arXiv subjects

Xiaoxuan Liu

Publications and source records attributed to Xiaoxuan Liu.

31 records · Page 2Linked to original sources

arXiv2024

MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs

Efficient deployment of large language models, particularly Mixture of Experts (MoE), on resource-constrained platforms presents significant challenges, especially in terms of computational efficiency and memory utilization. The MoE architecture, renowned for its ability to increase model capacity without a proportional increase in inference cost, greatly reduces the token generation latency compared with dense models. However, the large model size makes MoE models inaccessible to individuals without high-end GPUs. In this paper, we propose a high-throughput MoE batch inference system, that significantly outperforms past work. MoE-Lightning introduces a novel CPU-GPU-I/O pipelining schedule, CGOPipe, with paged weights to achieve high resource utilization, and a performance model, HRM, based on a Hierarchical Roofline Model we introduce to help find policies with higher throughput than existing systems. MoE-Lightning can achieve up to 10.3x higher throughput than state-of-the-art offloading-enabled LLM inference systems for Mixtral 8x7B on a single T4 GPU (16GB). When the theoretical system throughput is bounded by the GPU memory, MoE-Lightning can reach the throughput upper bound with 2-3x less CPU memory, significantly increasing resource utilization. MoE-Lightning also supports efficient batch inference for much larger MoEs (e.g., Mixtral 8x22B and DBRX) on multiple low-cost GPUs (e.g., 2-4 T4).

cs.DC↗

arXiv2024

UNetMamba: An Efficient UNet-Like Mamba for Semantic Segmentation of High-Resolution Remote Sensing Images

Semantic segmentation of high-resolution remote sensing images is vital in downstream applications such as land-cover mapping, urban planning and disaster assessment.Existing Transformer-based methods suffer from the constraint between accuracy and efficiency, while the recently proposed Mamba is renowned for being efficient. Therefore, to overcome the dilemma, we propose UNetMamba, a UNet-like semantic segmentation model based on Mamba. It incorporates a mamba segmentation decoder (MSD) that can efficiently decode the complex information within high-resolution images, and a local supervision module (LSM), which is train-only but can significantly enhance the perception of local contents. Extensive experiments demonstrate that UNetMamba outperforms the state-of-the-art methods with mIoU increased by 0.87% on LoveDA and 0.39% on ISPRS Vaihingen, while achieving high efficiency through the lightweight design, less memory footprint and reduced computational cost. The source code is available at https://github.com/EnzeZhu2001/UNetMamba.

cs.CV↗

arXiv2024

Mélange: Cost Efficient Large Language Model Serving by Exploiting GPU Heterogeneity

Large language models (LLMs) are increasingly integrated into many online services, yet they remain cost-prohibitive to deploy due to the requirement of expensive GPU instances. Prior work has addressed the high cost of LLM serving by improving the inference engine, but less attention has been given to selecting the most cost-efficient GPU type(s) for a specific LLM service. There is a large and growing landscape of GPU types and, within these options, higher cost does not always lead to increased performance. Instead, through a comprehensive investigation, we find that three key LLM service characteristics (request size, request rate, SLO) strongly influence GPU cost efficiency, and differing GPU types are most cost efficient for differing LLM service settings. As a result, the most cost-efficient allocation for a given service is typically a mix of heterogeneous GPU types. Based on this analysis, we introduce Mélange, a GPU allocation framework that navigates these diverse LLM service characteristics and heterogeneous GPU option space to automatically and efficiently derive the minimal-cost GPU allocation for a given LLM service. We formulate the GPU allocation task as a cost-aware bin packing problem where GPUs are bins and items are slices of the service workload. Our formulation's constraints account for a service's unique characteristics, allowing Mélange to be flexible to support diverse service settings and heterogeneity-aware to adapt the GPU allocation to a specific service. Compared to using only a single GPU type, Mélange reduces deployment costs by up to 77% in conversational settings, 33% in document-based settings, and 51% in a mixed setting.

cs.DC↗

arXiv2024

Learned Best-Effort LLM Serving

Many applications must provide low-latency LLM service to users or risk unacceptable user experience. However, over-provisioning resources to serve fluctuating request patterns is often prohibitively expensive. In this work, we present a best-effort serving system that employs deep reinforcement learning to adjust service quality based on the task distribution and system load. Our best-effort system can maintain availability with over 10x higher client request rates, serves above 96% of peak performance 4.1x more often, and serves above 98% of peak performance 2.3x more often than static serving on unpredictable workloads. Our learned router is robust to shifts in both the arrival and task distribution. Compared to static serving, learned best-effort serving allows for cost-efficient serving through increased hardware utility. Additionally, we argue that learned best-effort LLM serving is applicable in wide variety of settings and provides application developers great flexibility to meet their specific needs.

