arXiv subjects

Xinglin Pan

Publications and source records attributed to Xinglin Pan.

At least 19 recordsLinked to original sources

arXiv2026

Compass: Dissecting Communication and Computation Operators for Efficient LLM Training

Overlapping communication and computation operators is a common practice to hide communication overheads, accelerating large language models (LLMs) training on GPU clusters. Existing systems achieve this through either intra-operator fusion (IntraFusion), which packs operators into a single large kernel, or inter-operator decomposition (InterDecom), which splits a tensor into multiple parts for pipelined execution. However, current IntraFusion methods underutilize network topology, causing suboptimal bandwidth usage on multi-GPU systems, while InterDecom struggles to determine the optimal number of decomposed parts for peak performance. To address these issues, we introduce Compass, which employs systematic optimization and comprehensive modeling. First, we design a novel IntraFusion algorithm leveraging double-ring communications to maximize bandwidth utilization in hybrid NVLink-PCIe systems, achieving 1.5x-2.5x speedups. Second, we develop a decomposition model that mathematically derives the optimal tensor decomposition degree for InterDecom, improving performance by up to 1.3x. Finally, we develop a unified performance framework that accurately determines the best strategy for different scenarios. We validate Compass through extensive evaluation across 288 configurations and end-to-end experiments on real-world applications. The results demonstrate that Compass consistently selects the optimal strategy, achieving up to a 1.42x end-to-end speedup compared to the Megatron-LM baseline.

cs.PF↗

arXiv2026

ZipCCL: Efficient Lossless Data Compression of Communication Collectives for Accelerating LLM Training

Communication has emerged as a critical bottleneck in the distributed training of large language models (LLMs). While numerous approaches have been proposed to reduce communication overhead, the potential of lossless compression has remained largely underexplored since compression and decompression typically consume larger overheads than the benefits of reduced communication traffic. We observe that the communication data, including activations, gradients and parameters, during training often follows a near-Gaussian distribution, which is a key feature for data compression. Thus, we introduce ZipCCL, a lossless compressed communication library of collectives for LLM training. ZipCCL is equipped with our novel techniques: (1) theoretically grounded exponent coding that exploits the Gaussian distribution of LLM tensors to accelerate compression without expensive online statistics, (2) GPU-optimized compression and decompression kernels that carefully design memory access patterns and pipeline using communication-aware data layout, and (3) adaptive communication strategies that dynamically switch collective operations based on workload patterns and system characteristics. Evaluated on a 64-GPU cluster using both mixture-of-experts and dense transformer models, ZipCCL reduces communication time by up to 1.35$\times$ and achieves end-to-end training speedups of up to 1.18$\times$ without any impact on model quality.

cs.DC↗

arXiv2026

ZipServ: Fast and Memory-Efficient LLM Inference with Hardware-Aware Lossless Compression

Lossless model compression holds tremendous promise for alleviating the memory and bandwidth bottlenecks in bit-exact Large Language Model (LLM) serving. However, existing approaches often result in substantial inference slowdowns due to fundamental design mismatches with GPU architectures: at the kernel level, variable-length bitstreams produced by traditional entropy codecs break SIMT parallelism; at the system level, decoupled pipelines lead to redundant memory traffic. We present ZipServ, a lossless compression framework co-designed for efficient LLM inference. ZipServ introduces Tensor-Core-Aware Triple Bitmap Encoding (TCA-TBE), a novel fixed-length format that enables constant-time, parallel decoding, together with a fused decompression-GEMM (ZipGEMM) kernel that decompresses weights on-the-fly directly into Tensor Core registers. This "load-compressed, compute-decompressed" design eliminates intermediate buffers and maximizes compute intensity. Experiments show that ZipServ reduces the model size by up to 30%, achieves up to 2.21x kernel-level speedup over NVIDIA's cuBLAS, and expedites end-to-end inference by an average of 1.22x over vLLM. ZipServ is the first lossless compression system that provides both storage savings and substantial acceleration for LLM inference on GPUs.

