PyTorchKR Papers

최신 논문

총 3,273편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2505.10117

Learning Virtual Machine Scheduling in Cloud Computing through Language Agents

In cloud services, virtual machine (VM) scheduling is a typical Online Dynamic Multidimensional Bin Packing (ODMBP) problem, characterized by large-scale complexity and fluctuating demands. Traditional optimization methods struggle to adapt to real-time changes, domain-expert-designed heuristic approaches suffer from rigid strategies, and existing learning-based methods often lack generalizability and interpretability. To address these limitations, this paper proposes a hierarchical language agent framework named MiCo, which provides a large language model (LLM)-driven heuristic design paradigm for solving ODMBP. Specifically, ODMBP is formulated as a Semi-Markov Decision Process with Options (SMDP-Option), enabling dynamic scheduling through a two-stage architecture, i.e., Option Miner and Option Composer. Option Miner utilizes LLMs to discover diverse and useful non-context-aware strategies by interacting with constructed environments. Option Composer employs LLMs to discover a composing strategy that integrates the non-context-aware strategies with the contextual ones. Extensive experiments on real-world enterprise datasets demonstrate that MiCo achieves a 96.9\% competitive ratio in large-scale scenarios involving more than 10,000 virtual machines. It maintains high performance even under nonstationary request flows and diverse configurations, thus validating its effectiveness in complex and large-scale cloud environments.

arXiv:2505.09792

Interim Report on Human-Guided Adaptive Hyperparameter Optimization with Multi-Fidelity Sprints

This case study applies a phased hyperparameter optimization process to compare multitask natural language model variants that utilize multiphase learning rate scheduling and optimizer parameter grouping. We employ short, Bayesian optimization sessions that leverage multi-fidelity, hyperparameter space pruning, progressive halving, and a degree of human guidance. We utilize the Optuna TPE sampler and Hyperband pruner, as well as the Scikit-Learn Gaussian process minimization. Initially, we use efficient low-fidelity sprints to prune the hyperparameter space. Subsequent sprints progressively increase their model fidelity and employ hyperband pruning for efficiency. A second aspect of our approach is using a meta-learner to tune threshold values to resolve classification probabilities during inference. We demonstrate our method on a collection of variants of the 2021 Joint Entity and Relation Extraction model proposed by Eberts and Ulges.

arXiv:2505.10468

AI Agents vs. Agentic AI: A Conceptual Taxonomy, Applications and Challenges

This study critically distinguishes between AI Agents and Agentic AI, offering a structured conceptual taxonomy, application mapping, and challenge analysis to clarify their divergent design philosophies and capabilities. We begin by outlining the search strategy and foundational definitions, characterizing AI Agents as modular systems driven by Large Language Models (LLMs) and Large Image Models (LIMs) for narrow, task-specific automation. Generative AI is positioned as a precursor, with AI Agents advancing through tool integration, prompt engineering, and reasoning enhancements. In contrast, Agentic AI systems represent a paradigmatic shift marked by multi-agent collaboration, dynamic task decomposition, persistent memory, and orchestrated autonomy. Through a sequential evaluation of architectural evolution, operational mechanisms, interaction styles, and autonomy levels, we present a comparative analysis across both paradigms. Application domains such as customer support, scheduling, and data summarization are contrasted with Agentic AI deployments in research automation, robotic coordination, and medical decision support. We further examine unique challenges in each paradigm including hallucination, brittleness, emergent behavior, and coordination failure and propose targeted solutions such as ReAct loops, RAG, orchestration layers, and causal modeling. This work aims to provide a definitive roadmap for developing robust, scalable, and explainable AI agent and Agentic AI-driven systems. >AI Agents, Agent-driven, Vision-Language-Models, Agentic AI Decision Support System, Agentic-AI Applications

