Beyond the Buzz: A Pragmatic Take on Inference Disaggregation

• URL: http://arxiv.org/abs/2506.05508v1
• Date: Thu, 05 Jun 2025 18:47:49 GMT
• Title: Beyond the Buzz: A Pragmatic Take on Inference Disaggregation
• Authors: Tiyasa Mitra, Ritika Borkar, Nidhi Bhatia, Ramon Matas, Shivam Raj, Dheevatsa Mudigere, Ritchie Zhao, Maximilian Golub, Arpan Dutta, Sailaja Madduri, Dharmesh Jani, Brian Pharris, Bita Darvish Rouhani,
• Abstract summary: We present the first systematic study of disaggregated inference at scale.<n>We find that disaggregation is most effective for prefill-heavy traffic patterns and larger models.
• Score: 2.9938991029619064
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As inference scales to multi-node deployments, disaggregation - splitting inference into distinct phases - offers a promising path to improving the throughput-interactivity Pareto frontier. Despite growing enthusiasm and a surge of open-source efforts, practical deployment of disaggregated serving remains limited due to the complexity of the optimization search space and system-level coordination. In this paper, we present the first systematic study of disaggregated inference at scale, evaluating hundreds of thousands of design points across diverse workloads and hardware configurations. We find that disaggregation is most effective for prefill-heavy traffic patterns and larger models. Our results highlight the critical role of dynamic rate matching and elastic scaling in achieving Pareto-optimal performance. Our findings offer actionable insights for efficient disaggregated deployments to navigate the trade-off between system throughput and interactivity.