cs.DC↗

arXiv2025

Efficient MoE Inference with Fine-Grained Scheduling of Disaggregated Expert Parallelism

The mixture-of-experts (MoE) architecture scales model size with sublinear computational increase but suffers from memory-intensive inference due to KV caches and sparse expert activation. Recent disaggregated expert parallelism (DEP) distributes attention and experts to dedicated GPU groups but lacks support for shared experts and efficient task scheduling, limiting performance. We propose FinDEP, a fine-grained task scheduling algorithm for DEP that maximizes task overlap to improve MoE inference throughput. FinDEP introduces three innovations: 1) partitioning computation/communication into smaller tasks for fine-grained pipelining, 2) formulating a scheduling optimization supporting variable granularity and ordering, and 3) developing an efficient solver for this large search space. Experiments on four GPU systems with DeepSeek-V2 and Qwen3-MoE show FinDEP improves throughput by up to 1.61x over prior methods, achieving up to 1.24x speedup on a 32-GPU system.

cs.DC↗

arXiv2025

HierMoE: Accelerating MoE Training with Hierarchical Token Deduplication and Expert Swap

The sparsely activated mixture-of-experts (MoE) transformer has become a common architecture for large language models (LLMs) due to its sparsity, which requires fewer computational demands while easily scaling the model size. In MoE models, each MoE layer requires to dynamically choose tokens to activate particular experts for computation while the activated experts may not be located in the same device or GPU as the token. However, this leads to substantial communication and load imbalances across all GPUs, which obstructs the scalability of distributed systems within a GPU cluster. To this end, we introduce HierMoE to accelerate the training of MoE models by two topology-aware techniques: 1) token deduplication to reduce the communication traffic, and 2) expert swap to balance the workloads among all GPUs. To enable the above two proposed approaches to be more general, we build theoretical models aimed at achieving the best token duplication and expert swap strategy under different model configurations and hardware environments. We implement our prototype HierMoE system atop Megatron-LM and conduct experiments on a 32-GPU cluster with DeepSeek-V3 and Qwen3-30B-A3B models. Experimental results show that our HierMoE achieves $1.55\times$ to $3.32\times$ faster communication and delivers $1.18\times$ to $1.27\times$ faster end-to-end training compared to state-of-the-art MoE training systems, Tutel-2DH, SmartMoE, and Megatron-LM.

cs.DC↗

arXiv2025

DreamDDP: Accelerating Data Parallel Distributed LLM Training with Layer-wise Scheduled Partial Synchronization

The growth of large language models (LLMs) increases challenges of accelerating distributed training across multiple GPUs in different data centers. Moreover, concerns about data privacy and data exhaustion have heightened interest in geo-distributed data centers. Communication in geo-distributed data parallel training (DDP) with stochastic gradient descent (S-SGD) is the main bottleneck in low-bandwidth environments. Local SGD mitigates communication overhead by reducing synchronization frequency, and recent studies have successfully applied it to geo-distributedly pre-train LLMs. However, we identify that its model synchronization mechanism prevents overlapping communication and computation, which makes the system lose opportunities to overlap communication and computation. To overcome this limitation, we expand the design space of local SGD by layer-wisely decoupling model synchronization. In each iteration, only some layers are synchronized instead of the entire model after a specific number of iterations. Leveraging this methodology, we introduce DreamDDP, a training framework to accelerate low-bandwidth distributed training with three key innovations: (1) partial local SGD with theoretical assurances of convergence rates comparable to S-SGD; (2) overlapping parameter synchronization with computation without extra GPU memory occupation; (3) identifying and exploiting three properties to schedule the communication and computation to reduce the training time based on fine-grained profiling of layer-wise communication and computation time. Empirical evaluations conducted on 32 GPUs using prominent deep learning models, including ResNet-18, ResNet-50, GPT-2, and Llama-2, demonstrate that DreamDDP enhances the convergence properties of Local SGD (and Adam) and achieves speedups ranging from $1.49\times$ to $3.91\times$ over leading baseline methods.

