhao ⋅ Jiaming Tang ⋅ Kan Zhu ⋅ Zihao Ye ⋅ Chi-Chih Chang ⋅ Chaofan Lin ⋅ Jongseok Park ⋅ Guangxuan Xiao ⋅ Mohamed Abdelfattah ⋅ Mingyu Gao ⋅ Baris Kasikci ⋅ Song Han ⋅ Ion Stoica
Go to Event Page
Oral
8:30 AM - 10:00 AM
5 Events in this session
ADS: AN AGENTIC DETECTION SYSTEM FOR ENTERPRISE AGENTIC AI SECURITY
Chenning Li ⋅ Pan Hu ⋅ Justin Xu ⋅ Baris Ozbas ⋅ Olivia Liu ⋅ Caroline Van ⋅ ⋅ Wei Zhou ⋅ Mohammad Alizadeh ⋅ Pengyu Zhang
PROMPTS: PeRformance Optimization via Multi-Agent Planning for LLM Training and Serving
Yuran Ding ⋅ Ruobing Han ⋅ Xiaofan Zhang ⋅ Xinwei Chen
Agentic Operator Generation for ML ASICs
Alec Hammond ⋅ Aram Markosyan ⋅ Aman Dontula ⋅ ⋅ Zacharias Fisches ⋅ Dmitrii Pedchenko ⋅ Keyur Muzumdar ⋅ ⋅ Mark Saroufim ⋅ Joe Isaacson ⋅ Laura Wang ⋅ Warren Hunt ⋅ ⋅ ⋅ Gabriel Synnaeve ⋅ ⋅ Jacob Kahn ⋅
Cost-aware Duration Prediction for Software Upgrades in Datacenters
Yi Ding ⋅ Henry (Hank) Hoffmann
The OpenHands Software Agent SDK: A Composable and Extensible Foundation for Production Agents
Xingyao Wang ⋅ ⋅ Juan Michelini ⋅ Calvin Smith ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
Go to Event Page
Invited Talk
10:30 AM - 11:30 AM

Large language model training follows a standard pipeline: tokenization, pretraining, possibly mid-training, and post training or alignment. Despite its wild success, we understand relatively little about this recipe and are almost certainly missing many opportunities to improve it. In this talk, I will focus on three such cases. I’ll describe our work on data efficient post training (e.g. LIMA, ALMA, and s1) where we argue that nearly all advanced model capabilities ultimately come from the pretraining data, even if effective alignment is still essential for controlling model behavior. I will also describe new methods for extracting more signal from the pretraining data, including new hierarchical architectures for byte-level language models (e.g. BLT) that are both tokenizer-free and scale better than traditional BPE-based methods, especially in the long tail. Finally, I will discuss decentralized, modular training algorithms (e.g. BTM) that better isolate and control the influence of specific data on specific model components and behaviors. Together, these methods promise to simplify training and improve scaling, by centering and amplifying the influence of data in architecture design.

... more
Speaker Bio
Luke Zettlemoyer
Luke Zettlemoyer is a Professor in the Paul G. Allen School of Computer Science & Engineering at the University of Washington and a Senior Research Director at Meta. His research interests are in the intersections of natural language processing, machine learning, and decision making under uncertainty, with a recent emphasis on the science of training both text-based and multi-modal language models. Luke did postdoctoral research at the University of Edinburgh, earned his PhD at MIT, and was an undergraduate at NC State University. His honors include numerous paper awards, being named a Schmidt AI 2050 Senior Follow in 2025, elected President of the Association for Computational Linguistics (ACL) in 2024, named a Fellow of the ACL in 2021 along with winning the Presidential Early Career Award for Scientists and Engineers (PECASE) award in 2016, an Allen Distinguished Investigator Award in 2014, and the National Science Foundation (NSF) International Research Fellowship in 2009.