Related papers

• Decomposing the Entropy-Performance Exchange: The Missing Keys to Unlocking Effective Reinforcement Learning [106.68304931854038]
  Reinforcement learning with verifiable rewards (RLVR) has been widely used for enhancing the reasoning abilities of large language models (LLMs)<n>We conduct a systematic empirical analysis of the entropy-performance exchange mechanism of RLVR across different levels of granularity.<n>Our analysis reveals that, in the rising stage, entropy reduction in negative samples facilitates the learning of effective reasoning patterns.<n>In the plateau stage, learning efficiency strongly correlates with high-entropy tokens present in low-perplexity samples and those located at the end of sequences.
  arXiv  Detail & Related papers  (2025-08-04T10:08:10Z)
• Multi-Agent Collaboration via Evolving Orchestration [61.93162413517026]
  Large language models (LLMs) have achieved remarkable results across diverse downstream tasks, but their monolithic nature restricts scalability and efficiency in complex problem-solving.<n>We propose a puppeteer-style paradigm for LLM-based multi-agent collaboration, where a central orchestrator dynamically directs agents in response to evolving task states.<n> Experiments on closed- and open-domain scenarios show that this method achieves superior performance with reduced computational costs.
  arXiv  Detail & Related papers  (2025-05-26T07:02:17Z)
• Reducing Unimodal Bias in Multi-Modal Semantic Segmentation with Multi-Scale Functional Entropy Regularization [66.10528870853324]
  Fusing and balancing multi-modal inputs from novel sensors for dense prediction tasks is critically important.<n>One major limitation is the tendency of multi-modal frameworks to over-rely on easily learnable modalities.<n>We propose a plug-and-play regularization term based on functional entropy, which introduces no additional parameters.
  arXiv  Detail & Related papers  (2025-05-10T12:58:15Z)
• Q-function Decomposition with Intervention Semantics with Factored Action Spaces [51.01244229483353]
  We consider Q-functions defined over a lower dimensional projected subspace of the original action space, and study the condition for the unbiasedness of decomposed Q-functions.<n>This leads to a general scheme which we call action decomposed reinforcement learning that uses the projected Q-functions to approximate the Q-function in standard model-free reinforcement learning algorithms.
  arXiv  Detail & Related papers  (2025-04-30T05:26:51Z)
• Review, Refine, Repeat: Understanding Iterative Decoding of AI Agents with Dynamic Evaluation and Selection [71.92083784393418]
  Inference-time methods such as Best-of-N (BON) sampling offer a simple yet effective alternative to improve performance.<n>We propose Iterative Agent Decoding (IAD) which combines iterative refinement with dynamic candidate evaluation and selection guided by a verifier.
  arXiv  Detail & Related papers  (2025-04-02T17:40:47Z)
• Framework for Progressive Knowledge Fusion in Large Language Models Through Structured Conceptual Redundancy Analysis [0.0]
  The organization of latent knowledge within large-scale models poses unique challenges when addressing overlapping representations and optimizing contextual accuracy.<n>A framework was proposed to restructure these redundancies through advanced clustering techniques and dynamic thresholding.<n> Evaluations revealed improved memory efficiency and faster inference times, alongside better alignment in latent knowledge clusters that enhanced interpretability.
  arXiv  Detail & Related papers  (2025-01-23T11:34:04Z)
• DisCo-DSO: Coupling Discrete and Continuous Optimization for Efficient Generative Design in Hybrid Spaces [12.729697787995892]
  DisCo-DSO is a novel approach that uses a generative model to learn a joint distribution over discrete and continuous design variables.<n>In particular, we illustrate DisCo-DSO's superiority over the state-of-the-art methods for interpretable reinforcement learning with decision trees.
  arXiv  Detail & Related papers  (2024-12-15T04:51:54Z)
• Efficient Pareto Manifold Learning with Low-Rank Structure [31.082432589391953]
  Multi-task learning is inherently a multi-objective optimization problem. We propose a novel approach that integrates a main network with several low-rank matrices. It significantly reduces the number of parameters and facilitates the extraction of shared features.
  arXiv  Detail & Related papers  (2024-07-30T11:09:27Z)
• Unifying Feature and Cost Aggregation with Transformers for Semantic and Visual Correspondence [51.54175067684008]
  This paper introduces a Transformer-based integrative feature and cost aggregation network designed for dense matching tasks. We first show that feature aggregation and cost aggregation exhibit distinct characteristics and reveal the potential for substantial benefits stemming from the judicious use of both aggregation processes. Our framework is evaluated on standard benchmarks for semantic matching, and also applied to geometric matching, where we show that our approach achieves significant improvements compared to existing methods.
  arXiv  Detail & Related papers  (2024-03-17T07:02:55Z)
• Bayesian Off-Policy Evaluation and Learning for Large Action Spaces [13.001601860404426]
  In interactive systems, actions are often correlated, presenting an opportunity for more sample-efficient off-policy evaluation and learning.<n>We introduce a unified Bayesian framework to capture these correlations through structured and informative priors.<n>We propose sDM, a generic Bayesian approach for OPE and OPL, grounded in both algorithmic and theoretical foundations.
  arXiv  Detail & Related papers  (2024-02-22T16:09:45Z)
• BOtied: Multi-objective Bayesian optimization with tied multivariate ranks [33.414682601242006]
  In this paper, we show a natural connection between non-dominated solutions and the extreme quantile of the joint cumulative distribution function. Motivated by this link, we propose the Pareto-compliant CDF indicator and the associated acquisition function, BOtied. Our experiments on a variety of synthetic and real-world problems demonstrate that BOtied outperforms state-of-the-art MOBO acquisition functions.
  arXiv  Detail & Related papers  (2023-06-01T04:50:06Z)
• ECO-TR: Efficient Correspondences Finding Via Coarse-to-Fine Refinement [80.94378602238432]
  We propose an efficient structure named Correspondence Efficient Transformer (ECO-TR) by finding correspondences in a coarse-to-fine manner. To achieve this, multiple transformer blocks are stage-wisely connected to gradually refine the predicted coordinates. Experiments on various sparse and dense matching tasks demonstrate the superiority of our method in both efficiency and effectiveness against existing state-of-the-arts.
  arXiv  Detail & Related papers  (2022-09-25T13:05:33Z)
• Fusion and Orthogonal Projection for Improved Face-Voice Association [15.938463726577128]
  We study the problem of learning association between face and voice. We propose a light-weight, plug-and-play mechanism that exploits the complementary cues in both modalities to form enriched fused embeddings.
  arXiv  Detail & Related papers  (2021-12-20T12:33:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.