Daily Papers

Large Language Model (LLM) serving systems remain fundamentally fragile, where frequent hardware faults in hyperscale clusters trigger disproportionate service outages in the software stack. Current recovery mechanisms are prohibitively slow, often requiring up to 10 minutes to reinitialize resources and reload massive model weights. We introduce KevlarFlow, a fault tolerant serving architecture designed to bridge the gap between hardware unreliability and service availability. KevlarFlow leverages 1) decoupled model parallelism initialization, 2) dynamic traffic rerouting, and 3) background KV cache replication to maintain high throughput during partial failures. Our evaluation demonstrates that KevlarFlow reduces mean-time-to-recovery (MTTR) by 20x and, under failure conditions, improves average latency by 3.1x, 99th percentile (p99) latency by 2.8x, average time-to-first-token (TTFT) by 378.9x, and p99 TTFT by 574.6x with negligible runtime overhead in comparison to state-of-the-art LLM serving systems.

Large Language Models (LLMs) have recently experienced great success, as evident in the widespread popularity of ChatGPT. Existing LLM serving systems are stateless across requests. Consequently, when LLMs are used in the common setting of multi-turn conversations, a growing log of the conversation history must be processed alongside any request by the serving system at each turn, resulting in repeated history processing. In this paper, we design Pensieve, a system optimized for multi-turn conversation LLM serving. Pensieve maintains the conversation state across requests by caching previously processed history to avoid duplicate processing. Pensieve's multi-tier caching strategy can utilize both GPU and CPU memory to efficiently store and retrieve cached data. Pensieve also generalizes the recent PagedAttention kernel to support attention between multiple input tokens with a GPU cache spread over non-contiguous memory. Our evaluation shows that Pensieve is able to achieve 1.51-1.95x throughput compared to vLLM and reduce latency by 60-75%.

Large language models (LLMs) are increasingly integrated into many online services. However, a major challenge in deploying LLMs is their high cost, due primarily to the use of expensive GPU instances. To address this problem, we find that the significant heterogeneity of GPU types presents an opportunity to increase GPU cost efficiency and reduce deployment costs. The broad and growing market of GPUs creates a diverse option space with varying costs and hardware specifications. Within this space, we show that there is not a linear relationship between GPU cost and performance, and identify three key LLM service characteristics that significantly affect which GPU type is the most cost effective: model request size, request rate, and latency service-level objective (SLO). We then present M\'elange, a framework for navigating the diversity of GPUs and LLM service specifications to derive the most cost-efficient set of GPUs for a given LLM service. We frame the task of GPU selection as a cost-aware bin-packing problem, where GPUs are bins with a capacity and cost, and items are request slices defined by a request size and rate. Upon solution, M\'elange derives the minimal-cost GPU allocation that adheres to a configurable latency SLO. Our evaluations across both real-world and synthetic datasets demonstrate that M\'elange can reduce deployment costs by up to 77% as compared to utilizing only a single GPU type, highlighting the importance of making heterogeneity-aware GPU provisioning decisions for LLM serving. Our source code is publicly available at https://github.com/tyler-griggs/melange-release.

Large language models (LLMs) often incorporate multiple text chunks in their inputs to provide the necessary contexts. To speed up the prefill of the long LLM inputs, one can pre-compute the KV cache of a text and re-use the KV cache when the context is reused as the prefix of another LLM input. However, the reused text chunks are not always the input prefix, and when they are not, their precomputed KV caches cannot be directly used since they ignore the text's cross-attention with the preceding text in the LLM input. Thus, the benefits of reusing KV caches remain largely unrealized. This paper tackles just one question: when an LLM input contains multiple text chunks, how to quickly combine their precomputed KV caches in order to achieve the same generation quality as the expensive full prefill (i.e., without reusing KV cache)? We present CacheBlend, a scheme that reuses the pre-computed KV caches, regardless prefix or not, and selectively recomputes the KV values of a small subset of tokens to partially update each reused KV cache. In the meantime,the small extra delay for recomputing some tokens can be pipelined with the retrieval of KV caches within the same job,allowing CacheBlend to store KV caches in slower devices with more storage capacity while retrieving them without increasing the inference delay. By comparing CacheBlend with the state-of-the-art KV cache reusing schemes on three open-source LLMs of various sizes and four popular benchmark datasets of different tasks, we show that CacheBlend reduces time-to-first-token (TTFT) by 2.2-3.3X and increases the inference throughput by 2.8-5X, compared with full KV recompute, without compromising generation quality or incurring more storage cost.

Large language models (LLMs) have excelled in various applications, yet serving them at scale is challenging due to their substantial resource demands and high latency. Our real-world studies reveal that over 70% of user requests to LLMs have semantically similar counterparts, suggesting the potential for knowledge transfer among requests. However, naively caching and reusing past responses leads to a big quality drop. In this paper, we introduce IC-Cache, a caching system that enables live LLM capability augmentation to improve serving efficiency: by leveraging historical request-response pairs from larger models as in-context examples, IC-Cache empowers small LLMs to imitate and even exceed the compositional abilities (e.g., reasoning) of their larger counterparts, enabling selective offloading of requests to reduce cost and latency. Achieving this live augmentation at scale introduces intricate trade-offs between response quality, latency, and system throughput. For a new request, IC-Cache efficiently selects similar, high-utility examples to prepend them to the new request's input. At scale, it adaptively routes requests across LLMs of varying capabilities, accounting for response quality and serving loads. IC-Cache employs a cost-aware cache replay mechanism that refines example quality offline to maximize online cache utility and efficiency. Evaluations on millions of realistic requests demonstrate that IC-Cache improves LLM serving throughput by 1.4-5.9x and reduces latency by 28-71% without hurting response quality.

Large Language Models (LLMs) represent a revolutionary advancement in the contemporary landscape of artificial general intelligence (AGI). As exemplified by ChatGPT, LLM-based applications necessitate minimal response latency and maximal throughput for inference serving. However, due to the unpredictability of LLM execution, the first-come-first-serve (FCFS) scheduling policy employed by current LLM serving systems suffers from head-of-line (HoL) blocking issues and long job response times. In this paper, we propose a new efficient LLM inference serving framework, named ALISE. The key design paradigm of ALISE is to leverage a novel speculative scheduler by estimating the execution time for each job and exploiting such prior knowledge to assign appropriate job priority orders, thus minimizing potential queuing delays for heterogeneous workloads. Furthermore, to mitigate the memory overhead of the intermediate key-value (KV) cache, we employ a priority-based adaptive memory management protocol and quantization-based compression techniques. Evaluations demonstrate that in comparison to the state-of-the-art solution vLLM, ALISE improves the throughput of inference serving by up to 1.8x and 2.1x under the same latency constraint on the Alpaca and ShareGPT datasets, respectively.

Practical large language model (LLM) services may involve a long system prompt, which specifies the instructions, examples, and knowledge documents of the task and is reused across numerous requests. However, the long system prompt causes throughput/latency bottlenecks as the cost of generating the next token grows w.r.t. the sequence length. This paper aims to improve the efficiency of LLM services that involve long system prompts. Our key observation is that handling these system prompts requires heavily redundant memory accesses in existing causal attention computation algorithms. Specifically, for batched requests, the cached hidden states (i.e., key-value pairs) of system prompts are transferred from off-chip DRAM to on-chip SRAM multiple times, each corresponding to an individual request. To eliminate such a redundancy, we propose RelayAttention, an attention algorithm that allows reading these hidden states from DRAM exactly once for a batch of input tokens. RelayAttention is a free lunch: it maintains the generation quality while requiring no model retraining, as it is based on a mathematical reformulation of causal attention.

High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks dynamically. When managed inefficiently, this memory can be significantly wasted by fragmentation and redundant duplication, limiting the batch size. To address this problem, we propose PagedAttention, an attention algorithm inspired by the classical virtual memory and paging techniques in operating systems. On top of it, we build vLLM, an LLM serving system that achieves (1) near-zero waste in KV cache memory and (2) flexible sharing of KV cache within and across requests to further reduce memory usage. Our evaluations show that vLLM improves the throughput of popular LLMs by 2-4times with the same level of latency compared to the state-of-the-art systems, such as FasterTransformer and Orca. The improvement is more pronounced with longer sequences, larger models, and more complex decoding algorithms. vLLM's source code is publicly available at https://github.com/vllm-project/vllm

Inference serving for large language models (LLMs) is the key to unleashing their potential in people's daily lives. However, efficient LLM serving remains challenging today because the requests are inherently heterogeneous and unpredictable in terms of resource and latency requirements, as a result of the diverse applications and the dynamic execution nature of LLMs. Existing systems are fundamentally limited in handling these characteristics and cause problems such as severe queuing delays, poor tail latencies, and SLO violations. We introduce Llumnix, an LLM serving system that reacts to such heterogeneous and unpredictable requests by runtime rescheduling across multiple model instances. Similar to context switching across CPU cores in modern operating systems, Llumnix reschedules requests to improve load balancing and isolation, mitigate resource fragmentation, and differentiate request priorities and SLOs. Llumnix implements the rescheduling with an efficient and scalable live migration mechanism for requests and their in-memory states, and exploits it in a dynamic scheduling policy that unifies the multiple rescheduling scenarios elegantly. Our evaluations show that Llumnix improves tail latencies by an order of magnitude, accelerates high-priority requests by up to 1.5x, and delivers up to 36% cost savings while achieving similar tail latencies, compared against state-of-the-art LLM serving systems. Llumnix is publicly available at https://github.com/AlibabaPAI/llumnix.