cs.DC↗

arXiv2025

Mediator: Memory-efficient LLM Merging with Less Parameter Conflicts and Uncertainty Based Routing

Model merging aggregates Large Language Models (LLMs) finetuned on different tasks into a stronger one. However, parameter conflicts between models leads to performance degradation in averaging. While model routing addresses this issue by selecting individual models during inference, it imposes excessive storage and compute costs, and fails to leverage the common knowledge from different models. In this work, we observe that different layers exhibit varying levels of parameter conflicts. Building on this insight, we average layers with minimal parameter conflicts and use a novel task-level expert routing for layers with significant conflicts. To further reduce storage costs, inspired by task arithmetic sparsity, we decouple multiple fine-tuned experts into a dense expert and several sparse experts. Considering the out-of-distribution samples, we select and merge appropriate experts based on the task uncertainty of the input data. We conduct extensive experiments on both LLaMA and Qwen with varying parameter scales, and evaluate on real-world reasoning tasks. Results demonstrate that our method consistently achieves significant performance improvements while requiring less system cost.

cs.LG↗

arXiv2025

FSMoE: A Flexible and Scalable Training System for Sparse Mixture-of-Experts Models

Recent large language models (LLMs) have tended to leverage sparsity to reduce computations, employing the sparsely activated mixture-of-experts (MoE) technique. MoE introduces four modules, including token routing, token communication, expert computation, and expert parallelism, that impact model quality and training efficiency. To enable versatile usage of MoE models, we introduce FSMoE, a flexible training system optimizing task scheduling with three novel techniques: 1) Unified abstraction and online profiling of MoE modules for task scheduling across various MoE implementations. 2) Co-scheduling intra-node and inter-node communications with computations to minimize communication overheads. 3) To support near-optimal task scheduling, we design an adaptive gradient partitioning method for gradient aggregation and a schedule to adaptively pipeline communications and computations. We conduct extensive experiments with configured MoE layers and real-world MoE models on two GPU clusters. Experimental results show that 1) our FSMoE supports four popular types of MoE routing functions and is more efficient than existing implementations (with up to a 1.42$\times$ speedup), and 2) FSMoE outperforms the state-of-the-art MoE training systems (DeepSpeed-MoE and Tutel) by 1.18$\times$-1.22$\times$ on 1458 MoE layers and 1.19$\times$-3.01$\times$ on real-world MoE models based on GPT-2 and Mixtral using a popular routing function.

cs.LG↗

arXiv2024

Task Scheduling for Efficient Inference of Large Language Models on Single Moderate GPU Systems

Large language models~(LLMs) are known for their high demand on computing resources and memory due to their substantial model size, which leads to inefficient inference on moderate GPU systems. Techniques like quantization or pruning can shrink model sizes but often impair accuracy, making them unsuitable for practical applications. In this work, we introduce \modelname{}, a high-performance inference engine designed to speed up LLM inference without compromising model accuracy. \modelname{} incorporates three innovative methods to increase inference efficiency: 1) model partitioning to allow asynchronous processing of tasks across CPU computation, GPU computation, and CPU-GPU communication, 2) an adaptive partition algorithm to optimize the use of CPU, GPU, and PCIe communication capabilities, and 3) a token assignment strategy to handle diverse prompt and generation tasks during LLM inference. Comprehensive experiments were conducted with various LLMs such as Mixtral, LLaMA-2, Qwen, and PhiMoE across three test environments featuring different CPUs and GPUs. The experimental findings demonstrate that \modelname{} achieves speeds between $1.11\times$ to $1.80\times$ faster in decoding and $1.69\times$ to $6.33\times$ faster in pre-filling, leading to an overall speedup ranging from $1.25\times$ to $2.04\times$ compared to state-of-the-art solutions, llama.cpp and Fiddler.

cs.CE↗

arXiv2024

Should We Really Edit Language Models? On the Evaluation of Edited Language Models

