# MACE: A Hybrid LLM Serving System with Colocated SLO-aware Continuous Retraining Alignment

arXiv: https://arxiv.org/abs/2510.03283 · HTML: https://mlsanity.com/paper/2510.03283/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Serving systems and request scheduling, one of 12 directions in Efficient LLM inference and serving.
- Builds on 38 of its 57 references that are in the map.
- First posted on arXiv 28 Sep 2025.

## Placed in

- Efficient LLM inference and serving / [Serving systems and request scheduling](https://mlsanity.com/inference/serving-systems-and-request-scheduling/)

## Builds on (most cited first, 30 of 38)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [SGLang: Efficient Execution of Structured Language Model Programs](https://mlsanity.com/paper/2312.07104/) (2023)
- [A General Language Assistant as a Laboratory for Alignment](https://mlsanity.com/paper/2112.00861/) (2021)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks](https://mlsanity.com/paper/1312.6211/) (2013)
- [Efficient Large-Scale Language Model Training on GPU Clusters Using Megatron-LM](https://mlsanity.com/paper/2104.04473/) (2021)
- [DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving](https://mlsanity.com/paper/2401.09670/) (2024)
- [Splitwise: Efficient generative LLM inference using phase splitting](https://mlsanity.com/paper/2311.18677/) (2023)
- [Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Serve](https://mlsanity.com/paper/2403.02310/) (2024)
- [A Comprehensive Survey on Test-Time Adaptation under Distribution Shifts](https://mlsanity.com/paper/2303.15361/) (2023)
- [Understanding Dataset Difficulty with V-Usable Information](https://mlsanity.com/paper/2110.08420/) (2021)
- [Surgical Fine-Tuning Improves Adaptation to Distribution Shifts](https://mlsanity.com/paper/2210.11466/) (2022)
- [Llumnix: Dynamic Scheduling for Large Language Model Serving](https://mlsanity.com/paper/2406.03243/) (2024)
- [Fast Distributed Inference Serving for Large Language Models](https://mlsanity.com/paper/2305.05920/) (2023)
- [Parrot: Efficient Serving of LLM-based Applications with Semantic Variable](https://mlsanity.com/paper/2405.19888/) (2024)
- [Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning](https://mlsanity.com/paper/2410.19258/) (2024)
- [Noise Contrastive Alignment of Language Models with Explicit Rewards](https://mlsanity.com/paper/2402.05369/) (2024)
- [AutoAct: Automatic Agent Learning from Scratch for QA via Self-Planning](https://mlsanity.com/paper/2401.05268/) (2024)
- [ServerlessLLM: Low-Latency Serverless Inference for Large Language Models](https://mlsanity.com/paper/2401.14351/) (2024)
- [Ekya: Continuous Learning of Video Analytics Models on Edge Compute Servers](https://mlsanity.com/paper/2012.10557/) (2020)
- [Large Language Models Empowered Autonomous Edge AI for Connected Intelligence](https://mlsanity.com/paper/2307.02779/) (2023)
- [D'ej`aVu: KV-cache Streaming for Fast, Fault-tolerant Generative LLM Serving](https://mlsanity.com/paper/2403.01876/) (2024)
- [Aryl: An Elastic Cluster Scheduler for Deep Learning](https://mlsanity.com/paper/2202.07896/) (2022)
- [Creativity in LLM-based Multi-Agent Systems: A Survey](https://mlsanity.com/paper/2505.21116/) (2025)
- [Understanding Layer Significance in LLM Alignment](https://mlsanity.com/paper/2410.17875/) (2024)

## Built on by (most cited first, 2 of 2)

- [PIMbot: A Self-Adaptive Attack Framework for Adversarial Manipulation of Multi-Robot Reinforcement Learning](https://mlsanity.com/paper/2605.23027/) (2026)
- [Not Every Sync Is Safe: Calibrated DiLoCo Scheduling for Shared AI Infrastructure](https://mlsanity.com/paper/2607.02544/) (2026)

## Abstract (arXiv)

Large language models (LLMs) deployed on edge servers are increasingly used in latency-sensitive applications such as personalized assistants, recommendation, and content moderation. However, the non-stationary nature of user data necessitates frequent retraining, which introduces a fundamental tension between inference latency and model accuracy under constrained GPU resources. Existing retraining strategies either delay model updates, over-commit resources to retraining, or overlook iteration-level retraining granularity. In this paper, we identify that iteration-level scheduling is crucial for adapting retraining frequency to model drift without violating service-level objectives (SLOs). We propose MACE, a hybrid LLM system that colocates concurrent inference (prefill, decode) and fine-tuning, with intelligent memory management to maximize task performance while promising inference throughput. MACE leverages the insight that not all model updates equally affect output alignment and allocates GPU cycles accordingly to balance throughput, latency, and update freshness. Our trace-driven evaluation shows that MACE matches or exceeds continuous retraining while reducing inference latency by up to 63% and maintaining throughput under resource constraints. Compared to periodic retraining, MACE improves latency breakdown across prefill, decode, and finetune stages, and sustains GPU utilization above 85% in NVIDIA AGX Orin. These results demonstrate that iteration-level hybrid scheduling is a promising direction for deploying LLMs with continual learning capabilities on edge platforms.