arXiv:2505.10328공식 코드 · Papers with Code

A Comparative Study of SMT and MILP for the Nurse Rostering Problem

The effects of personnel scheduling on the quality of care and working conditions for healthcare personnel have been thoroughly documented. However, the ever-present demand and large variation of constraints make healthcare scheduling particularly challenging. This problem has been studied for decades, with limited research aimed at applying Satisfiability Modulo Theories (SMT). SMT has gained momentum within the formal verification community in the last decades, leading to the advancement of SMT solvers that have been shown to outperform standard mathematical programming techniques. In this work, we propose generic constraint formulations that can model a wide range of real-world scheduling constraints. Then, the generic constraints are formulated as SMT and MILP problems and used to compare the respective state-of-the-art solvers, Z3 and Gurobi, on academic and real-world inspired rostering problems. Experimental results show how each solver excels for certain types of problems; the MILP solver generally performs better when the problem is highly constrained or infeasible, while the SMT solver performs better otherwise. On real-world inspired problems containing a more varied set of shifts and personnel, the SMT solver excels. Additionally, it was noted during experimentation that the SMT solver was more sensitive to the way the generic constraints were formulated, requiring careful consideration and experimentation to achieve better performance. We conclude that SMT-based methods present a promising avenue for future research within the domain of personnel scheduling.

arXiv:2505.10183

KAITIAN: A Unified Communication Framework for Enabling Efficient Collaboration Across Heterogeneous Accelerators in Embodied AI Systems

Embodied Artificial Intelligence (AI) systems, such as autonomous robots and intelligent vehicles, are increasingly reliant on diverse heterogeneous accelerators (e.g., GPGPUs, NPUs, FPGAs) to meet stringent real-time processing and energy-efficiency demands. However, the proliferation of vendor-specific proprietary communication libraries creates significant interoperability barriers, hindering seamless collaboration between different accelerator types and leading to suboptimal resource utilization and performance bottlenecks in distributed AI workloads. This paper introduces KAITIAN, a novel distributed communication framework designed to bridge this gap. KAITIAN provides a unified abstraction layer that intelligently integrates vendor-optimized communication libraries for intra-group efficiency with general-purpose communication protocols for inter-group interoperability. Crucially, it incorporates a load-adaptive scheduling mechanism that dynamically balances computational tasks across heterogeneous devices based on their real-time performance characteristics. Implemented as an extension to PyTorch and rigorously evaluated on a testbed featuring NVIDIA GPUs and Cambricon MLUs, KAITIAN demonstrates significant improvements in resource utilization and scalability for distributed training tasks. Experimental results show that KAITIAN can accelerate training time by up to 42% compared to baseline homogeneous systems, while incurring minimal communication overhead (2.8--4.3%) and maintaining model accuracy. KAITIAN paves the way for more flexible and powerful heterogeneous computing in complex embodied AI applications.

arXiv:2505.09142

ELIS: Efficient LLM Iterative Scheduling System with Response Length Predictor

We propose ELIS, a serving system for Large Language Models (LLMs) featuring an Iterative Shortest Remaining Time First (ISRTF) scheduler designed to efficiently manage inference tasks with the shortest remaining tokens. Current LLM serving systems often employ a first-come-first-served scheduling strategy, which can lead to the "head-of-line blocking" problem. To overcome this limitation, it is necessary to predict LLM inference times and apply a shortest job first scheduling strategy. However, due to the auto-regressive nature of LLMs, predicting the inference latency is challenging. ELIS addresses this challenge by training a response length predictor for LLMs using the BGE model, an encoder-based state-of-the-art model. Additionally, we have devised the ISRTF scheduling strategy, an optimization of shortest remaining time first tailored to existing LLM iteration batching. To evaluate our work in an industrial setting, we simulate streams of requests based on our study of real-world user LLM serving trace records. Furthermore, we implemented ELIS as a cloud-native scheduler system on Kubernetes to evaluate its performance in production environments. Our experimental results demonstrate that ISRTF reduces the average job completion time by up to 19.6%.