... more
Session
10:30 AM - 11:30 AM
Oral
1:00 PM - 2:30 PM
5 Events in this session
IntAttention: A Fully Integer Attention Pipeline for Efficient Edge Inference
Wanli Zhong ⋅ Haibo Feng ⋅ Zirui Zhou ⋅ Hanyang Peng ⋅ Shiqi Yu
Massive-Scale Out-Of-Core UMAP on the GPU
Jinsol Park ⋅ Corey Nolet ⋅ Edward Raff ⋅ Tim Oates ⋅ Akira Naruse
REPARO: LOSS-RESILIENT GENERATIVE CODEC FOR VIDEO CONFERENCING
Tianhong Li ⋅ Vibhaalakshmi Sivaraman ⋅ Pantea Karimi ⋅ Lijie Fan ⋅ Mohammad Alizadeh ⋅ Dina Katabi
Rethinking DVFS for Mobile LLMs: Unified Energy-Aware Scheduling with CORE
Zongpu Zhang ⋅ ⋅ Y. Charlie Hu ⋅ Qiang Xu ⋅ Jian Li ⋅ Haibing Guan
EarthSight: A Distributed Framework for Low-Latency Satellite Intelligence
Ansel Erol ⋅ Seungjun Lee ⋅ Divya Mahajan
Go to Event Page
Oral
1:00 PM - 2:30 PM
6 Events in this session
SHIP: SRAM-Based Huge Inference Pipelines for Fast LLM Serving
⋅ ⋅ ⋅ ⋅ ⋅ Sahil Parmar ⋅ ⋅ ⋅ ⋅ ⋅
BatchLLM: Optimizing Large Batched LLM Inference with Global Prefix Sharing and Throughput-oriented Token Batching
Zhen Zheng ⋅ Xin Ji ⋅ Taosong Fang ⋅ Fanghao Zhou ⋅ Chuanjie Liu ⋅ Gang Peng
Beyond the Buzz: A Pragmatic Take on Inference Disaggregation
Tiyasa Mitra ⋅ Ritika Borkar ⋅ Nidhi Bhatia ⋅ Shivam Raj ⋅ hongkuan zhou ⋅ Yan Ru Pei ⋅ ⋅ Kyle Kranen ⋅ Ramon Matas ⋅ Dheevatsa Mudigere ⋅ Ritchie Zhao ⋅ Maximilian Golub ⋅ Arpan Dutta ⋅ ⋅ Sailaja Madduri ⋅ Dharmesh Jani ⋅ Brian Pharris ⋅ Itay Neeman ⋅ Bita Darvish Rouhani
Optimizing Deployment Configurations for LLM Inference
Sungmin Cho ⋅ Jaewon Lee ⋅ Chunqiang Tang ⋅ Yejin Lee ⋅ Geonhwa Jeong ⋅ ⋅ Scott Batura ⋅ ⋅ ⋅ ⋅ Sijia Chen ⋅ ⋅ Bradley Davis ⋅ Summer Deng ⋅ ⋅ Emad El-Haraty ⋅ ⋅ Lu Fang ⋅ Lu Fang ⋅ Joshua Fromm ⋅ ⋅ ⋅ Liangpeng Guo ⋅ ⋅ ⋅ Jianyu Huang ⋅ Aya Ibrahim ⋅ ⋅ Hongyi Jia ⋅ Changkyu Kim ⋅ ⋅ ⋅ ⋅ ⋅ Xiaozhu Meng ⋅ Vlad Tiberiu Mihailescu ⋅ ⋅ Maxim Naumov ⋅ Michal Ostrowski ⋅ ⋅ ⋅ Sarunya Pumma ⋅ ⋅ ⋅ Jeremy Francis Reizenstein ⋅ Rajasi Saha ⋅ ⋅ Zhan Shu ⋅ Ruan Silva ⋅ ⋅ Jon Swenson ⋅ ⋅ Chris Thi ⋅ ⋅ Yunfan Wang ⋅ Pengchao Wang ⋅ Wenchen Wang ⋅ ⋅ Bram Wasti ⋅ ⋅ ⋅ Jingyi Yang ⋅ ⋅ ⋅ Jing Zhang ⋅ Yi Zhen ⋅
Meeting SLOs, Slashing Hours: Automated Enterprise LLM Optimization with OptiKIT
Nicholas Santavas ⋅ Kareem Eissa ⋅ ⋅ Piotr Florek ⋅ Matteo Nulli ⋅ Stefan Vasilev ⋅ Seyyed Hashemi ⋅ Antonios Gasteratos ⋅ Shahram Khadivi
Scaling Up Large Language Models Serving Systems for Semantic Job Search
Kayhan Behdin ⋅ Qingquan Song ⋅ Sriram Vasudevan ⋅ Jian Sheng ⋅ Xiaojing Ma ⋅ Zhengze Zhou ⋅ Chuanrui Zhu ⋅ Guoyao Li ⋅ Chanh Nguyen ⋅ ⋅ Hejian Sang ⋅ Ata Fatahi ⋅ ⋅ Xiaoqing Wang ⋅ Qing Lan ⋅ ⋅ Qi Guo ⋅ Caleb Johnson ⋅ Zhipeng Wang ⋅
Go to Event Page
Oral
2:45 PM - 4:15 PM
6 Events in this session
Sparing Strategies to Minimize Reliability Impact On Large Training Jobs