The increasing usage of Large Language Models (LLMs) has resulted in a surging demand for planet-scale serving systems, where tens of thousands of GPUs continuously serve hundreds of millions of users. Consequently, throughput (under reasonable latency constraints) has emerged as a key metric that determines serving systems' performance. To boost throughput, various methods of inter-device parallelism (e.g., data, tensor, pipeline) have been explored. However, existing methods do not consider overlapping the utilization of different resources within a single device, leading to underutilization and sub-optimal performance. We propose NanoFlow, a novel serving framework that exploits intra-device parallelism, which overlaps the usage of resources including compute, memory, and network within a single device through operation co-scheduling. To exploit intra-device parallelism, NanoFlow introduces two key innovations: First, NanoFlow splits requests into nano-batches at the granularity of operations, which breaks the dependency of sequential operations in LLM inference and enables overlapping; then, to get benefit from overlapping, NanoFlow uses an operation-level pipeline with execution unit scheduling, which partitions the device's functional units and simultaneously executes different operations in each unit. NanoFlow automates the pipeline setup using a parameter search algorithm, which enables easily porting NanoFlow to different models. We implement NanoFlow on NVIDIA GPUs and evaluate end-to-end serving throughput on several popular models such as LLaMA-2-70B, Mixtral 8x7B, LLaMA-3-8B, etc.. With practical workloads, NanoFlow provides 1.91x throughput boost compared to state-of-the-art serving systems achieving 59% to 72% of optimal throughput across ported models.

This paper introduces PowerInfer, a high-speed Large Language Model (LLM) inference engine on a personal computer (PC) equipped with a single consumer-grade GPU. The key underlying the design of PowerInfer is exploiting the high locality inherent in LLM inference, characterized by a power-law distribution in neuron activation. This distribution indicates that a small subset of neurons, termed hot neurons, are consistently activated across inputs, while the majority, cold neurons, vary based on specific inputs. PowerInfer exploits such an insight to design a GPU-CPU hybrid inference engine: hot-activated neurons are preloaded onto the GPU for fast access, while cold-activated neurons are computed on the CPU, thus significantly reducing GPU memory demands and CPU-GPU data transfers. PowerInfer further integrates adaptive predictors and neuron-aware sparse operators, optimizing the efficiency of neuron activation and computational sparsity. Evaluation shows that PowerInfer attains an average token generation rate of 13.20 tokens/s, with a peak of 29.08 tokens/s, across various LLMs (including OPT-175B) on a single NVIDIA RTX 4090 GPU, only 18% lower than that achieved by a top-tier server-grade A100 GPU. This significantly outperforms llama.cpp by up to 11.69x while retaining model accuracy.

In the rapidly evolving landscape of artificial intelligence (AI), generative large language models (LLMs) stand at the forefront, revolutionizing how we interact with our data. However, the computational intensity and memory consumption of deploying these models present substantial challenges in terms of serving efficiency, particularly in scenarios demanding low latency and high throughput. This survey addresses the imperative need for efficient LLM serving methodologies from a machine learning system (MLSys) research perspective, standing at the crux of advanced AI innovations and practical system optimizations. We provide in-depth analysis, covering a spectrum of solutions, ranging from cutting-edge algorithmic modifications to groundbreaking changes in system designs. The survey aims to provide a comprehensive understanding of the current state and future directions in efficient LLM serving, offering valuable insights for researchers and practitioners in overcoming the barriers of effective LLM deployment, thereby reshaping the future of AI.

DistServe improves the performance of large language models (LLMs) serving by disaggregating the prefill and decoding computation. Existing LLM serving systems colocate the two phases and batch the computation of prefill and decoding across all users and requests. We find that this strategy not only leads to strong prefill-decoding interferences but also couples the resource allocation and parallelism plans for both phases. LLM applications often emphasize individual latency for each phase: time to first token (TTFT) for the prefill phase and time per output token (TPOT) of each request for the decoding phase. In the presence of stringent latency requirements, existing systems have to prioritize one latency over the other, or over-provision compute resources to meet both. DistServe assigns prefill and decoding computation to different GPUs, hence eliminating prefill-decoding interferences. Given the application's TTFT and TPOT requirements, DistServe co-optimizes the resource allocation and parallelism strategy tailored for each phase. DistServe also places the two phases according to the serving cluster's bandwidth to minimize the communication caused by disaggregation. As a result, DistServe significantly improves LLM serving performance in terms of the maximum rate that can be served within both TTFT and TPOT constraints on each GPU. Our evaluations show that on various popular LLMs, applications, and latency requirements, DistServe can serve 4.48x more requests or 10.2x tighter SLO, compared to state-of-the-art systems, while staying within latency constraints for > 90% of requests.

We survey the large language model (LLM) serving area to understand the intricate dynamics between cost-efficiency and accuracy, which is magnified by the growing need for longer contextual understanding when deploying models at a massive scale. Our findings reveal that works in this space optimize along three distinct but conflicting goals: improving serving context length (C), improving serving accuracy (A), and improving serving performance (P). Drawing inspiration from the CAP theorem in databases, we propose a CAP principle for LLM serving, which suggests that any optimization can improve at most two of these three goals simultaneously. Our survey categorizes existing works within this framework. We find the definition and continuity of user-perceived measurement metrics are crucial in determining whether a goal has been met, akin to prior CAP databases in the wild. We recognize the CAP principle for LLM serving as a guiding principle, rather than a formal theorem, to inform designers of the inherent and dynamic trade-offs in serving models. As serving accuracy and performance have been extensively studied, this survey focuses on works that extend serving context length and address the resulting challenges.

This paper introduces DeepFlow, a scalable and serverless AI platform designed to efficiently serve large language models (LLMs) at scale in cloud environments. DeepFlow addresses key challenges such as resource allocation, serving efficiency, and cold start latencies through four main design components. First, it uses a simple serverless abstraction called the request-job-task model, which helps manage AI workloads across post-training and model serving tasks. Second, it builds an in-house serving engine FlowServe using a microkernel-inspired design, NPU-centric execution, and SPMD-based parallelism to optimize LLM serving. The system also includes novel scheduling policies tailored for both PD-disaggregated and PD-colocated configurations. With optimizations like pre-warmed pods, DRAM pre-loading, and NPU-fork, DeepFlow can scale up to 64 instances in seconds. DeepFlow has been in production for over a year, operating on a large Ascend NPU cluster and providing industrystandard APIs for fine-tuning, agent serving, and model serving to our customers.

Despite the rapid advancements of large language models (LLMs), LLM serving systems remain memory-intensive and costly. The key-value (KV) cache, which stores KV tensors during autoregressive decoding, is crucial for enabling low-latency, high-throughput LLM inference serving. In this survey, we focus on system-aware KV infrastructure for serving LLMs (abbreviated as sKis). We revisit recent work from a system behavior perspective, organizing existing efforts into three dimensions: execution and scheduling (temporal), placement and migration (spatial), and representation and retention (structural). Furthermore, we analyze cross-behavior co-design affinity and behavior-objective links, highlighting future opportunities. Our work systematizes a rapidly evolving area, providing a foundation for understanding and innovating KV cache designs in modern LLM serving infrastructure.

Key-value (KV) caching is a powerful technique for accelerating large language model inference and generation. Inference workloads are large and diverse, which makes them difficult to cache effectively. Existing cache management strategies adopt the least-recently-used policy for evicting cache blocks. However, LRU leads to multiple unrelated workloads flushing each other's caches. To address this, we integrate adaptive caching that dynamically allocates cache space between recently and frequently occurring KV blocks. Evaluations show that it improves the KV cache hit rate by up to 10.8% and reduces time to first token by up to 12.6% over naive vLLM on synthetic document question answering workloads, and 2.1% and 2.0% respectively on real-world conversation workloads. The method generalizes well to batch inference and demonstrates clear interpretability while effectively accommodating diverse workloads.

With the widespread adoption of Large Language Models (LLMs), serving LLM inference requests has become an increasingly important task, attracting active research advancements. Practical workloads play an essential role in this process: they are critical for motivating and benchmarking serving techniques and systems. However, the existing understanding of real-world LLM serving workloads is limited due to the lack of a comprehensive workload characterization. Prior analyses remain insufficient in scale and scope, thus failing to fully capture intricate workload characteristics. In this paper, we fill the gap with an in-depth characterization of LLM serving workloads collected from our worldwide cloud inference serving service, covering not only language models but also emerging multimodal and reasoning models, and unveiling important new findings in each case. Moreover, based on our findings, we propose ServeGen, a principled framework for generating realistic LLM serving workloads by composing them on a per-client basis. A practical use case in production validates that ServeGen avoids 50% under-provisioning compared to naive workload generation, demonstrating ServeGen's advantage in performance benchmarking. We will open-source ServeGen to foster future research.