Model editing has become an increasingly popular alternative for efficiently updating knowledge within language models. Current methods mainly focus on reliability, generalization, and locality, with many methods excelling across these criteria. Some recent works disclose the pitfalls of these editing methods such as knowledge distortion or conflict. However, the general abilities of post-edited language models remain unexplored. In this paper, we perform a comprehensive evaluation on various editing methods and different language models, and have following findings. (1) Existing editing methods lead to inevitable performance deterioration on general benchmarks, indicating that existing editing methods maintain the general abilities of the model within only a few dozen edits. When the number of edits is slightly large, the intrinsic knowledge structure of the model is disrupted or even completely damaged. (2) Instruction-tuned models are more robust to editing, showing less performance drop on general knowledge after editing. (3) Language model with large scale is more resistant to editing compared to small model. (4) The safety of the edited model, is significantly weakened, even for those safety-aligned models. Our findings indicate that current editing methods are only suitable for small-scale knowledge updates within language models, which motivates further research on more practical and reliable editing methods. The details of code and reproduction can be found in https://github.com/lqinfdim/EditingEvaluation.

cs.AI↗

arXiv2024

FusionLLM: A Decentralized LLM Training System on Geo-distributed GPUs with Adaptive Compression

To alleviate hardware scarcity in training large deep neural networks (DNNs), particularly large language models (LLMs), we present FusionLLM, a decentralized training system designed and implemented for training DNNs using geo-distributed GPUs across different computing clusters or individual devices. Decentralized training faces significant challenges regarding system design and efficiency, including: 1) the need for remote automatic differentiation (RAD), 2) support for flexible model definitions and heterogeneous software, 3) heterogeneous hardware leading to low resource utilization or the straggler problem, and 4) slow network communication. To address these challenges, in the system design, we represent the model as a directed acyclic graph of operators (OP-DAG). Each node in the DAG represents the operator in the DNNs, while the edge represents the data dependency between operators. Based on this design, 1) users are allowed to customize any DNN without caring low-level operator implementation; 2) we enable the task scheduling with the more fine-grained sub-tasks, offering more optimization space; 3) a DAG runtime executor can implement RAD withour requiring the consistent low-level ML framework versions. To enhance system efficiency, we implement a workload estimator and design an OP-Fence scheduler to cluster devices with similar bandwidths together and partition the DAG to increase throughput. Additionally, we propose an AdaTopK compressor to adaptively compress intermediate activations and gradients at the slowest communication links. To evaluate the convergence and efficiency of our system and algorithms, we train ResNet-101 and GPT-2 on three real-world testbeds using 48 GPUs connected with 8 Mbps~10 Gbps networks. Experimental results demonstrate that our system and method can achieve 1.45 - 9.39x speedup compared to baseline methods while ensuring convergence.

cs.DC↗

arXiv2024

Multi-Task Domain Adaptation for Language Grounding with 3D Objects

The existing works on object-level language grounding with 3D objects mostly focus on improving performance by utilizing the off-the-shelf pre-trained models to capture features, such as viewpoint selection or geometric priors. However, they have failed to consider exploring the cross-modal representation of language-vision alignment in the cross-domain field. To answer this problem, we propose a novel method called Domain Adaptation for Language Grounding (DA4LG) with 3D objects. Specifically, the proposed DA4LG consists of a visual adapter module with multi-task learning to realize vision-language alignment by comprehensive multimodal feature representation. Experimental results demonstrate that DA4LG competitively performs across visual and non-visual language descriptions, independent of the completeness of observation. DA4LG achieves state-of-the-art performance in the single-view setting and multi-view setting with the accuracy of 83.8% and 86.8% respectively in the language grounding benchmark SNARE. The simulation experiments show the well-practical and generalized performance of DA4LG compared to the existing methods. Our project is available at https://sites.google.com/view/da4lg.

cs.CV↗

arXiv2024

Parm: Efficient Training of Large Sparsely-Activated Models with Dedicated Schedules

