54 citations · 78 across the 17 of their papers we have counts for
20 papers
When Does Task-Aware Imputation Matter? An Analysis of Robustness Under Train-Test Mismatch
Maria V. Salvador, Ana F. Sequeira, Jaime S. Cardoso
Missing data is a pervasive challenge in machine learning, often degrading predictive performance and complicating model deployment. Most existing approaches treat imputation as a…
Reshaping Rollout Workloads for Asynchronous RL Post-Training on Heterogeneous Accelerators
Jiahui Li, Hao Nie, Yibo Zhu +5
Reinforcement learning (RL) post-training increasingly relies on long-horizon, multi-turn rollouts. As post-training jobs outgrow a single cluster, rollout pools assembled across c…
StepAudio 3 Realtime Technical Report
Bin Lin, Bo Zhao, Boyang Zhang +87
Realtime spoken interaction demands deep reasoning, prompt responses, and fluid turn-taking. We present StepAudio 3 Realtime, an audio-language foundation model organized around a…
Scheduling Mixed RL Rollouts Beyond Prefix Locality
Zetao Hong, Song Yuan, Yuanhao Ding +4
Modern reinforcement learning (RL) post-training pipelines for large language models (LLMs) increasingly combine rollout workloads across multiple domains and feedback paradigms. P…
TensorCast: The Missing Tensor Management Layer in Large Language Model Infrastructure
Yuhan Zhou, Yuchu Luo, Hao Nie +5
Modern LLM infrastructure increasingly manages tensors not only as computation data, but also as persistent states shared across distributed components. Existing systems optimize i…
Don't Predict, Prioritize: Rethinking GPU Reliability Assessment
Difeng Ma, Changhua Pei, Yuanwei Lu +7
The reliability of Graphics Processing Units (GPUs) is a criticalbottleneck for modern large-scale AI infrastructure, where a sin-gle node failure can disrupt synchronous training…
JetSpec: Breaking the Scaling Ceiling of Speculative Decoding with Parallel Tree Drafting
Lanxiang Hu, Zhaoxiang Feng, Yulun Wu +9
Speculative decoding (SD) accelerates autoregressive Large Language Models (LLMs) by drafting multiple tokens and verifying them in parallel, but it faces a scaling limitation: inc…
Frontier: Towards Comprehensive and Accurate LLM Inference Simulation
Yicheng Feng, Xin Tan, Yangtao Deng +3
Modern LLM serving is no longer homogeneous or monolithic. Production systems now combine disaggregated execution, complex parallelism, runtime optimizations, and stateful workload…
StepAudio 2.5 Technical Report
Bin Lin, Bo Zhao, Boyong Wu +98
Unified audio-language modeling has emerged as a prominent trend in modern speech systems, promising to bring the reasoning capabilities of large language models to auditory tasks.…
OrchestrRL: Dynamic Compute and Network Orchestration for Disaggregated RL
Xin Tan, Yicheng Feng, Yu Zhou +3
Disaggregating the generation and training stages in RL is widely adopted to scale LLM post-training. There are two critical challenges here. First, the generation stage often beco…
Efficient Long-context Language Model Training by Core Attention Disaggregation
Yonghao Zhuang, Junda Chen, Bo Pang +6
We present core attention disaggregation (CAD), a technique that improves long-context large language model training by decoupling the core attention computation, softmax(QK^T)V, f…
Frontier: Simulating the Next Generation of LLM Inference Systems
Yicheng Feng, Xin Tan, Kin Hang Sew +3
Large Language Model (LLM) inference is growing increasingly complex with the rise of Mixture-of-Experts (MoE) models and disaggregated architectures that decouple components like…
StreamRL: Scalable, Heterogeneous, and Elastic RL for LLMs with Disaggregated Stream Generation
Yinmin Zhong, Zili Zhang, Xiaoniu Song +11
Reinforcement learning (RL) has become the core post-training technique for large language models (LLMs). RL for LLMs involves two stages: generation and training. The LLM first ge…
DIP: Efficient Large Multimodal Model Training with Dynamic Interleaved Pipeline
Zhenliang Xue, Hanpeng Hu, Xing Chen +7
Large multimodal models (LMMs) have demonstrated excellent capabilities in both understanding and generation tasks with various modalities. While these models can accept flexible c…
DSV: Exploiting Dynamic Sparsity to Accelerate Large-Scale Video DiT Training
Xin Tan, Yuetao Chen, Yimin Jiang +6
Diffusion Transformers (DiTs) have shown remarkable performance in generating high-quality videos. However, the quadratic complexity of 3D full attention remains a bottleneck in sc…
InfiniteHBD: Building Datacenter-Scale High-Bandwidth Domain for LLM with Optical Circuit Switching Transceivers
Chenchen Shou, Guyue Liu, Hao Nie +11
Scaling Large Language Model (LLM) training relies on multi-dimensional parallelism, where High-Bandwidth Domains (HBDs) are critical for communication-intensive parallelism like T…
DistTrain: Addressing Model and Data Heterogeneity with Disaggregated Training for Multimodal Large Language Models
Zili Zhang, Yinmin Zhong, Yimin Jiang +6
Multimodal large language models (LLMs) empower LLMs to ingest inputs and generate outputs in multiple forms, such as text, image, and audio. However, the integration of multiple m…
CDMPP: A Device-Model Agnostic Framework for Latency Prediction of Tensor Programs
Hanpeng Hu, Junwei Su, Juntao Zhao +4
Deep Neural Networks (DNNs) have shown excellent performance in a wide range of machine learning applications. Knowing the latency of running a DNN model or tensor program on a spe…
Accelerating Distributed MoE Training and Inference with Lina
Jiamin Li, Yimin Jiang, Yibo Zhu +2
Scaling model parameters improves model quality at the price of high computation overhead. Sparsely activated models, usually in the form of Mixture of Experts (MoE) architecture,…
Aryl: An Elastic Cluster Scheduler for Deep Learning
Jiamin Li, Hong Xu, Yibo Zhu +3
Companies build separate training and inference GPU clusters for deep learning, and use separate schedulers to manage them. This leads to problems for both training and inference:…