Serving numerous users and requests concurrently requires good fairness in Large Language Models (LLMs) serving system. This ensures that, at the same cost, the system can meet the Service Level Objectives (SLOs) of more users , such as time to first token (TTFT) and time between tokens (TBT), rather than allowing a few users to experience performance far exceeding the SLOs. To achieve better fairness, the preemption-based scheduling policy dynamically adjusts the priority of each request to maintain balance during runtime. However, existing systems tend to overly prioritize throughput, overlooking the overhead caused by preemption-induced context switching, which is crucial for maintaining fairness through priority adjustments. In this work, we identify three main challenges that result in this overhead. 1) Inadequate I/O utilization. 2) GPU idleness. 3) Unnecessary I/O transmission during multi-turn conversations. Our key insight is that the block-based KV cache memory policy in existing systems, while achieving near-zero memory waste, leads to discontinuity and insufficient granularity in the KV cache memory. To respond, we introduce FastSwitch, a fairness-aware serving system that not only aligns with existing KV cache memory allocation policy but also mitigates context switching overhead. Our evaluation shows that FastSwitch outperforms the state-of-the-art LLM serving system vLLM with speedups of 1.4-11.2x across different tail TTFT and TBT.

Fine-tuning large language models (LLMs) for downstream tasks can greatly improve model quality, however serving many different fine-tuned LLMs concurrently for users in multi-tenant environments is challenging. Dedicating GPU memory for each model is prohibitively expensive and naively swapping large model weights in and out of GPU memory is slow. Our key insight is that fine-tuned models can be quickly swapped in and out of GPU memory by extracting and compressing the delta between each model and its pre-trained base model. We propose DeltaZip, an LLM serving system that efficiently serves multiple full-parameter fine-tuned models concurrently by aggressively compressing model deltas by a factor of 6times to 8times while maintaining high model quality. DeltaZip increases serving throughput by 1.5times to 3times and improves SLO attainment compared to a vanilla HuggingFace serving system.

Heterogeneous LLM serving stacks split scheduling into two layers that optimize in isolation: model routers pick a model from quality and cost signals while ignoring instance load, and serving load balancers optimize queues while ignoring quality. We present RouteBalance, a serving-aware scheduling layer that fuses both into a single online assignment over concrete model instances, jointly trading off quality, latency, and cost. A batched in-process predictor stack and dead-reckoned instance state keep the joint decision cheap on the request hot path (approx32 ms at 12 req/s). On a 13-instance, 28-GPU heterogeneous cluster serving four model sizes, a single deployed RouteBalance stack traces the upper region of the three-way quality-cost-throughput frontier. Sweeping one weight vector reaches both the highest routing-decision quality (DeepEval 0.419, +0.013 over the strongest baseline, 95% CI [{+}0.005,{+}0.022]; the ordering holds when a second judge re-scores the actually served text) and, at its cost-priority corner, per-request cost that ties the cheapest baseline. With router engineering equalized against concurrent-scoring baseline variants we build, its balanced preset serves at 2.8 s and 30 req/s, leading 2.6 to 4.1times ahead of enhanced BEST-Route at high load. (Deploying those routers as published, one serial scoring call per request, makes them collapse 23times under load, a deployment-architecture effect we isolate separately, not the routing result.) A four-arm isolation shows the benefit follows from pricing latency at model-selection time; the learned predictors contribute calibration and SLO headroom rather than the headline frontier. Code: https://github.com/AKafakA/route-balance

While model serving has unlocked unprecedented capabilities, the high cost of serving large-scale models continues to be a significant barrier to widespread accessibility and rapid innovation. Compiler optimizations have long driven substantial performance improvements, but existing compilers struggle with neural workloads due to the exponentially large and highly interdependent space of possible transformations. Although existing stochastic search techniques can be effective, they are often sample-inefficient and fail to leverage the structural context underlying compilation decisions. We set out to investigate the research question of whether reasoning with large language models (LLMs), without any retraining, can leverage the context-aware decision space of compiler optimizations to significantly improve sample efficiency. To that end, we introduce a novel compilation framework (dubbed REASONING COMPILER) that formulates optimization as a sequential, context-aware decision process guided by a large language model and structured Monte Carlo tree search (MCTS). The LLM acts as a proposal mechanism, suggesting hardware-informed transformations that reflect the current program state and accumulated performance feedback. MCTS incorporates the LLM-generated proposals to balance exploration and exploitation, facilitating a structured, context-sensitive traversal of the expansive compiler optimization space. By achieving substantial speedups with markedly fewer samples than leading neural compilers, our approach demonstrates the potential of LLM-guided reasoning to transform the landscape of compiler optimization.

Multi-turn LLM serving accumulates dialogue history whose Key-Value (KV) cache grows with every turn and every user, quickly exceeding the model weights themselves and making memory -- not compute -- the binding constraint on throughput. Non-uniform KV compression, which allocates heterogeneous budgets across attention heads, preserves accuracy far better than uniform schemes, yet remains impractical: modern serving stacks assume identical KV lengths across heads, so heterogeneity traps freed memory as page fragmentation, spends up to 25% of prefill time reclaiming scattered pages, and skews GPU workloads that inflate decode latency by up to 1.7times or burn 15--20% of each decode step on re-planning. We observe that this heterogeneity need not be discovered at runtime: head-wise retention follows a two-level structural regularity -- an input-invariant head ranking with narrowly bounded per-head ratios -- that can be calibrated offline from as few as 50 samples. Building on this insight, we present Tangram, a serving framework that statically resolves what prior systems handle dynamically: Budget Reservation fixes each head's post-compression footprint at scheduling time, eliminating page reclamation; Ragged Paging clusters similar-budget heads into independent page tables, turning fragmentation into reclaimable memory; and Ahead-of-Time Load Balancing precomputes balanced GPU partitions with zero runtime planning. Implemented on vLLM, Tangram serves as a drop-in substrate for existing non-uniform compression methods, matching their accuracy while improving end-to-end throughput by up to 2.6times over the full-KV baseline. Our implementation is publicly available at https://github.com/aiha-lab/TANGRAM.

Multimodal large language models (MLLMs) extend LLMs to handle images, videos, and audio by incorporating feature extractors and projection modules. However, these additional components -- combined with complex inference pipelines and heterogeneous workloads -- introduce significant inference overhead. Therefore, efficiently serving MLLMs remains a major challenge. Current tightly coupled serving architectures struggle to distinguish between mixed request types or adapt parallelism strategies to different inference stages, leading to increased time-to-first-token (TTFT) latency and poor resource utilization. To address this, we introduce Elastic Multimodal Parallelism (EMP), a new serving paradigm that elastically adapts to resource heterogeneity across request types and inference stages. Building upon EMP, we develop ElasticMM, an MLLM serving system that (1) separates requests into independent modality groups with dynamic resource allocation via a modality-aware load balancer; (2) decouples inference stages and enables parallelism adjustment and adaptive scaling via elastic partition scheduling; and (3) improves inference efficiency through unified multimodal prefix caching and non-blocking encoding. Experiments on diverse real-world datasets show that ElasticMM outperforms state-of-the-art (SOTA) serving systems, reducing TTFT by up to 4.2x and achieving 3.2-4.5x higher throughput while meeting service-level objectives (SLOs).

The growing demand for large language models (LLMs) requires serving systems to handle many concurrent requests with diverse service level objectives (SLOs). This exacerbates head-of-line (HoL) blocking during the compute-intensive prefill phase, where long-running requests monopolize resources and delay higher-priority ones, leading to widespread time-to-first-token (TTFT) SLO violations. While chunked prefill enables interruptibility, it introduces an inherent trade-off between responsiveness and throughput: reducing chunk size improves response latency but degrades computational efficiency, whereas increasing chunk size maximizes throughput but exacerbates blocking. This necessitates an adaptive preemption mechanism. However, dynamically balancing execution granularity against scheduling overheads remains a key challenge. In this paper, we propose FlowPrefill, a TTFT-goodput-optimized serving system that resolves this conflict by decoupling preemption granularity from scheduling frequency. To achieve adaptive prefill scheduling, FlowPrefill introduces two key innovations: 1) Operator-Level Preemption, which leverages operator boundaries to enable fine-grained execution interruption without the efficiency loss associated with fixed small chunking; and 2) Event-Driven Scheduling, which triggers scheduling decisions only upon request arrival or completion events, thereby supporting efficient preemption responsiveness while minimizing control-plane overhead. Evaluation on real-world production traces shows that FlowPrefill improves maximum goodput by up to 5.6times compared to state-of-the-art systems while satisfying heterogeneous SLOs.