arXiv:2505.07842

RAN Cortex: Memory-Augmented Intelligence for Context-Aware Decision-Making in AI-Native Networks

As Radio Access Networks (RAN) evolve toward AI-native architectures, intelligent modules such as xApps and rApps are expected to make increasingly autonomous decisions across scheduling, mobility, and resource management domains. However, these agents remain fundamentally stateless, treating each decision as isolated, lacking any persistent memory of prior events or outcomes. This reactive behavior constrains optimization, especially in environments where network dynamics exhibit episodic or recurring patterns. In this work, we propose RAN Cortex, a memory-augmented architecture that enables contextual recall in AI-based RAN decision systems. RAN Cortex introduces a modular layer composed of four elements: a context encoder that transforms network state into high-dimensional embeddings, a vector-based memory store of past network episodes, a recall engine to retrieve semantically similar situations, and a policy interface that supplies historical context to AI agents in real time or near-real time. We formalize the retrieval-augmented decision problem in the RAN, present a system architecture compatible with O-RAN interfaces, and analyze feasible deployments within the Non-RT and Near-RT RIC domains. Through illustrative use cases such as stadium traffic mitigation and mobility management in drone corridors, we demonstrate how contextual memory improves adaptability, continuity, and overall RAN intelligence. This work introduces memory as a missing primitive in AI-native RAN designs and provides a framework to enable "learning agents" without the need for retraining or centralized inference

arXiv:2505.07598공식 코드 · Papers with Code

Wireless Link Scheduling with State-Augmented Graph Neural Networks

We consider the problem of optimal link scheduling in large-scale wireless ad hoc networks. We specifically aim for the maximum long-term average performance, subject to a minimum transmission requirement for each link to ensure fairness. With a graph structure utilized to represent the conflicts of links, we formulate a constrained optimization problem to learn the scheduling policy, which is parameterized with a graph neural network (GNN). To address the challenge of long-term performance, we use the state-augmentation technique. In particular, by augmenting the Lagrangian dual variables as dynamic inputs to the scheduling policy, the GNN can be trained to gradually adapt the scheduling decisions to achieve the minimum transmission requirements. We verify the efficacy of our proposed policy through numerical simulations and compare its performance with several baselines in various network settings.

arXiv:2505.08682

Joint Communication Scheduling and Resource Allocation for Distributed Edge Learning: Seamless Integration in Next-Generation Wireless Networks

Distributed edge learning (DL) is considered a cornerstone of intelligence enablers, since it allows for collaborative training without the necessity for local clients to share raw data with other parties, thereby preserving privacy and security. Integrating DL into the 6G networks requires a coexistence design with existing services such as high-bandwidth (HB) traffic like eMBB. Current designs in the literature mainly focus on communication round-wise designs that assume a rigid resource allocation throughout each communication round (CR). However, rigid resource allocation within a CR is a highly inefficient and inaccurate representation of the system's realistic behavior. This is due to the heterogeneous nature of the system, as clients inherently may need to access the network at different times. This work zooms into one arbitrary CR, and demonstrates the importance of considering a time-dependent resource sharing design with HB traffic. We first formulate a time-step-wise optimization problem to minimize the consumed time by DL within the CR while constrained by a DL energy budget. Due to its intractability, a session-based optimization problem is formulated assuming a CR lasts less than a large-scale coherence time. Some scheduling properties of such multi-server joint communication scheduling and resource allocation framework have been established. An iterative algorithm has been designed to solve such non-convex and non-block-separable-constrained problems. Simulation results confirm the importance of the efficient and accurate integration design proposed in this work.