cs.LG↗

arXiv2024

Online Speculative Decoding

Speculative decoding is a pivotal technique to accelerate the inference of large language models (LLMs) by employing a smaller draft model to predict the target model's outputs. However, its efficacy can be limited due to the low predictive accuracy of the draft model, particularly when faced with diverse text inputs and a significant capability gap between the draft and target models. We introduce online speculative decoding to address this challenge. The main idea is to continuously update the (multiple) draft model(s) on observed user query data. Adapting to query distribution mitigates the shifts between the training distribution of the draft model and the query distribution, enabling the draft model to more accurately predict the target model's outputs. We develop a prototype of online speculative decoding based on knowledge distillation and evaluate it using both synthetic and real query data. The results show a substantial increase in the token acceptance rate by 0.1 to 0.65, bringing 1.42x to 2.17x latency reduction. Our code is available at https://github.com/LiuXiaoxuanPKU/OSD.

cs.AI↗

arXiv2024

Towards Clinical AI Fairness: Filling Gaps in the Puzzle

The ethical integration of Artificial Intelligence (AI) in healthcare necessitates addressing fairness-a concept that is highly context-specific across medical fields. Extensive studies have been conducted to expand the technical components of AI fairness, while tremendous calls for AI fairness have been raised from healthcare. Despite this, a significant disconnect persists between technical advancements and their practical clinical applications, resulting in a lack of contextualized discussion of AI fairness in clinical settings. Through a detailed evidence gap analysis, our review systematically pinpoints several deficiencies concerning both healthcare data and the provided AI fairness solutions. We highlight the scarcity of research on AI fairness in many medical domains where AI technology is increasingly utilized. Additionally, our analysis highlights a substantial reliance on group fairness, aiming to ensure equality among demographic groups from a macro healthcare system perspective; in contrast, individual fairness, focusing on equity at a more granular level, is frequently overlooked. To bridge these gaps, our review advances actionable strategies for both the healthcare and AI research communities. Beyond applying existing AI fairness methods in healthcare, we further emphasize the importance of involving healthcare professionals to refine AI fairness concepts and methods to ensure contextually relevant and ethically sound AI applications in healthcare.

cs.AI↗

arXiv2024

Computing in the Era of Large Generative Models: From Cloud-Native to AI-Native

In this paper, we investigate the intersection of large generative AI models and cloud-native computing architectures. Recent large models such as ChatGPT, while revolutionary in their capabilities, face challenges like escalating costs and demand for high-end GPUs. Drawing analogies between large-model-as-a-service (LMaaS) and cloud database-as-a-service (DBaaS), we describe an AI-native computing paradigm that harnesses the power of both cloud-native technologies (e.g., multi-tenancy and serverless computing) and advanced machine learning runtime (e.g., batched LoRA inference). These joint efforts aim to optimize costs-of-goods-sold (COGS) and improve resource accessibility. The journey of merging these two domains is just at the beginning and we hope to stimulate future research and development in this area.

cs.DC↗

arXiv2024

The Jet Composition of GRB 230307A: Poynting-Flux-Dominated Outflow?

The jet composition of GRB plays an important role in understanding the energy dissipation and radiation mechanisms in GRB physics, but it is poorly constrained from the observational data. Recently, an interesting long-duration GRB 230307A with redshift $z=$0.065 has attracted great attention. The lack of detected thermal emission and mini-structure of prompt emission lightcurve of this burst suggest that the outflow is Poynting-flux-dominated and point towards the ICMART model. In this paper, we invoke two independent methods to investigate the jet composition of GRB 230307A. The high magnetization parameter ($σ>7$ or ever large) for$R_0=10^{10}$ cm that is used to suppress thermal component, strongly suggests that a significant fraction of the outflow energy is likely in a Poynting flux entrained with the baryonic matter. Moreover, it is found that the radiation efficiency of this burst for typical values $ε_e=0.1$ and $ε_B=0.01$ can reach as high as $~50\%$ which disfavors the internal shock model, but is consistent with ICMART model. Finally, a possible unified picture to produce GRB 230307A originated from a compact star merger is also discussed.