Multi-modal Large Language Models (MLLMs) serving systems commonly employ KV-cache compression to reduce memory footprint. However, existing compression methods introduce significant processing overhead and queuing delays, particularly in concurrent serving scenarios. We present FastCache, a novel serving framework that effectively addresses these challenges through two key innovations: (1) a dynamic batching strategy that optimizes request scheduling across prefill, compression, and decode stages, and (2) an efficient KV-cache memory pool mechanism that eliminates memory fragmentation while maintaining high GPU utilization. Our comprehensive experiments on the GQA and MileBench datasets demonstrate that FastCache achieves up to 19.3times reduction in Time-To-First-Token (TTFT) and 12.1times improvement in throughput compared to state-of-the-art baselines. The system maintains stable performance under high-concurrency scenarios (up to 40 req/s) while reducing average memory consumption by 20\%. These results establish FastCache as an efficient solution for real-world LLM serving systems with KV-cache compression.

This paper presents Block, a distributed scheduling framework designed to optimize load balancing and auto-provisioning across instances in large language model serving frameworks by leveraging contextual information from incoming requests. Unlike popular model serving systems that rely on monolithic and heuristic task schedulers, Block operates as a fully distributed, stateless, and predictive scheduling system to achieve low overhead, reliability, and scalability. It leverages the deterministic and predictable characteristics of LLM inferences, such as host configurations, response lengths, and hardware performance, to make scheduling decisions based on accurately predicted metrics. Evaluation on a 12 GPUs cluster shows that Block significantly outperforms heuristic schedulers, boosting serving capacity by up to 16.7\% and reducing P99 tail latency by up to 49.5\%. These performance gains remain consistent across diverse models, workloads and configurations. Code and data are open-sourced.

LLMs are widely adopted in production, pushing inference systems to their limits. Disaggregated LLM serving (e.g., PD separation and KV state disaggregation) improves scalability and cost efficiency, but it also turns KV into an explicit payload crossing network and storage boundaries, making KV a dominant end-to-end bottleneck. Existing KV compression are typically static runtime configurations, despite production service context varies over time in workload mix, bandwidth, and SLO/quality budgets. As a result, a fixed choice can be suboptimal or even increase latency. We present KVServe, the first service-aware and adaptive KV communication compression framework for disaggregated LLM serving: KVServe (1) unifies KV compression into a modular strategy space with new components and cross-method recomposition; (2) introduces Bayesian Profiling Engine that efficiently searches this space and distills a 3D Pareto candidate set, reducing 50times offline search overhead; and (3) deploys a Service-Aware Online Controller that combines an analytical latency model with a lightweight bandit to select profiles under constraints and correct offline-to-online mismatch. Integrated into vLLM and evaluated across datasets, models, GPUs and networks, KVServe achieves up to 9.13times JCT speedup in PD-separated serving and up to 32.8times TTFT reduction in KV-disaggregated serving.

Quantization can accelerate large language model (LLM) inference. Going beyond INT8 quantization, the research community is actively exploring even lower precision, such as INT4. Nonetheless, state-of-the-art INT4 quantization techniques only accelerate low-batch, edge LLM inference, failing to deliver performance gains in large-batch, cloud-based LLM serving. We uncover a critical issue: existing INT4 quantization methods suffer from significant runtime overhead (20-90%) when dequantizing either weights or partial sums on GPUs. To address this challenge, we introduce QoQ, a W4A8KV4 quantization algorithm with 4-bit weight, 8-bit activation, and 4-bit KV cache. QoQ stands for quattuor-octo-quattuor, which represents 4-8-4 in Latin. QoQ is implemented by the QServe inference library that achieves measured speedup. The key insight driving QServe is that the efficiency of LLM serving on GPUs is critically influenced by operations on low-throughput CUDA cores. Building upon this insight, in QoQ algorithm, we introduce progressive quantization that can allow low dequantization overhead in W4A8 GEMM. Additionally, we develop SmoothAttention to effectively mitigate the accuracy degradation incurred by 4-bit KV quantization. In the QServe system, we perform compute-aware weight reordering and take advantage of register-level parallelism to reduce dequantization latency. We also make fused attention memory-bound, harnessing the performance gain brought by KV4 quantization. As a result, QServe improves the maximum achievable serving throughput of Llama-3-8B by 1.2x on A100, 1.4x on L40S; and Qwen1.5-72B by 2.4x on A100, 3.5x on L40S, compared to TensorRT-LLM. Remarkably, QServe on L40S GPU can achieve even higher throughput than TensorRT-LLM on A100. Thus, QServe effectively reduces the dollar cost of LLM serving by 3x. Code is available at https://github.com/mit-han-lab/qserve.

Prefill-decode (PD) disaggregation has become the standard architecture for large-scale LLM serving, but in practice its deployment boundary is still determined by KVCache transfer. In conventional dense-attention models, prefill generates huge KVCache traffics that keep prefill and decode tightly coupled within a single high-bandwidth network domain, limiting heterogeneous deployment and resource elasticity. Recent hybrid-attention architectures substantially reduce KVCache size, making cross-cluster KVCache transport increasingly plausible. However, smaller KVCache alone does not make heterogeneous cross-datacenter PD serving practical: real workloads remain bursty, request lengths are highly skewed, prefix caches are unevenly distributed, and inter-cluster bandwidth fluctuates. A naive design that fully externalizes prefill can therefore still suffer from congestion, unstable queueing, and poor utilization. We present Prefill-as-a-Service (PrfaaS), a cross-datacenter serving architecture that selectively offloads long-context prefill to standalone, compute-dense prefill clusters and transfers the resulting KVCache over commodity Ethernet to local PD clusters for decode. Rather than treating reduced KVCache as sufficient, PrfaaS combines model-side KV efficiency with system-side selective offloading, bandwidth-aware scheduling, and cache-aware request placement. This design removes the requirement that heterogeneous accelerators share the same low-latency RDMA fabric, enabling independent scaling of prefill and decode capacity across loosely coupled clusters. In a case study using an internal 1T-parameter hybrid model, a PrfaaS-augmented heterogeneous deployment achieves 54% higher serving throughput and 64% lower P90 TTFT than a homogeneous PD baseline, with approximately 15% throughput gain at equal cost, while consuming only modest cross-datacenter bandwidth.

This technical report introduces Aya 23, a family of multilingual language models. Aya 23 builds on the recent release of the Aya model (\"Ust\"un et al., 2024), focusing on pairing a highly performant pre-trained model with the recently released Aya collection (Singh et al., 2024). The result is a powerful multilingual large language model serving 23 languages, expanding state-of-art language modeling capabilities to approximately half of the world's population. The Aya model covered 101 languages whereas Aya 23 is an experiment in depth vs breadth, exploring the impact of allocating more capacity to fewer languages that are included during pre-training. Aya 23 outperforms both previous massively multilingual models like Aya 101 for the languages it covers, as well as widely used models like Gemma, Mistral and Mixtral on an extensive range of discriminative and generative tasks. We release the open weights for both the 8B and 35B models as part of our continued commitment for expanding access to multilingual progress.

Serving large language models (LLMs) at scale necessitates efficient key-value (KV) cache management. KV caches can be reused across conversation turns via shared-prefix prompts that are common in iterative code editing and chat. However, stale caches consume scarce GPU memory, require offloading, or force recomputation. We present KVTC, a lightweight transform coder that compresses KV caches for compact on-GPU and off-GPU storage. Drawing on classical media compression, KVTC combines PCA-based feature decorrelation, adaptive quantization, and entropy coding. It requires only a brief initial calibration and leaves model parameters unchanged. By exploiting redundancies in KV caches, KVTC achieves up to 20times compression while maintaining reasoning and long-context accuracy, and 40times or higher for specific use cases. We test KVTC with Llama 3, Mistral NeMo, and R1-Qwen 2.5 models across benchmarks including AIME25, LiveCodeBench, GSM8K, MMLU, Qasper, RULER, and MATH-500. It consistently outperforms inference-time baselines such as token eviction, quantization, and SVD-based methods, while achieving higher compression ratios. These results support KVTC as a practical building block for memory-efficient LLM serving with reusable KV caches.

Hybrid models that combine the language modeling capabilities of Attention layers with the efficiency of Recurrent layers (e.g., State Space Models) have gained traction in practically supporting long contexts in Large Language Model serving. Yet, the unique properties of these models complicate the usage of complementary efficiency optimizations such as prefix caching that skip redundant computations across requests. Most notably, their use of in-place state updates for recurrent layers precludes rolling back cache entries for partial sequence overlaps, and instead mandates only exact-match cache hits; the effect is a deluge of (large) cache entries per sequence, most of which yield minimal reuse opportunities. We present Marconi, the first system that supports efficient prefix caching with Hybrid LLMs. Key to Marconi are its novel admission and eviction policies that more judiciously assess potential cache entries based not only on recency, but also on (1) forecasts of their reuse likelihood across a taxonomy of different hit scenarios, and (2) the compute savings that hits deliver relative to memory footprints. Across diverse workloads and Hybrid models, Marconi achieves up to 34.4times higher token hit rates (71.1% or 617 ms lower TTFT) compared to state-of-the-art prefix caching systems.