arXiv:2505.07833

Patchwork: A Unified Framework for RAG Serving

Retrieval Augmented Generation (RAG) has emerged as a new paradigm for enhancing Large Language Model reliability through integration with external knowledge sources. However, efficient deployment of these systems presents significant technical challenges due to their inherently heterogeneous computational pipelines comprising LLMs, databases, and specialized processing components. We introduce Patchwork, a comprehensive end-to-end RAG serving framework designed to address these efficiency bottlenecks. Patchwork's architecture offers three key innovations: First, it provides a flexible specification interface enabling users to implement custom RAG pipelines. Secondly, it deploys these pipelines as distributed inference systems while optimizing for the unique scalability characteristics of individual RAG components. Third, Patchwork incorporates an online scheduling mechanism that continuously monitors request load and execution progress, dynamically minimizing SLO violations through strategic request prioritization and resource auto-scaling. Our experimental evaluation across four distinct RAG implementations demonstrates that Patchwork delivers substantial performance improvements over commercial alternatives, achieving throughput gains exceeding 48% while simultaneously reducing SLO violations by ~24%.

arXiv:2505.07451

Optimal Low Emission Zones scheduling as an example of transport policy backcasting

This study presents a backcasting approach that considers the passenger car fleet dynamics to determine optimal policy roadmaps in transport systems. As opposed to the scenario-based approach, backcasting sets emission reduction targets first, then identifies policies that meet the constraint. The policy is the implementation of Low Emission Zones (LEZs), in the Ile-de-France region as a case study. The aim is to minimize the number of scrapped vehicles due to LEZs under CO2 emission targets and to deduce an interdiction schedule of polluting vehicles by 2050. To explore potential solutions, we use a genetic algorithm that provides a first insight into optimal policy pathways.

arXiv:2505.08451

Adaptive Bias Generalized Rollout Policy Adaptation on the Flexible Job-Shop Scheduling Problem

The Flexible Job-Shop Scheduling Problem (FJSSP) is an NP-hard combinatorial optimization problem, with several application domains, especially for manufacturing purposes. The objective is to efficiently schedule multiple operations on dissimilar machines. These operations are gathered into jobs, and operations pertaining to the same job need to be scheduled sequentially. Different methods have been previously tested to solve this problem, such as Constraint Solving, Tabu Search, Genetic Algorithms, or Monte Carlo Tree Search (MCTS). We propose a novel algorithm derived from the Generalized Nested Rollout Policy Adaptation, developed to solve the FJSSP. We report encouraging experimental results, as our algorithm performs better than other MCTS-based approaches, even if makespans obtained on large instances are still far from known upper bounds.

arXiv:2505.07293

AttentionInfluence: Adopting Attention Head Influence for Weak-to-Strong Pretraining Data Selection

Recently, there has been growing interest in collecting reasoning-intensive pretraining data to improve LLMs' complex reasoning ability. Prior approaches typically rely on supervised classifiers to identify such data, which requires labeling by humans or LLMs, often introducing domain-specific biases. Due to the attention heads being crucial to in-context reasoning, we propose AttentionInfluence, a simple yet effective, training-free method without supervision signal. Our approach enables a small pretrained language model to act as a strong data selector through a simple attention head masking operation. Specifically, we identify retrieval heads and compute the loss difference when masking these heads. We apply AttentionInfluence to a 1.3B-parameter dense model to conduct data selection on the SmolLM corpus of 241B tokens, and mix the SmolLM corpus with the selected subset comprising 73B tokens to pretrain a 7B-parameter dense model using 1T training tokens and WSD learning rate scheduling. Our experimental results demonstrate substantial improvements, ranging from 1.4pp to 3.5pp, across several knowledge-intensive and reasoning-heavy benchmarks (i.e., MMLU, MMLU-Pro, AGIEval-en, GSM8K, and HumanEval). This demonstrates an effective weak-to-strong scaling property, with small models improving the final performance of larger models-offering a promising and scalable path for reasoning-centric data selection.

arXiv:2505.07680

SpecRouter: Adaptive Routing for Multi-Level Speculative Decoding in Large Language Models