Sparsely-activated Mixture-of-Expert (MoE) layers have found practical applications in enlarging the model size of large-scale foundation models, with only a sub-linear increase in computation demands. Despite the wide adoption of hybrid parallel paradigms like model parallelism, expert parallelism, and expert-sharding parallelism (i.e., MP+EP+ESP) to support MoE model training on GPU clusters, the training efficiency is hindered by communication costs introduced by these parallel paradigms. To address this limitation, we propose Parm, a system that accelerates MP+EP+ESP training by designing two dedicated schedules for placing communication tasks. The proposed schedules eliminate redundant computations and communications and enable overlaps between intra-node and inter-node communications, ultimately reducing the overall training time. As the two schedules are not mutually exclusive, we provide comprehensive theoretical analyses and derive an automatic and accurate solution to determine which schedule should be applied in different scenarios. Experimental results on an 8-GPU server and a 32-GPU cluster demonstrate that Parm outperforms the state-of-the-art MoE training system, DeepSpeed-MoE, achieving 1.13$\times$ to 5.77$\times$ speedup on 1296 manually configured MoE layers and approximately 3$\times$ improvement on two real-world MoE models based on BERT and GPT-2.

cs.DC↗

arXiv2024

Pruner-Zero: Evolving Symbolic Pruning Metric from scratch for Large Language Models

Despite the remarkable capabilities, Large Language Models (LLMs) face deployment challenges due to their extensive size. Pruning methods drop a subset of weights to accelerate, but many of them require retraining, which is prohibitively expensive and computationally demanding. Recently, post-training pruning approaches introduced novel metrics, enabling the pruning of LLMs without retraining. However, these metrics require the involvement of human experts and tedious trial and error. To efficiently identify superior pruning metrics, we develop an automatic framework for searching symbolic pruning metrics using genetic programming. In particular, we devise an elaborate search space encompassing the existing pruning metrics to discover the potential symbolic pruning metric. We propose an opposing operation simplification strategy to increase the diversity of the population. In this way, Pruner-Zero allows auto-generation of symbolic pruning metrics. Based on the searched results, we explore the correlation between pruning metrics and performance after pruning and summarize some principles. Extensive experiments on LLaMA and LLaMA-2 on language modeling and zero-shot tasks demonstrate that our Pruner-Zero obtains superior performance than SOTA post-training pruning methods. Code at: \url{https://github.com/pprp/Pruner-Zero}.

cs.LG↗

arXiv2024

ParZC: Parametric Zero-Cost Proxies for Efficient NAS

Recent advancements in Zero-shot Neural Architecture Search (NAS) highlight the efficacy of zero-cost proxies in various NAS benchmarks. Several studies propose the automated design of zero-cost proxies to achieve SOTA performance but require tedious searching progress. Furthermore, we identify a critical issue with current zero-cost proxies: they aggregate node-wise zero-cost statistics without considering the fact that not all nodes in a neural network equally impact performance estimation. Our observations reveal that node-wise zero-cost statistics significantly vary in their contributions to performance, with each node exhibiting a degree of uncertainty. Based on this insight, we introduce a novel method called Parametric Zero-Cost Proxies (ParZC) framework to enhance the adaptability of zero-cost proxies through parameterization. To address the node indiscrimination, we propose a Mixer Architecture with Bayesian Network (MABN) to explore the node-wise zero-cost statistics and estimate node-specific uncertainty. Moreover, we propose DiffKendall as a loss function to directly optimize Kendall's Tau coefficient in a differentiable manner so that our ParZC can better handle the discrepancies in ranking architectures. Comprehensive experiments on NAS-Bench-101, 201, and NDS demonstrate the superiority of our proposed ParZC compared to existing zero-shot NAS methods. Additionally, we demonstrate the versatility and adaptability of ParZC by transferring it to the Vision Transformer search space.

cs.CV↗

arXiv2023

Dissecting the Runtime Performance of the Training, Fine-tuning, and Inference of Large Language Models