The reasoning large language model (RLLM) has been proven competitive in solving complex reasoning tasks such as mathematics, coding, compared to general LLM. However, the serving performance and behavior of RLLM remains unexplored, which may undermine the deployment and utilization of RLLM in real-world scenario. To close this gap, in this paper, we conduct a comprehensive study of RLLM service. We first perform a pilot study on comparing the serving performance between RLLM and traditional LLM and reveal that there are several distinct differences regarding serving behavior: (1) significant memory usage and fluctuations; (2) straggler requests; (3) adaptive running time; (4) domain preference. Then we further investigate whether existing inference optimization techniques are valid for RLLM. Our main takeaways are that model quantization methods and speculative decoding can improve service system efficiency with small compromise to RLLM accuracy, while prefix caching, KV cache quantization may even degrade accuracy or serving performance for small RLLM. Lastly, we conduct evaluation under real world workload modeled by Gamma distribution to verify our findings. Empirical results of real world workload evaluation across different dataset are aligned with our main findings regarding RLLM serving. We hope our work can provide the research community and industry with insights to advance RLLM inference serving.

The increasing demand for large language model (LLM) serving has necessitated significant advancements in the optimization and profiling of LLM inference systems. As these models become integral to a wide range of applications, the need for efficient and scalable serving solutions has grown exponentially. This work introduces TokenSim, a comprehensive hardware and software exploration system designed specifically for LLM inference. TokenSim is characterized by its support for extensible system optimizations including scheduling and memory management. We validate the results with systems running with realworld datasets, achieving an error rate of less than 1%. Furthermore, TokenSim facilitates various insightful explorations into the performance and optimization of LLM serving systems.

Large language model (LLM) serving has transformed from stateless to stateful systems, utilizing techniques like context caching and disaggregated inference. These optimizations extend the lifespan and domain of the KV cache, necessitating a new architectural approach. We present MemServe, a unified system that integrates both inter-request and intra-request optimizations. MemServe introduces MemPool, an elastic memory pool managing distributed memory and KV caches across serving instances. Using MemPool APIs, MemServe combines context caching with disaggregated inference for the first time, supported by a global scheduler that enhances cache reuse through a global prompt tree-based locality-aware policy. Tests show that MemServe significantly improves job completion time and time-to-first-time.

Large language model (LLM) serving infrastructures are undergoing a shift toward heterogeneity and disaggregation. Modern deployments increasingly integrate diverse accelerators and near-memory processing technologies, introducing significant hardware heterogeneity, while system software increasingly separates computation, memory, and model components across distributed resources to improve scalability and efficiency. As a result, LLM serving performance is no longer determined by hardware or software choices in isolation, but by their runtime interaction through scheduling, data movement, and interconnect behavior. However, understanding these interactions remains challenging, as existing simulators lack the ability to jointly model heterogeneous hardware and disaggregated serving techniques within a unified, runtime-driven framework. This paper presents LLMServingSim 2.0, a unified system-level simulator designed to make runtime-driven hardware-software interactions in heterogeneous and disaggregated LLM serving infrastructures explicit and analyzable. LLMServingSim 2.0 embeds serving decisions and hardware behavior into a single runtime loop, enabling interaction-aware modeling of batching, routing, offloading, memory, and power. The simulator supports extensible integration of emerging accelerators and memory systems through profile-based modeling, while capturing dynamic serving behavior and system-level effects. We validate LLMServingSim 2.0 against real deployments, showing that it reproduces key performance, memory, and power metrics with an average error of 0.95%, while maintaining simulation times of around 10 minutes even for complex configurations. These results demonstrate that LLMServingSim 2.0 provides a practical bridge between hardware innovation and serving-system design, enabling systematic exploration and co-design for next-generation LLM serving infrastructures.

Large Language Model (LLM) serving has become a critical cloud workload, and realistic traces are essential for motivating and benchmarking serving systems. However, existing LLM serving workload studies remain limited in scale and scope. They often observe short time periods and provide limited visibility into how users interact with models in production. As a result, they do not fully capture how LLM serving workloads evolve over time or how user-model interactions shape production traffic. In this work, we further the understanding of real-world LLM serving workloads through both a global characterization and a longitudinal study of a one-year production trace from CompanyX. Unlike prior studies, our trace captures full production behavior across many models and users, including both popular and long-tail models. We analyze the workload from aggregate, temporal, model-level, and user-level perspectives, revealing workload evolution and user-model structure that are typically hidden behind aggregate views. To support future research, we publicly release the full one-year trace, enabling downstream studies of production behavior without relying on sampled or synthetically generated workloads. The trace is available at https://github.com/HarvardMadSys/chutes_workload.

Large language model (LLM) serving requires substantial memory to store model weights and KV caches. As models grow larger and contexts become longer, memory capacity and bandwidth increasingly become bottlenecks for serving performance. Agentic workloads compound this pressure through repeated interactions over growing contexts, making it increasingly important to retain KV state for reuse. High-bandwidth flash (HBF) offers a way to expand accelerator memory capacity for large language model (LLM) serving, but its access costs and limited write endurance complicate its use. We evaluate HBF for high-throughput agentic serving across system design and scheduling choices to understand when additional capacity improves serving performance and energy efficiency. We introduce an HBM-HBF-host hierarchical storage system and buffered cache-aware scheduling, and use trace-driven simulations to analyze their effects on performance, energy consumption, and HBF write lifetime. Across the evaluated workloads, the fastest HBF-augmented systems reduce completion time by 36.1-87.7% relative to HBM-only systems. Modeled energy savings reach 59.1%, with benefits depending on the workload and weight placement. Buffered cache-aware scheduling extends estimated HBF write lifetime from 1.21 to 14.82 years in the evaluated configuration. These results demonstrate the importance of coordinating data placement and scheduling to improve serving efficiency while sustaining a practical HBF write lifetime.

Existing large language model (LLM) serving systems fall into two categories: 1) a unified system where prefill phase and decode phase are co-located on the same GPU, sharing the unified computational resource and storage, and 2) a disaggregated system where the two phases are disaggregated to different GPUs. The design of the disaggregated system addresses the latency interference and sophisticated scheduling issues in the unified system but leads to storage challenges including 1) replicated weights for both phases that prevent flexible deployment, 2) KV cache transfer overhead between the two phases, 3) storage imbalance that causes substantial wasted space of the GPU capacity, and 4) suboptimal resource adjustment arising from the difficulties in migrating KV cache. Such storage inefficiency delivers poor serving performance under high request rates. In this paper, we identify that the advantage of the disaggregated system lies in the disaggregated computation, i.e., partitioning the computational resource to enable the asynchronous computation of two phases. Thus, we propose a novel LLM serving system, semi-PD, characterized by disaggregated computation and unified storage. In semi-PD, we introduce a computation resource controller to achieve disaggregated computation at the streaming multi-processor (SM) level, and a unified memory manager to manage the asynchronous memory access from both phases. semi-PD has a low-overhead resource adjustment mechanism between the two phases, and a service-level objective (SLO) aware dynamic partitioning algorithm to optimize the SLO attainment. Compared to state-of-the-art systems, semi-PD maintains lower latency at higher request rates, reducing the average end-to-end latency per request by 1.27-2.58x on DeepSeek series models, and serves 1.55-1.72x more requests adhering to latency constraints on Llama series models.

The complexity of large language model (LLM) serving workloads has substantially increased due to the integration with external tool invocations, such as ChatGPT plugins. In this paper, we identify a new opportunity for efficient LLM serving for requests that trigger tools: tool partial execution alongside LLM decoding. To this end, we design Conveyor, an efficient LLM serving system optimized for handling requests involving external tools. We introduce a novel interface for tool developers to expose partial execution opportunities to the LLM serving system and a request scheduler that facilitates partial tool execution. Our results demonstrate that tool partial execution can improve request completion latency by up to 38.8%.

As the input length of large language model (LLM) serving continues to grow, the KV cache has become a dominant bottleneck in AI infrastructure. It limits GPU memory capacity, serving concurrency, cache reuse, and distributed scalability. Multiple important problems, including position-independent KV cache, prefix KV cache compression, hot/cold KV cache separation, and distributed KV cache management, all depend on how the KV cache is represented and managed. However, existing serving systems largely rely on a monolithic KV cache abstraction, where the KV cache is treated as a homogeneous sequence of token-level memory blocks and managed with similar policies across attention heads and serving scenarios. We observe that KV cache utility is highly structured across KV heads: different heads exhibit different functional roles, attention distances, and runtime importance. Therefore, a full KV cache is not always necessary for every head, token range, or serving scenario. We present RedKnot, a head-aware KV cache management system for LLM serving. RedKnot breaks the conventional monolithic KV cache abstraction by decomposing the KV cache along KV heads, whose importance and effective attention ranges vary significantly across serving scenarios. This head-level decomposition turns the KV cache from a monolithic tensor abstraction into a structured memory object, enabling RedKnot to uniformly support position-independent KV reuse, prefix KV compression, hot/cold KV separation, and distributed KV placement while preserving output fidelity and improving resource efficiency, without requiring model retraining or fine-tuning. RedKnot establishes a new foundation for AI infrastructure by transforming the KV cache from a monolithic, passive runtime artifact into a dynamic, model-aware runtime substrate for scalable LLM serving.