astro-ph.HE↗

arXiv2023

An Evaluation of Memory Optimization Methods for Training Neural Networks

As models continue to grow in size, the development of memory optimization methods (MOMs) has emerged as a solution to address the memory bottleneck encountered when training large models. To comprehensively examine the practical value of various MOMs, we have conducted a thorough analysis of existing literature from a systems perspective. Our analysis has revealed a notable challenge within the research community: the absence of standardized metrics for effectively evaluating the efficacy of MOMs. The scarcity of informative evaluation metrics hinders the ability of researchers and practitioners to compare and benchmark different approaches reliably. Consequently, drawing definitive conclusions and making informed decisions regarding the selection and application of MOMs becomes a challenging endeavor. To address the challenge, this paper summarizes the scenarios in which MOMs prove advantageous for model training. We propose the use of distinct evaluation metrics under different scenarios. By employing these metrics, we evaluate the prevailing MOMs and find that their benefits are not universal. We present insights derived from experiments and discuss the circumstances in which they can be advantageous.

cs.LG↗

arXiv2022

Leveraging Application Data Constraints to Optimize Database-Backed Web Applications

Exploiting the relationships among data is a classical query optimization technique. As persistent data is increasingly being created and maintained programmatically, prior work that infers data relationships from data statistics misses an important opportunity. We present ConstrOpt, the first tool that identifies data relationships by analyzing database-backed applications. Once identified, ConstrOpt leverages the constraints to optimize the application's physical design and query execution. Instead of developing a fixed set of predefined rewriting rules, ConstrOpt employs an enumerate-test-verify technique to automatically exploit the discovered data constraints to improve query execution. Each resulting rewrite is provably equivalent to the original query. Using 14 real-world web applications, our experiments show that ConstrOpt can discover numerous data constraints from code analysis and improve real-world application performance significantly.

cs.DB↗

arXiv2022

GACT: Activation Compressed Training for Generic Network Architectures

Training large neural network (NN) models requires extensive memory resources, and Activation Compressed Training (ACT) is a promising approach to reduce training memory footprint. This paper presents GACT, an ACT framework to support a broad range of machine learning tasks for generic NN architectures with limited domain knowledge. By analyzing a linearized version of ACT's approximate gradient, we prove the convergence of GACT without prior knowledge on operator type or model architecture. To make training stable, we propose an algorithm that decides the compression ratio for each tensor by estimating its impact on the gradient at run time. We implement GACT as a PyTorch library that readily applies to any NN architecture. GACT reduces the activation memory for convolutional NNs, transformers, and graph NNs by up to 8.1x, enabling training with a 4.2x to 24.7x larger batch size, with negligible accuracy loss. We implement GACT as a PyTorch library at https://github.com/LiuXiaoxuanPKU/GACT-ICML.

cs.LG↗

arXiv2022

Long-run User Value Optimization in Recommender Systems through Content Creation Modeling

Content recommender systems are generally adept at maximizing immediate user satisfaction but to optimize for the \textit{long-run} user value, we need more statistically sophisticated solutions than off-the-shelf simple recommender algorithms. In this paper we lay out such a solution to optimize \textit{long-run} user value through discounted utility maximization and a machine learning method we have developed for estimating it. Our method estimates which content producers are most likely to create the highest long-run user value if their content is shown more to users who enjoy it in the present. We do this estimation with the help of an A/B test and heterogeneous effects machine learning model. We have used such models in Facebook's feed ranking system, and such a model can be used in other recommender systems.

cs.IR↗

arXiv2020

Order-Preserving Key Compression for In-Memory Search Trees

We present the High-speed Order-Preserving Encoder (HOPE) for in-memory search trees. HOPE is a fast dictionary-based compressor that encodes arbitrary keys while preserving their order. HOPE's approach is to identify common key patterns at a fine granularity and exploit the entropy to achieve high compression rates with a small dictionary. We first develop a theoretical model to reason about order-preserving dictionary designs. We then select six representative compression schemes using this model and implement them in HOPE. These schemes make different trade-offs between compression rate and encoding speed. We evaluate HOPE on five data structures used in databases: SuRF, ART, HOT, B+tree, and Prefix B+tree. Our experiments show that using HOPE allows the search trees to achieve lower query latency (up to 40\% lower) and better memory efficiency (up to 30\% smaller) simultaneously for most string key workloads.

cs.DB↗