⋅ ⋅ Ehsan K. Ardestani ⋅ ⋅ ⋅ ⋅ Zhaodong Wang ⋅ ⋅ Xu Zhang ⋅ ⋅ Ying Zhang
FlexScale: Flexible and High-Performance FSDP at Scale
Zezhou Wang ⋅ Youjie Li ⋅ Zhiqi Lin ⋅ Jiacheng Yang ⋅ Cong Xie ⋅ ⋅ ZHENG ZHONG ⋅ Ziyue Huang ⋅ Hongyu Zhu ⋅ Zhi Zhang ⋅ Xin Liu ⋅ Yanghua Peng
AXLearn: Modular, Hardware-Agnostic Large Model Training
Mark Lee ⋅ Tom Gunter ⋅ Chang Lan ⋅ ⋅ Hanzhi Zhou ⋅ ⋅ Sneha Bangalore ⋅ ⋅ ⋅ Xianzhi Du ⋅ Philipp Dufter ⋅ ⋅ Ruixuan Hou ⋅ Haoshuo Huang ⋅ ⋅ Xiang Kong ⋅ Jinhao Lei ⋅ Tao Lei ⋅ Meng Li ⋅ Li Li ⋅ Jiarui Lu ⋅ Zhiyun Lu ⋅ ⋅ ⋅ ⋅ ⋅ Zhucheng Tu ⋅ Chong Wang ⋅ Jianyu Wang ⋅ ⋅ Zirui Wang ⋅ ⋅ Sam Wiseman ⋅ Guoli Yin ⋅ ⋅ Xiyou Zhou ⋅ Danyang Zhuo ⋅ ⋅ Ruoming Pang
NodeSweep: Practical Straggler Detection and Health Monitoring for Large-Scale Foundation Model Training
Guanliang Liu ⋅ Zoe Zeng ⋅ ⋅ Cong Cheng ⋅ ⋅ ⋅ ⋅ ⋅ Abhinandan Patni ⋅ vijay rajakumar ⋅ Alexander Zhipa ⋅ Ashvin Nihalani ⋅ Binxuan Huang ⋅ Parthasarathy Govindarajen ⋅
MoEBlaze: Breaking the Memory Wall for Efficient MoE Training on Modern GPUs
Jiyuan Zhang ⋅ Yining Liu ⋅ Siqi Yan ⋅ Lisen Deng ⋅ Jennifer Cao ⋅ Shuqi Yang ⋅ Bi Xue ⋅ Min Ni ⋅ Shen Li
FreeScale: Distributed Training for Sequence Recommendation Models with Minimal Scaling Cost
Chenhao Feng ⋅ Haoli Zhang ⋅ Shakhzod Ali-zade ⋅ Yanli Zhao ⋅ Liang Luo ⋅ Jennifer Cao ⋅ Lisen Deng ⋅ ⋅ Chenyu Zhao ⋅ ⋅ ⋅ ⋅ Tiantu Xu ⋅ Yi Zhang ⋅ Evgenii Kolpakov ⋅ Siqi Yan ⋅ Chuanhao Zhuge ⋅ Min Ni ⋅ Bi Xue ⋅ Qunshu Zhang ⋅ Shen Li
Go to Event Page
Oral
2:45 PM - 4:15 PM
4 Events in this session
Charon: A Unified and Fine-Grained Simulator for Large-Scale LLM Training and Inference
Mengtian Yang ⋅ Zhekun Zhang ⋅ Mingheng Wu ⋅ jianwen yan ⋅ Hanshi Sun ⋅ Li-Wen Chang
Hawkeye: Reproducing GPU-Level Non-Determinism
⋅ Dan Boneh ⋅ Ilan Komargodski ⋅ Megha Srivastava
DriftBench: Measuring and Predicting Infrastructure Drift in LLM Serving Systems
Gianluigi Vitale
Parrot: Persuasion and Agreement Robustness Rating of Output Truth
Yusuf Çelebi ⋅ Mahmoud ElHussieni ⋅ Özay Ezerceli
Go to Event Page
Oral
4:30 PM - 6:00 PM
6 Events in this session
XProf: An Open, Scalable, and Extensible Profiling System for the Modern ML Stack
Clive Verghese ⋅ Prasanna Rengasamy ⋅ ⋅ Yin Zhang ⋅ Jiya Zhang ⋅ ⋅ Charles Alaras ⋅ Aditya Sharma ⋅ ⋅ ⋅ Rushabh Lalwani ⋅ Sannidhya Chauhan ⋅ Sai Ganesh Bandiatmakuri ⋅ ⋅ Ani Udipi ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ Naveen Kumar ⋅ ⋅ Sayce Falk ⋅ ⋅
AIRS: Scaling Live Inference in Resource Constrained Environments
Nilesh Jagnik ⋅ ⋅ ⋅ ⋅ Harshvardhan GM
SAKURAONE: An Open Ethernet–Based AI HPC System and Its Observed Workload Dynamics in a Single-Tenant LLM Development Environment