In the context of Machine Learning as a Service (MLaaS) clouds, the extensive use of Large Language Models (LLMs) often requires efficient management of significant query loads. When providing real-time inference services, several challenges arise. Firstly, increasing the number of GPUs may lead to a decrease in inference speed due to heightened communication overhead, while an inadequate number of GPUs can lead to out-of-memory errors. Secondly, different deployment strategies need to be evaluated to guarantee optimal utilization and minimal inference latency. Lastly, inefficient orchestration of inference queries can easily lead to significant Service Level Objective (SLO) violations. Lastly, inefficient orchestration of inference queries can easily lead to significant Service Level Objective (SLO) violations. To address these challenges, we propose a Unified and Efficient approach for Large Language Model inference serving (UELLM), which consists of three main components: 1) resource profiler, 2) batch scheduler, and 3) LLM deployer. UELLM minimizes resource overhead, reduces inference latency, and lowers SLO violation rates. Compared with state-of-the-art (SOTA) techniques, UELLM reduces the inference latency by 72.3% to 90.3%, enhances GPU utilization by 1.2X to 4.1X, and increases throughput by 1.92X to 4.98X, it can also serve without violating the inference latency SLO.

Deploying multiple models within shared GPU clusters is a key strategy to improve resource efficiency in large language model (LLM) serving. Existing multi-LLM serving systems improve GPU utilization at the cost of degraded inference performance, particularly time-to-first-token (TTFT). We attribute this degradation to the lack of awareness regarding future workload characteristics. In contrast, recent analyses have shown the strong periodicity and long-term predictability of real-world LLM serving workloads. In this paper, we propose one-for-many GPU prewarming, which proactively loads parameters from multiple models onto GPUs based on workload forecasts. These prewarmed weights enable the system to promptly instantiate serving instances upon encountering request bursts. We design and implement WarmServe, a multi-LLM serving system incorporating three key techniques: (1) a model placement algorithm that optimizes prewarming decisions to minimize cross-model prewarming interference, (2) a KV cache reservation strategy that repurposes idle KV cache space on running GPUs for prewarming new models, and (3) an efficient GPU memory switching mechanism for tensor management. Evaluation on real-world datasets shows that WarmServe reduces tail TTFT by up to 50.8times compared to the state-of-the-art autoscaling-based system, while supporting up to 2.5times higher request throughput than the GPU-sharing system.

Large Language Model (LLM) serving faces a fundamental tension between stringent latency Service Level Objectives (SLOs) and limited GPU memory capacity. When high request rates exhaust the KV cache budget, existing LLM inference systems often suffer severe head-of-line (HOL) blocking. While prior work explored PCIe-based offloading, these approaches cannot sustain responsiveness under high request rates, often failing to meet tight Time-To-First-Token (TTFT) and Time-Between-Tokens (TBT) SLOs. We present SuperInfer, a high-performance LLM inference system designed for emerging Superchips (e.g., NVIDIA GH200) with tightly coupled GPU-CPU architecture via NVLink-C2C. SuperInfer introduces RotaSched, the first proactive, SLO-aware rotary scheduler that rotates requests to maintain responsiveness on Superchips, and DuplexKV, an optimized rotation engine that enables full-duplex transfer over NVLink-C2C. Evaluations on GH200 using various models and datasets show that SuperInfer improves TTFT SLO attainment rates by up to 74.7% while maintaining comparable TBT and throughput compared to state-of-the-art systems, demonstrating that SLO-aware scheduling and memory co-design unlocks the full potential of Superchips for responsive LLM serving.

With the ubiquitous use of modern large language models (LLMs) across industries, the inference serving for these models is ever expanding. Given the high compute and memory requirements of modern LLMs, more and more top-of-the-line GPUs are being deployed to serve these models. Energy availability has come to the forefront as the biggest challenge for data center expansion to serve these models. In this paper, we present the trade-offs brought up by making energy efficiency the primary goal of LLM serving under performance SLOs. We show that depending on the inputs, the model, and the service-level agreements, there are several knobs available to the LLM inference provider to use for being energy efficient. We characterize the impact of these knobs on the latency, throughput, as well as the energy. By exploring these trade-offs, we offer valuable insights into optimizing energy usage without compromising on performance, thereby paving the way for sustainable and cost-effective LLM deployment in data center environments.

Autoregressive decoding remains a primary bottleneck in large language model (LLM) serving, motivating speculative decoding methods that reduce expensive teacher-model invocations by verifying multiple candidate tokens per step. Tree-structured speculation further increases parallelism, but is often brittle when ported across heterogeneous backends and accelerator stacks, where attention masking, KV-cache layouts, and indexing semantics are not interchangeable. We present EAGLE-Pangu, a reproducible system that ports EAGLE-3-style tree speculative decoding to a Pangu teacher backend on Ascend NPUs. EAGLE-Pangu contributes (i) an explicit branch/commit cache manager built on the Cache API, (ii) accelerator-safe tree tensorization that removes undefined negative indices by construction and validates structural invariants, and (iii) a fused-kernel-compatible teacher verification path with a debuggable eager fallback. On 240 turns from MT-Bench and HumanEval-style prompts, EAGLE-Pangu improves end-to-end decoding throughput by 1.27x on average, up to 2.46x at p99, over teacher-only greedy decoding in the fused-kernel performance path. We also provide a fused-kernel-free reference path with structured traces and invariant checks to support reproducible debugging and ablation across execution modes and tree budgets.

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.

Our work tackles the challenge of securing user inputs in cloud-hosted large language model (LLM) serving while ensuring output invariance, model confidentiality, and compute efficiency. We introduce secure multi-party decoding (SMD), which leverages confidential computing to confine user prompts to a trusted execution environment (TEE), namely a confidential virtual machine (CVM), while allowing service providers to generate tokens efficiently. We also introduce a novel cryptographic method, prompt obfuscation (PO), to ensure robustness against reconstruction attacks on SMD. We demonstrate that our approach preserves both prompt confidentiality and LLM serving efficiency. Our solution can enable privacy-preserving cloud LLM serving that handles sensitive prompts, such as clinical records, financial data, and personal information.

Large language model (LLM) inference serving is increasingly constrained by memory rather than compute. As long-context and long-form reasoning workloads become more prevalent, the key-value (KV) cache dominates both memory footprint and memory traffic during LLM token generation, i.e., decode. In particular, HBM capacity has become a scarce and costly resource that heavily limits inference batch size and system throughput. This paper presents OasisKV, a memory-centric LLM inference system design that alleviates HBM capacity pressure by decoupling full KV-cache storage from HBM during LLM decoding. Because decode-time attention is naturally sparse, OasisKV keeps only the KV entries of the most relevant tokens in HBMs for attention computation. We observe that future important tokens can be predicted accurately in advance using lookahead tokens drafted by speculative decoding (SD). OasisKV employs an efficient attention background pipeline to identify important KV blocks. They are then prefetched from higher-capacity memory tiers (e.g., host or remote memory) and staged in HBMs before being used in the next decode step. We implement OasisKV based on vLLM. The lookahead prediction is accurate enough to keep accuracy within 0.7 points of full attention under a 2,048-token KV budget. This lets OasisKV turn sparsity into throughput gain: 1.69times over dense vLLM on the reasoning workload at 0.1 points of accuracy loss, and up to 2.1times on multi-GPU long-context serving. Under prefill--decode disaggregation, OasisKV reaches about 2times dense throughput while admitting each request with 6.5--9.7times less KV and holding 2.2-2.6 less decode-node host memory than full KV transfer.

Large language model (LLM) agents serving different users often solve related tasks, yet separate user histories can leave reusable experience inaccessible to other agents. Pooling memories expands access but risks transferring preferences that conflict with the receiving user's requirements. We introduce ShareMem, a memory architecture that shares reusable experience while grounding its application in the receiving user's own preferences. Shared experiences indicate how to act and which preferences to consult; the receiving user's memory supplies their concrete values. Two-stage consolidation refines experience locally before integrating accepted edits into a shared pool. During execution, scope-first retrieval jointly selects local and shared experiences under a common entry budget, while a user-bound channel supports initial and agent-initiated preference retrieval. We evaluate ShareMem across web navigation (Mind2Web), online personalized interaction (VitaBench~2.0), and multi-session coding (MemoryCode) with four backbone models. It improves step success, average task success, and dialogue-macro coding scores, respectively, over matched user-local memory across all four models. Ablations favor two-stage consolidation for smaller shared pools, lower induction token usage, and better downstream performance, and support complementarity between experience guidance and active preference retrieval. Further analyses show that sharing helps most when relevant local experience is scarce, while source quality and cross-user preference interference limit useful transfer.

Transformer-based large language model (LLM) inference serving is now the backbone of many cloud services. LLM inference consists of a prefill phase and a decode phase. However, existing LLM deployment practices often overlook the distinct characteristics of these phases, leading to significant interference. To mitigate interference, our insight is to carefully schedule and group inference requests based on their characteristics. We realize this idea in TetriInfer through three pillars. First, it partitions prompts into fixed-size chunks so that the accelerator always runs close to its computationsaturated limit. Second, it disaggregates prefill and decode instances so each can run independently. Finally, it uses a smart two-level scheduling algorithm augmented with predicted resource usage to avoid decode scheduling hotspots. Results show that TetriInfer improves time-to-first-token (TTFT), job completion time (JCT), and inference efficiency in turns of performance per dollar by a large margin, e.g., it uses 38% less resources all the while lowering average TTFT and average JCT by 97% and 47%, respectively.