Large Language Models (LLMs) present a critical trade-off between inference quality and computational cost: larger models offer superior capabilities but incur significant latency, while smaller models are faster but less powerful. Existing serving strategies often employ fixed model scales or static two-stage speculative decoding, failing to dynamically adapt to the varying complexities of user requests or fluctuations in system performance. This paper introduces \systemname{}, a novel framework that reimagines LLM inference as an adaptive routing problem solved through multi-level speculative decoding. \systemname{} dynamically constructs and optimizes inference "paths" (chains of models) based on real-time feedback, addressing the limitations of static approaches. Our contributions are threefold: (1) An \textbf{adaptive model chain scheduling} mechanism that leverages performance profiling (execution times) and predictive similarity metrics (derived from token distribution divergence) to continuously select the optimal sequence of draft and verifier models, minimizing predicted latency per generated token. (2) A \textbf{multi-level collaborative verification} framework where intermediate models within the selected chain can validate speculative tokens, reducing the verification burden on the final, most powerful target model. (3) A \textbf{synchronized state management} system providing efficient, consistent KV cache handling across heterogeneous models in the chain, including precise, low-overhead rollbacks tailored for asynchronous batch processing inherent in multi-level speculation. Preliminary experiments demonstrate the validity of our method.

arXiv:2505.06625

CaMDN: Enhancing Cache Efficiency for Multi-tenant DNNs on Integrated NPUs

With the rapid development of DNN applications, multi-tenant execution, where multiple DNNs are co-located on a single SoC, is becoming a prevailing trend. Although many methods are proposed in prior works to improve multi-tenant performance, the impact of shared cache is not well studied. This paper proposes CaMDN, an architecture-scheduling co-design to enhance cache efficiency for multi-tenant DNNs on integrated NPUs. Specifically, a lightweight architecture is proposed to support model-exclusive, NPU-controlled regions inside shared cache to eliminate unexpected cache contention. Moreover, a cache scheduling method is proposed to improve shared cache utilization. In particular, it includes a cache-aware mapping method for adaptability to the varying available cache capacity and a dynamic allocation algorithm to adjust the usage among co-located DNNs at runtime. Compared to prior works, CaMDN reduces the memory access by 33.4% on average and achieves a model speedup of up to 2.56$\times$ (1.88$\times$ on average).

arXiv:2505.07286공식 코드 · Papers with Code

Piloting Structure-Based Drug Design via Modality-Specific Optimal Schedule

Structure-Based Drug Design (SBDD) is crucial for identifying bioactive molecules. Recent deep generative models are faced with challenges in geometric structure modeling. A major bottleneck lies in the twisted probability path of multi-modalities -- continuous 3D positions and discrete 2D topologies -- which jointly determine molecular geometries. By establishing the fact that noise schedules decide the Variational Lower Bound (VLB) for the twisted probability path, we propose VLB-Optimal Scheduling (VOS) strategy in this under-explored area, which optimizes VLB as a path integral for SBDD. Our model effectively enhances molecular geometries and interaction modeling, achieving state-of-the-art PoseBusters passing rate of 95.9% on CrossDock, more than 10% improvement upon strong baselines, while maintaining high affinities and robust intramolecular validity evaluated on held-out test set. Code is available at https://github.com/AlgoMole/MolCRAFT.

arXiv:2505.05231

Adaptive Biased User Scheduling for Heterogeneous Wireless Federate Learning Network