Fumikazu KONISHI ⋅ Yuuki Tsubouchi ⋅ Hirofumi Tsuruta
MLCommons Chakra: Advancing Performance Benchmarking and Co-design using Standardized Execution Traces
Srinivas ⋅ Taekyung Heo ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ Hanjiang Wu ⋅ Changhai Man ⋅ Jinsun Yoo ⋅ Huan Xu ⋅ William Won ⋅ ⋅ Winston Liu ⋅ Andrey Balogh ⋅ Dan Mihailescu ⋅ Brad B ⋅ Vinay Ramakrishnaiah ⋅ Spandan More ⋅ Saeed Rashidi ⋅ Louis Feng ⋅ Ashwin Ramachandran ⋅ Puneet Sharma ⋅ ⋅ Vijay Janapa Reddi ⋅ David Kanter ⋅ Tushar Krishna
ProfInfer: An eBPF-based Fine-Grained LLM Inference Profiler
Bohua Zou ⋅ ⋅ ⋅ Weihao Xu ⋅ Binqi Sun ⋅ ⋅
ML Fleet Efficiency: Improving TPU Systems at Scale with ML Productivity Goodput
Arissa Wongpanich ⋅ Tayo Oguntebi ⋅ ⋅ Yu Wang ⋅ Phitchaya Phothilimthana ⋅ Ritwika Mitra ⋅ Zongwei Zhou ⋅ Naveen Kumar ⋅ Vijay Janapa Reddi
Go to Event Page
Oral
4:30 PM - 6:00 PM
6 Events in this session
ParallelKittens: Systematic and Practical Simplification of Multi-GPU AI Kernels
Stuart H. Sul ⋅ Simran Arora ⋅ Benjamin Spector ⋅ Christopher Ré
Event Tensor: A Unified Abstraction for Compiling Dynamic Megakernel
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 ⋅ ⋅ Xupeng Miao ⋅ Vinod Grover ⋅ Todd Mowry ⋅ Zhihao Jia ⋅ Tianqi Chen
AccelOpt: A Self-Improving LLM Agentic System for AI Accelerator Kernel Optimization
Genghan Zhang ⋅ Shaowei Zhu ⋅ ⋅ ⋅ Allen Nie ⋅ Zhen Jia ⋅ Nandita Vijaykumar ⋅ Yida Wang ⋅ Kunle Olukotun
SchedFlow: Transparent and Flexible Intra-Device Parallelism via Programmable Operator Scheduling
Yi Pan ⋅ Yile Gu ⋅ Luo Jinbin ⋅ Yibo Wu ⋅ Ziren Wang ⋅ ⋅ Ziyi Xu ⋅ Shengkai Lin ⋅ Stephanie Wang ⋅ Baris Kasikci
ApproxMLIR : Accuracy-Aware Compiler for Compound ML System
Hao Ren ⋅ Yi Mu ⋅ Sasa Misailovic
HipKittens: Fast and Furious AMD Kernels
William Hu ⋅ Drew Wadsworth ⋅ Sean Siddens ⋅ ⋅ Daniel Fu ⋅ Ryan Swann ⋅ Muhammad Osama ⋅ Christopher Ré ⋅ Simran Arora
Go to Event Page
Poster
6:00 PM - 8:00 PM
54 Events in this session
Accelerating Large-Scale Reasoning Model Inference with Sparse Self-Speculative Decoding
Yilong Zhao ⋅ Jiaming Tang ⋅ Kan Zhu ⋅ Zihao Ye ⋅ Chi-Chih Chang ⋅ Chaofan Lin ⋅ Jongseok Park ⋅ Guangxuan Xiao ⋅ Mohamed Abdelfattah ⋅ Mingyu Gao ⋅ Baris Kasikci ⋅ Song Han ⋅ Ion Stoica
AccelOpt: A Self-Improving LLM Agentic System for AI Accelerator Kernel Optimization
Genghan Zhang ⋅ Shaowei Zhu ⋅ ⋅ ⋅ Allen Nie ⋅ Zhen Jia ⋅ Nandita Vijaykumar ⋅ Yida Wang ⋅ Kunle Olukotun
ADS: AN AGENTIC DETECTION SYSTEM FOR ENTERPRISE AGENTIC AI SECURITY
Chenning Li ⋅ Pan Hu ⋅ Justin Xu ⋅ Baris Ozbas ⋅ Olivia Liu ⋅ Caroline Van ⋅ ⋅ Wei Zhou ⋅ Mohammad Alizadeh ⋅ Pengyu Zhang
Agentic Operator Generation for ML ASICs