Serving disaggregated large language models (LLMs) over tens of thousands of xPU devices (GPUs or NPUs) with reliable performance faces multiple challenges. 1) Ignoring the diversity (various prefixes and tidal requests), treating all the prompts in a mixed pool is inadequate. To facilitate the similarity per scenario and minimize the inner mismatch on P/D (prefill and decoding) processing, fine-grained organization is required, dynamically adjusting P/D ratios for better performance. 2) Due to inaccurate estimation on workload (queue status or maintained connections), the global scheduler easily incurs unnecessary timeouts in prefill. 3) Block-fixed device-to-device (D2D) KVCache transfer over cluster-level RDMA (remote direct memory access) fails to achieve desired D2D utilization as expected. To overcome previous problems, this paper proposes an end-to-end system P/D-Serve, complying with the paradigm of MLOps (machine learning operations), which models end-to-end (E2E) P/D performance and enables: 1) fine-grained P/D organization, mapping the service with RoCE (RDMA over converged ethernet) as needed, to facilitate similar processing and dynamic adjustments on P/D ratios; 2) on-demand forwarding upon rejections for idle prefill, decoupling the scheduler from regular inaccurate reports and local queues, to avoid timeouts in prefill; and 3) efficient KVCache transfer via optimized D2D access. P/D-Serve is implemented upon Ascend and MindSpore, has been deployed over tens of thousands of NPUs for more than eight months in commercial use, and further achieves 60\%, 42\% and 46\% improvements on E2E throughput, time-to-first-token (TTFT) SLO (service level objective) and D2D transfer time. As the E2E system with optimizations, P/D-Serve achieves 6.7x increase on throughput, compared with aggregated LLMs.

Parameter-efficient finetuning (PEFT) is a widely used technique to adapt large language models for different tasks. Service providers typically create separate systems for users to perform PEFT model finetuning and inference tasks. This is because existing systems cannot handle workloads that include a mix of inference and PEFT finetuning requests. As a result, shared GPU resources are underutilized, leading to inefficiencies. To address this problem, we present FlexLLM, the first system that can serve inference and parameter-efficient finetuning requests in the same iteration. Our system leverages the complementary nature of these two tasks and utilizes shared GPU resources to run them jointly, using a method called co-serving. To achieve this, FlexLLM introduces a novel token-level finetuning mechanism, which breaks down the finetuning computation of a sequence into smaller token-level computations and uses dependent parallelization and graph pruning, two static compilation optimizations, to minimize the memory overhead and latency for co-serving. Compared to existing systems, FlexLLM's co-serving approach reduces the activation GPU memory overhead by up to 8x, and the end-to-end GPU memory requirement of finetuning by up to 36% while maintaining a low inference latency and improving finetuning throughput. For example, under a heavy inference workload, FlexLLM can still preserve more than 80% of the peak finetuning throughput, whereas existing systems cannot make any progress with finetuning. The source code of FlexLLM is publicly available at https://github.com/flexflow/FlexFlow.

Large language models (LLMs) have revolutionized the field of natural language processing, extending their strong capabilities into multi-modal domains. Thus, it is vital to define proper and diversified metrics for the evaluation of LLMs. In this paper, we introduce matrix entropy, a novel metric rooted in information theory and geometry principles to quantify the data compression proficiency in LLMs. It reflects the model's ability to extract relevant information and eliminate unnecessary elements, thereby providing insight into the language model's intrinsic capability. Specifically, we demonstrate its applicability in both single-modal (language) and multi-modal settings. For language models, our findings reveal that the matrix entropy of representations follows a scaling law type reduction when the model scales up, serving as a complement to the traditional loss scaling law. For the multi-modal setting, we also propose an evaluation method based on matrix entropy for assessing alignment quality and we find that modern large multi-modal models exhibit great alignment performance.

Transformer-based large language models (LLMs) exhibit impressive performance in generative tasks but introduce significant challenges in real-world serving due to inefficient use of the expensive, computation-optimized accelerators. This mismatch arises from the autoregressive nature of LLMs, where the generation phase comprises operators with varying resource demands. Specifically, the attention operator is memory-intensive, exhibiting a memory access pattern that clashes with the strengths of modern accelerators, especially as context length increases. To enhance the efficiency and cost-effectiveness of LLM serving, we introduce the concept of attention offloading. This approach leverages a collection of cheap, memory-optimized devices for the attention operator while still utilizing high-end accelerators for other parts of the model. This heterogeneous setup ensures that each component is tailored to its specific workload, maximizing overall performance and cost efficiency. Our comprehensive analysis and experiments confirm the viability of splitting the attention computation over multiple devices. Also, the communication bandwidth required between heterogeneous devices proves to be manageable with prevalent networking technologies. To further validate our theory, we develop Lamina, an LLM inference system that incorporates attention offloading. Experimental results indicate that Lamina can provide 1.48x-12.1x higher estimated throughput per dollar than homogeneous solutions.

LLM serving systems typically treat user prompts as monolithic inputs, optimizing inference through decoding tricks or inter-query batching. However, many real-world prompts contain latent semantic parallelism--decomposable structures where subtasks can be executed independently to reduce latency while preserving meaning. We introduce PARALLELPROMPT, the first benchmark for measuring intra-query parallelism in natural user prompts. Our dataset comprises over 37,000 real-world prompts from public LLM chat logs, each annotated with a structured schema capturing task templates, shared context, and iteration inputs. These schemas are extracted using LLM-assisted prompting with rule-based multilingual validation. To evaluate the benefits of decomposition, we provide an execution suite that benchmarks serial vs. parallel strategies, measuring latency, structural adherence, and semantic fidelity. Our results show that intra-query parallelism can be successfully parsed in over 75% of curated datasets, unlocking up to 5x speedups on tasks like translation, comprehension, and comparative analysis, with minimal quality degradation. By releasing this benchmark, curation pipeline, and evaluation suite, we provide the first standardized testbed for studying structure-aware execution in LLM serving pipelines.

Large language models (LLMs), with their powerful generative capabilities and vast knowledge, empower various tasks in everyday life. However, these abilities are primarily concentrated in high-resource languages, leaving low-resource languages with weaker generative capabilities and relatively limited knowledge. Enhancing the multilingual capabilities of LLMs is therefore crucial for serving over 100 linguistic communities worldwide. An intuitive approach to enhance the multilingual capabilities would be to construct instruction data for various languages, but constructing instruction data for over 100 languages is prohibitively costly. In this paper, we introduce BayLing 2, which efficiently transfers generative capabilities and knowledge from high-resource languages to low-resource languages through language alignment. To achieve this, we constructed a dataset of 3.2 million instructions, comprising high-resource language instructions (Chinese and English) and cross-lingual instructions for 100+ languages and performed instruction tuning based on the dataset to facilitate the capability transfer between languages. Using Llama as the foundation model, we developed BayLing-2-7B, BayLing-2-13B, and BayLing-2-8B, and conducted a comprehensive evaluation of BayLing. For multilingual translation across 100+ languages, BayLing shows superior performance compared to open-source models of similar scale. For multilingual knowledge and understanding benchmarks, BayLing achieves significant improvements across over 20 low-resource languages, demonstrating its capability of effective knowledge transfer from high-resource to low-resource languages. Furthermore, results on English benchmarks indicate that BayLing maintains high performance in highresource languages while enhancing the performance in low-resource languages. Demo, homepage, code and models of BayLing are available.

Mixed-precision inference techniques reduce the memory and computational demands of Large Language Models (LLMs) by applying hybrid precision formats to model weights, activations, and KV caches. This work introduces mixed-precision LLM inference techniques that encompass (i) systematic memory and compute optimization across hierarchical storage and tensor core architectures, and (ii) comprehensive end-to-end mixed-precision optimization across diverse precision formats and hardware configurations. Our approach features two novel mixed-precision pipelines designed for optimal hardware utilization: a General Matrix Multiply (GEMM) pipeline that optimizes matrix operations through offline weight packing and online acceleration, and an attention pipeline that enables efficient attention computation with arbitrary Query, Key, and Value precision combinations. The key implementation of the pipelines includes (i) hardware-aware weight packing for automatic format optimization, (ii) adaptive head alignment for efficient attention computation, (iii) instruction-level parallelism for memory hierarchy exploitation, and (iv) KV memory loading pipeline for enhanced inference efficiency. We conduct comprehensive evaluations across 16 popular LLMs and 4 representative GPU architectures. Results demonstrate that our approach achieves up to 61% lower serving latency (30% on average) and up to 156% higher throughput (58% on average) in mixed-precision workloads compared to existing mixed-precision frameworks, establishing consistent performance improvements across all tested configurations and hardware types. This work is integrated into TurboMind, a high-performance inference engine of the LMDeploy project, which is open-sourced and publicly available at https://github.com/InternLM/lmdeploy.