Federated Learning (FL) has revolutionized collaborative model training in distributed networks, prioritizing data privacy and communication efficiency. This paper investigates efficient deployment of FL in wireless heterogeneous networks, focusing on strategies to accelerate convergence despite stragglers. The primary objective is to minimize long-term convergence wall-clock time through optimized user scheduling and resource allocation. While stragglers may introduce delays in a single round, their inclusion can expedite subsequent rounds, particularly when they possess critical information. Moreover, balancing single-round duration with the number of cumulative rounds, compounded by dynamic training and transmission conditions, necessitates a novel approach beyond conventional optimization solutions. To tackle these challenges, convergence analysis with respect to adaptive and biased scheduling is derived. Then, by factoring in real-time system and statistical information, including diverse energy constraints and users' energy harvesting capabilities, a deep reinforcement learning approach, empowered by proximal policy optimization, is employed to adaptively select user sets. For the scheduled users, Lagrangian decomposition is applied to optimize local resource utilization, further enhancing system efficiency. Simulation results validate the effectiveness and robustness of the proposed framework for various FL tasks, demonstrating reduced task time compared to existing benchmarks under various settings.

arXiv:2505.05087

Predictive Control of EV Overnight Charging with Multi-Session Flexibility

The majority of electric vehicles (EVs) are charged domestically overnight, where the precise timing of power allocation is not important to the user, thus representing a source of flexibility that can be leveraged by charging control algorithms. In this paper, we relax the common assumption, that EVs require full charge every morning, enabling additional flexibility to defer charging of surplus energy to subsequent nights, which can enhance the performance of controlled charging. In particular, we consider a simple domestic smart plug, scheduling power delivery with the objective to minimize CO$_2$ emissions over prediction horizons of multiple sessions -- up to seven days ahead -- utilising model predictive control (MPC). Based on carbon intensity data from the UK National Grid, we demonstrate significant potential for emission reductions with multi-session planning of 40 to 46\% compared to uncontrolled charging and 19 to 26\% compared to single-session planning. Furthermore, we assess, how the driving and charging behaviour of EV users affects the available flexibility and consequentially the potential for emission reductions. Finally, using grid carbon intensity data from 14 different UK regions, we report significant variations in absolute emission reductions based on the local energy mix.

arXiv:2505.04532

Integrated equilibrium model for electrified logistics and power systems

This paper proposes an integrated equilibrium model to characterize the complex interactions between electrified logistics systems and electric power delivery systems. The model consists of two major players: an electrified logistics operator (ELO) and a power system operator (PSO). The ELO aims to maximize its profit by strategically scheduling and routing its electric delivery vehicles (e-trucks) for deliveries and charging, in response to the locational marginal price (LMP) set by the PSO. The routing, delivery, and charging behaviors of e-trucks are modeled by a perturbed utility Markov decision process (PU-MDP) while their collective operations are optimized to achieve the ELO's objective by designing rewards in the PU-MDP. On the other hand, PSO optimizes the energy price by considering both the spatiotemporal e-truck charging demand and the base electricity load. The equilibrium of the integrated system is formulated as a fixed point, proved to exist under mild assumptions, and solved for a case study on the Hawaii network via Anderson's fixed-point acceleration algorithm. Along with these numerical results, this paper provides both theoretical insights and practical guidelines to achieve sustainable and efficient operations in modern electrified logistics and power systems.

arXiv:2505.03763공식 코드 · Papers with Code

Splitwiser: Efficient LM inference with constrained resources

Efficient inference of LLMs remains a crucial challenge, with two main phases: a compute-intensive prompt computation and a memory-intensive token generation. Despite existing batching and scheduling techniques, token generation phases fail to fully utilize compute resources, especially when compared to prompt computation phases. To address these challenges, we propose Splitwiser, a methodology that splits the two phases of an LLM inference request onto the same GPU, thereby reducing overhead and improving memory access and cache utilization. By eliminating the need to transfer data across devices, Splitwiser aims to minimize network-related overheads. In this report, we describe the basic structure of our proposed pipeline while sharing preliminary results and analysis. We implement our proposed multiprocessing design on two widely-used and independent LLM architectures: Huggingface and vLLM. We open-source our code for the respective implementations: 1) Huggingface (https://github.com/asad-aali/splitwiser), and 2) vLLM (https://github.com/adney11/vllm-sysml).