Alec Hammond ⋅ Aram Markosyan ⋅ Aman Dontula ⋅ ⋅ Zacharias Fisches ⋅ Dmitrii Pedchenko ⋅ Keyur Muzumdar ⋅ ⋅ Site Cao ⋅ ⋅ Mark Saroufim ⋅ Joe Isaacson ⋅ Laura Wang ⋅ Warren Hunt ⋅ ⋅ ⋅ Gabriel Synnaeve ⋅ ⋅ Jacob Kahn ⋅
AIRS: Scaling Live Inference in Resource Constrained Environments
Nilesh Jagnik ⋅ ⋅ ⋅ ⋅ Harshvardhan GM
ApproxMLIR : Accuracy-Aware Compiler for Compound ML System
Hao Ren ⋅ Yi Mu ⋅ Sasa Misailovic
Attribution-based Sparse Activation in Large Language Models
Jifeng Song ⋅ Xiangyu Yin ⋅ Boyuan Yang ⋅ Kai Huang ⋅ Weichen Liu ⋅ Wei Gao
AXLearn: Modular, Hardware-Agnostic Large Model Training
Mark Lee ⋅ Chang Lan ⋅ Tom Gunter ⋅ John Peebles ⋅ Hanzhi Zhou ⋅ ⋅ Sneha Bangalore ⋅ ⋅ ⋅ Xianzhi Du ⋅ Philipp Dufter ⋅ ⋅ Ruixuan Hou ⋅ Haoshuo Huang ⋅ ⋅ Xiang Kong ⋅ Jinhao Lei ⋅ Tao Lei ⋅ Meng Li ⋅ Li Li ⋅ Jiarui Lu ⋅ Zhiyun Lu ⋅ ⋅ ⋅ ⋅ ⋅ Zhucheng Tu ⋅ Chong Wang ⋅ Jianyu Wang ⋅ ⋅ Zirui Wang ⋅ ⋅ Sam Wiseman ⋅ Guoli Yin ⋅ ⋅ Xiyou Zhou ⋅ Danyang Zhuo ⋅ ⋅ Ruoming Pang
BatchLLM: Optimizing Large Batched LLM Inference with Global Prefix Sharing and Throughput-oriented Token Batching
Zhen Zheng ⋅ Xin Ji ⋅ Taosong Fang ⋅ Fanghao Zhou ⋅ Chuanjie Liu ⋅ Gang Peng
Beyond the Buzz: A Pragmatic Take on Inference Disaggregation
Tiyasa Mitra ⋅ Ritika Borkar ⋅ Nidhi Bhatia ⋅ Shivam Raj ⋅ hongkuan zhou ⋅ Yan Ru Pei ⋅ ⋅ Kyle Kranen ⋅ Ramon Matas ⋅ Dheevatsa Mudigere ⋅ Ritchie Zhao ⋅ Maximilian Golub ⋅ Arpan Dutta ⋅ ⋅ Sailaja Madduri ⋅ Dharmesh Jani ⋅ Brian Pharris ⋅ Itay Neeman ⋅ Bita Darvish Rouhani
BLASST: Dynamic BLocked Attention Sparsity via Softmax Thresholding
Jiayi Yuan ⋅ Cameron Shinn ⋅ Kai Xu ⋅ Jingze Cui ⋅ George Klimiashvili ⋅ ⋅ Perkz Zheng ⋅ Bo Li ⋅ Zhou Yuxin ⋅ Zhouhai Ye ⋅ Weijie You ⋅ Tian Zheng ⋅ Dominic Brown ⋅ Pengbo Wang ⋅ Markus Hoehnerbach ⋅ Richard Cai ⋅ Julien Demouth ⋅ John D. Owens ⋅ Xia Hu ⋅ ⋅ Timmy Liu ⋅ Huizi Mao
CATWILD: Compiler Autotuning for TPU workloads in the Wild
Ignacio Cano ⋅ Yu Wang ⋅ Mike Burrows ⋅ ⋅ Matheus Camargo ⋅ Chao Wang ⋅ David Liu ⋅ Tengyu Sun ⋅ Alexander Wertheim ⋅ Arissa Wongpanich ⋅ Christof Angermueller ⋅ ⋅ ⋅ ⋅ Amit Sabne ⋅ ⋅ Mehrdad Khani ⋅ ⋅ Berkin Ilbeyi ⋅ Subhankar Shah ⋅ Ryan Lefever ⋅ Arjun Khare ⋅ Ankit Sinha ⋅ ⋅ ⋅ ⋅ Emily Donahue ⋅ Sami Abu-El-Haija ⋅ Nikhil Sarda ⋅ Vineetha Govindaraj ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ Naveen Kumar ⋅ Phitchaya Phothilimthana
CDLM: Consistency Diffusion Language Models for Faster Sampling
Minseo Kim ⋅ Chenfeng Xu ⋅ Coleman Hooper ⋅ Harman Singh ⋅ Ben Athiwaratkun ⋅ Ce Zhang ⋅ Kurt Keutzer ⋅ Amir Gholami
Charon: A Unified and Fine-Grained Simulator for Large-Scale LLM Training and Inference
Mengtian Yang ⋅ Zhekun Zhang ⋅ Mingheng Wu ⋅ jianwen yan ⋅ Hanshi Sun ⋅ Li-Wen Chang
Cost-aware Duration Prediction for Software Upgrades in Datacenters