Large Language Models (LLMs) have seen great advance in both academia and industry, and their popularity results in numerous open-source frameworks and techniques in accelerating LLM pre-training, fine-tuning, and inference. Training and deploying LLMs are expensive as it requires considerable computing resources and memory, hence many efficient approaches have been developed for improving system pipelines as well as operators. However, the runtime performance can vary significantly across hardware and software stacks, which makes it difficult to choose the best configuration. In this work, we aim to benchmark the performance from both macro and micro perspectives. First, we benchmark the end-to-end performance of pre-training, fine-tuning, and serving LLMs in different sizes , i.e., 7, 13, and 70 billion parameters (7B, 13B, and 70B) on three 8-GPU platforms with and without individual optimization techniques, including ZeRO, quantization, recomputation, FlashAttention. Then, we dive deeper to provide a detailed runtime analysis of the sub-modules, including computing and communication operators in LLMs. For end users, our benchmark and findings help better understand different optimization techniques, training and inference frameworks, together with hardware platforms in choosing configurations for deploying LLMs. For researchers, our in-depth module-wise analyses discover potential opportunities for future work to further optimize the runtime performance of LLMs.

cs.PF↗

arXiv2023

Generalized Category Discovery with Clustering Assignment Consistency

Generalized category discovery (GCD) is a recently proposed open-world task. Given a set of images consisting of labeled and unlabeled instances, the goal of GCD is to automatically cluster the unlabeled samples using information transferred from the labeled dataset. The unlabeled dataset comprises both known and novel classes. The main challenge is that unlabeled novel class samples and unlabeled known class samples are mixed together in the unlabeled dataset. To address the GCD without knowing the class number of unlabeled dataset, we propose a co-training-based framework that encourages clustering consistency. Specifically, we first introduce weak and strong augmentation transformations to generate two sufficiently different views for the same sample. Then, based on the co-training assumption, we propose a consistency representation learning strategy, which encourages consistency between feature-prototype similarity and clustering assignment. Finally, we use the discriminative embeddings learned from the semi-supervised representation learning process to construct an original sparse network and use a community detection method to obtain the clustering results and the number of categories simultaneously. Extensive experiments show that our method achieves state-of-the-art performance on three generic benchmarks and three fine-grained visual recognition datasets. Especially in the ImageNet-100 data set, our method significantly exceeds the best baseline by 15.5\% and 7.0\% on the \texttt{Novel} and \texttt{All} classes, respectively.

cs.CV↗

arXiv2023

Fault-Tolerant Hybrid-Parallel Training at Scale with Reliable and Efficient In-memory Checkpointing

To efficiently scale large model (LM) training, researchers transition from data parallelism (DP) to hybrid parallelism (HP) on GPU clusters, which frequently experience hardware and software failures. Existing works introduce in-memory checkpointing optimizations that snapshot parameters to device memory for rapid failure recovery. However, these methods introduce severe resource competition between checkpointing and training, which can work under DP but can hardly scale under resource-intensive HP. To ensure low checkpointing overhead for hybrid-parallel training, this paper introduces a distributed in-memory checkpointing system with near-zero in-memory saving overhead. It strives from two aspects to mitigate the on-host resource competition caused by in-memory checkpointing: (1) It introduces Hierarchical Asynchronous Snapshotting Coordination in the checkpoint saving stage. This approach uses three-level asynchronous on-device scheduling to enhance parallelism between snapshotting and training, thereby minimizing snapshotting overhead. (2) It proposes Hybrid In-memory Checkpoint Protection to enhance checkpoint completeness during hardware failures. Unlike methods that require inter-node communications, which may block training under HP, it creates intra-node redundancy with efficient resource utilization, protecting training against hardware failures with minimal overhead. With these methods, this work enables fast restart for failed HP training with Distributed In-memory Checkpoint Loading, bypassing inefficiencies in NFS reads. In our evaluation, we achieve zero in-memory checkpoint saving overhead on Frontier while training Llama-2-34B on 256 MI250X devices (512 GPUs).

cs.DC↗