Yi Ding ⋅ Aijia Gao ⋅ ⋅ Michal Sedlak ⋅ ⋅ Igor Marnat ⋅ Henry (Hank) Hoffmann
Dataflow Is All You Need
Darshan Gandhi ⋅ Pushkar Nandkar ⋅ David Koeplinger ⋅ ⋅ Romy Tsoupidi ⋅ ⋅ ⋅ Tuowen Zhao ⋅ Reid Goodbar ⋅ ⋅ Leon Zhang ⋅ ⋅ John Long ⋅ Han Wang ⋅ ⋅ ⋅ ⋅ Yun Du ⋅ Håkan Zeffer ⋅ ⋅ Raghu Prabhakar
DriftBench: Measuring and Predicting Infrastructure Drift in LLM Serving Systems
Gianluigi Vitale
DynaFlow: Transparent and Flexible Intra-Device Parallelism via Programmable Operator Scheduling
Yi Pan ⋅ Yile Gu ⋅ Luo Jinbin ⋅ Yibo Wu ⋅ Ziren Wang ⋅ ⋅ Ziyi Xu ⋅ Shengkai Lin ⋅ Baris Kasikci ⋅ Stephanie Wang
EarthSight: A Distributed Framework for Low-Latency Satellite Intelligence
Ansel Erol ⋅ Seungjun Lee ⋅ Divya Mahajan
Efficient, VRAM-Constrained xLM Inference on Clients
Aditya Ukarande ⋅ Deep Shekhar ⋅ Marc Blackstein ⋅ Ram Rangan
Event Tensor: A Unified Abstraction for Compiling Dynamic Megakernel
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 ⋅ ⋅ Xupeng Miao ⋅ Vinod Grover ⋅ Todd Mowry ⋅ Zhihao Jia ⋅ Tianqi Chen
ExecuTorch - A Unified PyTorch Solution to Run ML Models On-Device
Mergen Nachin ⋅ Digant Desai ⋅ Sicheng Jia ⋅ Chen Lai ⋅ Mengwei Liu ⋅ Jacob Szwejbka ⋅ ⋅ Robert Ascani ⋅ ⋅ Manuel Candales ⋅ ⋅ Yanan Cao ⋅ Zhengxu Chen ⋅ Soumith Chintala ⋅ Gregory Comer ⋅ ⋅ Songhao Jia ⋅ ⋅ Abhinay Kukkadapu ⋅ ⋅ Andrew Or ⋅ ⋅ ⋅ Lucy Qiu ⋅ Supriya Rao ⋅ ⋅ ⋅ Scott Roy ⋅ ⋅ ⋅ ⋅ Angela Yi ⋅ ⋅ Hansong Zhang ⋅ ⋅ Zhenrui Zhang ⋅ Shunting Zhang ⋅ Cemal Bilgin
FlashAttention-4: Algorithm and Kernel Pipelining Co-Design for Asymmetric Hardware Scaling
Ted Zadouri ⋅ Markus Hoehnerbach ⋅ Jay Shah ⋅ ⋅
Flashlight: PyTorch Compiler Extensions to Accelerate Attention Variants
Bozhi You ⋅ Irene Wang ⋅ ⋅ Abhinav Jangda ⋅ Angélica Moreira ⋅ Roshan Dathathri ⋅ Divya Mahajan ⋅ Keshav Pingali
FreeScale: Distributed Training for Sequence Recommendation Models with Minimal Scaling Cost
Chenhao Feng ⋅ Haoli Zhang ⋅ Shakhzod Ali-zade ⋅ Yanli Zhao ⋅ Liang Luo ⋅ Jennifer Cao ⋅ Lisen Deng ⋅ ⋅ Chenyu Zhao ⋅ ⋅ ⋅ ⋅ Tiantu Xu ⋅ Yi Zhang ⋅ Evgenii Kolpakov ⋅ Siqi Yan ⋅ Chuanhao Zhuge ⋅ Min Ni ⋅ Bi Xue ⋅ Qunshu Zhang ⋅ Shen Li
GUARD: SCALABLE STRAGGLER DETECTION AND NODE HEALTH MANAGEMENT FOR LARGE-SCALE TRAINING
Guanliang Liu ⋅ Abhinandan Patni ⋅ ⋅ Zoe Zeng ⋅ ⋅ ⋅ ⋅ ⋅ Alexander Zhipa ⋅ Ashvin Nihalani ⋅ Binxuan Huang ⋅ Cong Cheng ⋅ ⋅ vijay rajakumar ⋅ Rejith Joseph ⋅ Parthasarathy Govindarajen
Hawkeye: Reproducing GPU-Level Non-Determinism
⋅ Dan Boneh ⋅ Ilan Komargodski ⋅ Megha Srivastava
HipKittens: Fast and Furious AMD Kernels
William Hu ⋅ Drew Wadsworth ⋅ Sean Siddens ⋅ ⋅ Daniel Fu ⋅ Ryan Swann ⋅ Muhammad Osama ⋅ Christopher Ré ⋅ Simran Arora
IntAttention: A Fully Integer Attention Pipeline for Efficient Edge Inference
Wanli Zhong ⋅ Haibo Feng ⋅ Zirui Zhou ⋅ Hanyang Peng ⋅ Shiqi Yu
LEANN: A Low-Storage Overhead Vector Index
Yichuan Wang ⋅ Zhifei Li ⋅ Shu Liu ⋅ Yongji Wu ⋅ Ziming Mao ⋅ Yilong Zhao ⋅ Xiao Yan ⋅ Zhiying Xu ⋅ Yang Zhou ⋅ Ion Stoica ⋅ Sewon Min ⋅ Matei Zaharia ⋅ Joseph Gonzalez
Massive-Scale Out-Of-Core UMAP on the GPU
Jinsol Park ⋅ Corey Nolet ⋅ Edward Raff ⋅ Tim Oates ⋅ Akira Naruse
Meeting SLOs, Slashing Hours: Automated Enterprise LLM Optimization with OptiKIT
Nicholas Santavas ⋅ Kareem Eissa ⋅ ⋅ Piotr Florek ⋅ Matteo Nulli ⋅ Stefan Vasilev ⋅ Seyyed Hashemi ⋅ Antonios Gasteratos ⋅ Shahram Khadivi
MLCommons Chakra: Advancing Performance Benchmarking and Co-design using Standardized Execution Traces
Srinivas ⋅ Taekyung Heo ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ Hanjiang Wu ⋅ Changhai Man ⋅ Jinsun Yoo ⋅ Huan Xu ⋅ William Won ⋅ ⋅ Winston Liu ⋅ Andrey Balogh ⋅ Dan Mihailescu ⋅ Brad B ⋅ Vinay Ramakrishnaiah ⋅ Spandan More ⋅ Saeed Rashidi ⋅ Louis Feng ⋅ Ashwin Ramachandran ⋅ Puneet Sharma ⋅ ⋅ Vijay Janapa Reddi ⋅ David Kanter ⋅ Tushar Krishna
ML Fleet Efficiency: Improving TPU Systems at Scale with ML Productivity Goodput
Arissa Wongpanich ⋅ Tayo Oguntebi ⋅ ⋅ Yu Wang ⋅ Phitchaya Phothilimthana ⋅ Ritwika Mitra ⋅ Zongwei Zhou ⋅ Naveen Kumar ⋅ Vijay Janapa Reddi
MoEBlaze: Breaking the Memory Wall for Efficient MoE Training on Modern GPUs
Jiyuan Zhang ⋅ Yining Liu ⋅ Siqi Yan ⋅ Lisen Deng ⋅ Jennifer Cao ⋅ Shuqi Yang ⋅ Bi Xue ⋅ Min Ni ⋅ Shen Li
Optimizing Deployment Configurations for LLM Inference
Sungmin Cho ⋅ Jaewon Lee ⋅ Chunqiang Tang ⋅ Yejin Lee ⋅ Geonhwa Jeong ⋅ ⋅ Scott Batura ⋅ ⋅ ⋅ ⋅ Sijia Chen ⋅ ⋅ Bradley Davis ⋅ Summer Deng ⋅ ⋅ Emad El-Haraty ⋅ ⋅ Lu Fang ⋅ Lu Fang ⋅ Joshua Fromm ⋅ ⋅ ⋅ Liangpeng Guo ⋅ ⋅ ⋅ Jianyu Huang ⋅ Aya Ibrahim ⋅ ⋅ Hongyi Jia ⋅ Changkyu Kim ⋅ ⋅ ⋅ ⋅ ⋅ Xiaozhu Meng ⋅ Vlad Tiberiu Mihailescu ⋅ ⋅ Maxim Naumov ⋅ Michal Ostrowski ⋅ ⋅ ⋅ Sarunya Pumma ⋅ ⋅ ⋅ Jeremy Francis Reizenstein ⋅ Rajasi Saha ⋅ ⋅ Zhan Shu ⋅ Ruan Silva ⋅ ⋅ Jon Swenson ⋅ ⋅ Chris Thi ⋅ ⋅ Yunfan Wang ⋅ Pengchao Wang ⋅ Wenchen Wang ⋅ ⋅ Bram Wasti ⋅ ⋅ ⋅ Jingyi Yang ⋅ ⋅ ⋅ Jing Zhang ⋅ Yi Zhen ⋅
ParallelKittens: Systematic and Practical Simplification of Multi-GPU AI Kernels
Stuart H. Sul ⋅ Simran Arora ⋅ Benjamin Spector ⋅ Christopher Ré
PARROT: Persuasion and Agreement Robustness Rating of Output Truth — A Sycophancy Robustness Benchmark for LLMs
Yusuf Çelebi ⋅ Özay Ezerceli ⋅ Mahmoud ElHussieni
PRISM: Parametrically Refactor Inference for Speculative Decoding Draft Models
Xuliang Wang ⋅ Yuetao Chen ⋅ Maochan Zhen ⋅ Fang LIU ⋅ Xinzhou Zheng ⋅ Xingwu Liu ⋅ Hong Xu ⋅ Ming Li
ProfInfer: An eBPF-based Fine-Grained LLM Inference Profiler
Bohua Zou ⋅ ⋅ ⋅ Weihao Xu ⋅ Binqi Sun ⋅ ⋅
PROMPTS: PeRformance Optimization via Multi-Agent Planning for LLM Training and Serving
Yuran Ding ⋅ Ruobing Han ⋅ Xiaofan Zhang ⋅ Xinwei Chen
REPARO: LOSS-RESILIENT GENERATIVE CODEC FOR VIDEO CONFERENCING
Tianhong Li ⋅ Vibhaalakshmi Sivaraman ⋅ Pantea Karimi ⋅ Lijie Fan ⋅ Mohammad Alizadeh ⋅ Dina Katabi
Rethinking DVFS for Mobile LLMs: Unified Energy-Aware Scheduling with CORE
Zongpu Zhang ⋅ ⋅ Qiang Xu ⋅ Y. Charlie Hu ⋅ Jian Li ⋅ Haibing Guan
SAKURAONE: An Open Ethernet–Based AI HPC System and Its Observed Workload Dynamics in a Single-Tenant LLM Development Environment
Fumikazu KONISHI ⋅ Yuuki Tsubouchi ⋅ Hirofumi Tsuruta
Scaling Up Large Language Models Serving Systems for Semantic Job Search
Kayhan Behdin ⋅ Qingquan Song ⋅ Sriram Vasudevan ⋅ Jian Sheng ⋅ Xiaojing Ma ⋅ Zhengze Zhou ⋅ Chuanrui Zhu ⋅ Guoyao Li ⋅ Chanh Nguyen ⋅ ⋅ Hejian Sang ⋅ Ata Fatahi ⋅ ⋅ Xiaoqing Wang ⋅ Qing Lan ⋅ ⋅ Qi Guo ⋅ Caleb Johnson ⋅ Zhipeng Wang ⋅
SHIP: SRAM-Based Huge Inference Pipelines for Fast LLM Serving
Andrew Bitar ⋅ ⋅ Baorui Zhou ⋅ Matthew Boyd ⋅ Charlie Wang ⋅ ⋅ Eugene Sha ⋅ ⋅ ⋅ Alex Bowe ⋅ ⋅ Santosh Raghavan ⋅ ⋅ ⋅ ⋅ Kris Kang ⋅ ⋅ ⋅ Mohamed Eldafrawy ⋅ ⋅ ⋅ ⋅ ⋅ Andrew Paprotskyi ⋅ Arash Taheri-Dezfouli ⋅ ⋅ Andrew Ling
Sparing Strategies to Minimize Reliability Impact On Large Training Jobs
⋅ ⋅ Ehsan K. Ardestani ⋅ ⋅ ⋅ ⋅ Zhaodong Wang ⋅ ⋅ Xu Zhang ⋅ ⋅ Ying Zhang ⋅ ⋅ Chunqiang Tang
SpecDiff-2: Scaling Diffusion Drafter Alignment For Faster Speculative Decoding
Jameson Sandler ⋅ Jacob K Christopher ⋅ ⋅ Ferdinando Fioretto
Speculative Decoding: Performance or Illusion?
Xiaoxuan Liu ⋅ Jiaxiang Yu ⋅ Jongseok Park ⋅ Ion Stoica ⋅ Alvin Cheung
Spira: Exploiting Voxel Data Structural Properties for Efficient Sparse Convolution in Point Cloud Networks
Dionysios Adamopoulos ⋅ Anastasia Poulopoulou ⋅ Georgios Goumas ⋅ Christina Giannoula
The OpenHands Software Agent SDK: A Composable and Extensible Foundation for Production Agents
Xingyao Wang ⋅ Simon Rosenberg ⋅ Juan Michelini ⋅ Calvin Smith ⋅ Hoang Tran ⋅ ⋅ Rohit Malhotra ⋅ Xuhui Zhou ⋅ Valerie Chen ⋅ Robert Brennan ⋅ Graham Neubig
veScale-FSDP: Flexible and High-Performance FSDP at Scale
Zezhou Wang ⋅ Youjie Li ⋅ Zhiqi Lin ⋅ Jiacheng Yang ⋅ Cong Xie ⋅ ⋅ ZHENG ZHONG ⋅ Ziyue Huang ⋅ Hongyu Zhu ⋅ Zhi Zhang ⋅ Xin Liu ⋅ Yanghua Peng
Wave: A Symbolic Python DSL And Compiler for High-Performance Machine Learning
Harsh Menon ⋅ Oleksandr Zinenko ⋅ Gaurav Verma ⋅ ⋅ ⋅ Nithin Meganathan ⋅ Sanket Pandit ⋅ William Gallard Hatch ⋅ ⋅ ⋅ Sahil FAIZAL ⋅ ⋅ ⋅ Martin P. Lücke
XProf: An Open, Scalable, and Extensible Profiling System for the Modern ML Stack
⋅ Naveen Kumar ⋅ ⋅ ⋅ Clive Verghese ⋅ Prasanna Rengasamy ⋅ ⋅ Jiya Zhang ⋅ Charles Alaras ⋅ Yin Zhang ⋅ ⋅ ⋅ Sai Ganesh Bandiatmakuri ⋅ ⋅ Ani Udipi ⋅
Go to Event Page