Intelligent agents stand out as a potential path toward artificial general intelligence (AGI). Thus, researchers have dedicated significant effort to diverse implementations for them. Benefiting from recent progress in large language models (LLMs), LLM-based agents that use universal natural language as an interface exhibit robust generalization capabilities across various applications -- from serving as autonomous general-purpose task assistants to applications in coding, social, and economic domains, LLM-based agents offer extensive exploration opportunities. This paper surveys current research to provide an in-depth overview of LLM-based intelligent agents within single-agent and multi-agent systems. It covers their definitions, research frameworks, and foundational components such as their composition, cognitive and planning methods, tool utilization, and responses to environmental feedback. We also delve into the mechanisms of deploying LLM-based agents in multi-agent systems, including multi-role collaboration, message passing, and strategies to alleviate communication issues between agents. The discussions also shed light on popular datasets and application scenarios. We conclude by envisioning prospects for LLM-based agents, considering the evolving landscape of AI and natural language processing.

The integration of a complex set of Electronic Design Automation (EDA) tools to enhance interoperability is a critical concern for circuit designers. Recent advancements in large language models (LLMs) have showcased their exceptional capabilities in natural language processing and comprehension, offering a novel approach to interfacing with EDA tools. This research paper introduces ChatEDA, an autonomous agent for EDA empowered by a large language model, AutoMage, complemented by EDA tools serving as executors. ChatEDA streamlines the design flow from the Register-Transfer Level (RTL) to the Graphic Data System Version II (GDSII) by effectively managing task planning, script generation, and task execution. Through comprehensive experimental evaluations, ChatEDA has demonstrated its proficiency in handling diverse requirements, and our fine-tuned AutoMage model has exhibited superior performance compared to GPT-4 and other similar LLMs.

As Large Language Models (LLMs) have made significant advancements across various tasks, such as question answering, translation, text summarization, and dialogue systems, the need for accuracy in information becomes crucial, especially for serious financial products serving billions of users like Alipay. To address this, Alipay has developed a Retrieval-Augmented Generation (RAG) system that grounds LLMs on the most accurate and up-to-date information. However, for a real-world product serving millions of users, the inference speed of LLMs becomes a critical factor compared to a mere experimental model. Hence, this paper presents a generic framework for accelerating the inference process, resulting in a substantial increase in speed and cost reduction for our RAG system, with lossless generation accuracy. In the traditional inference process, each token is generated sequentially by the LLM, leading to a time consumption proportional to the number of generated tokens. To enhance this process, our framework, named lookahead, introduces a multi-branch strategy. Instead of generating a single token at a time, we propose a Trie-based Retrieval (TR) process that enables the generation of multiple branches simultaneously, each of which is a sequence of tokens. Subsequently, for each branch, a Verification and Accept (VA) process is performed to identify the longest correct sub-sequence as the final output. Our strategy offers two distinct advantages: (1) it guarantees absolute correctness of the output, avoiding any approximation algorithms, and (2) the worst-case performance of our approach is equivalent to the conventional process. We conduct extensive experiments to demonstrate the significant improvements achieved by applying our inference acceleration framework. Code is avaliable: https://github.com/alipay/PainlessInferenceAcceleration.

Despite their sophisticated capabilities, large language models (LLMs) encounter a major hurdle in effective assessment. This paper first revisits the prevalent evaluation method-multiple choice question answering (MCQA), which allows for straightforward accuracy measurement. Through a comprehensive evaluation of 24 models across 11 benchmarks, we highlight several potential drawbacks of MCQA, for instance, the inconsistency between the MCQA evaluation and the generation of open-ended responses in practical scenarios. In response, we introduce an RWQ-Elo rating system, engaging 24 LLMs such as GPT-4, GPT-3.5, Google-Gemini-Pro and LLaMA-1/-2, in a two-player competitive format, with GPT-4 serving as the judge. Each LLM receives an Elo rating thereafter. This system is designed to mirror real-world usage, and for this purpose, we have compiled a new benchmark called ``Real-world questions'' (RWQ), comprising 20,772 authentic user inquiries. Additionally, we thoroughly analyze the characteristics of our system and compare it with prior leaderboards like AlpacaEval and MT-Bench. Our analysis reveals the stability of our RWQ-Elo system, the feasibility of registering new models, and its potential to reshape LLM leaderboards.

Confidential computing on GPUs, like NVIDIA H100, mitigates the security risks of outsourced Large Language Models (LLMs) by implementing strong isolation and data encryption. Nonetheless, this encryption incurs a significant performance overhead, reaching up to 52.8 percent and 88.2 percent throughput drop when serving OPT-30B and OPT-66B, respectively. To address this challenge, we introduce PipeLLM, a user-transparent runtime system. PipeLLM removes the overhead by overlapping the encryption and GPU computation through pipelining - an idea inspired by the CPU instruction pipelining - thereby effectively concealing the latency increase caused by encryption. The primary technical challenge is that, unlike CPUs, the encryption module lacks prior knowledge of the specific data needing encryption until it is requested by the GPUs. To this end, we propose speculative pipelined encryption to predict the data requiring encryption by analyzing the serving patterns of LLMs. Further, we have developed an efficient, low-cost pipeline relinquishing approach for instances of incorrect predictions. Our experiments on NVIDIA H100 GPU show that compared with vanilla systems without confidential computing (e.g., vLLM, PEFT, and FlexGen), PipeLLM incurs modest overhead (less than 19.6 percent in throughput) across various LLM sizes, from 13B to 175B.

Recent prevailing works on graph machine learning typically follow a similar methodology that involves designing advanced variants of graph neural networks (GNNs) to maintain the superior performance of GNNs on different graphs. In this paper, we aim to streamline the GNN design process and leverage the advantages of Large Language Models (LLMs) to improve the performance of GNNs on downstream tasks. We formulate a new paradigm, coined "LLMs-as-Consultants," which integrates LLMs with GNNs in an interactive manner. A framework named LOGIN (LLM Consulted GNN training) is instantiated, empowering the interactive utilization of LLMs within the GNN training process. First, we attentively craft concise prompts for spotted nodes, carrying comprehensive semantic and topological information, and serving as input to LLMs. Second, we refine GNNs by devising a complementary coping mechanism that utilizes the responses from LLMs, depending on their correctness. We empirically evaluate the effectiveness of LOGIN on node classification tasks across both homophilic and heterophilic graphs. The results illustrate that even basic GNN architectures, when employed within the proposed LLMs-as-Consultants paradigm, can achieve comparable performance to advanced GNNs with intricate designs. Our codes are available at https://github.com/QiaoYRan/LOGIN.

While foundation models update slowly due to resource-intensive training requirements, domain-specific models evolve between updates. Model merging aims to combine multiple expert models into a single, more capable model, thereby reducing storage and serving costs while supporting decentralized model development. Despite its potential, previous studies have primarily focused on merging visual classification models or Large Language Models (LLMs) for code and math tasks. Multimodal Large Language Models (MLLMs), which extend the capabilities of LLMs through large-scale multimodal training, have gained traction. However, there lacks a benchmark for model merging research that clearly divides the tasks for MLLM training and evaluation. In this paper, (i) we introduce the model merging benchmark for MLLMs, which includes multiple tasks such as VQA, Geometry, Chart, OCR, and Grounding, providing both LoRA and full fine-tuning models. Moreover, we explore how model merging can combine different modalities (e.g., vision-language, audio-language, and video-language models), moving toward the Omni-language model. (ii) We implement 10 model merging algorithms on the benchmark. Furthermore, we propose a novel method that removes noise from task vectors and robustly optimizes the merged vector based on a loss defined over task vector interactions, achieving an average performance gain of 2.48%. (iii) We find that model merging offers a promising way for building improved MLLMs without requiring data training. Our results also demonstrate that the complementarity among multiple modalities outperforms individual modalities.

In the era of large models, the autoregressive nature of decoding often results in latency serving as a significant bottleneck. We propose a non-autoregressive LM-fused ASR system that effectively leverages the parallelization capabilities of accelerator hardware. Our approach combines the Universal Speech Model (USM) and the PaLM 2 language model in per-segment scoring mode, achieving an average relative WER improvement across all languages of 10.8% on FLEURS and 3.6% on YouTube captioning. Furthermore, our comprehensive ablation study analyzes key parameters such as LLM size, context length, vocabulary size, fusion methodology. For instance, we explore the impact of LLM size ranging from 128M to 340B parameters on ASR performance. This study provides valuable insights into the factors influencing the effectiveness of practical large-scale LM-fused speech recognition systems.

Long-form video understanding represents a significant challenge within computer vision, demanding a model capable of reasoning over long multi-modal sequences. Motivated by the human cognitive process for long-form video understanding, we emphasize interactive reasoning and planning over the ability to process lengthy visual inputs. We introduce a novel agent-based system, VideoAgent, that employs a large language model as a central agent to iteratively identify and compile crucial information to answer a question, with vision-language foundation models serving as tools to translate and retrieve visual information. Evaluated on the challenging EgoSchema and NExT-QA benchmarks, VideoAgent achieves 54.1% and 71.3% zero-shot accuracy with only 8.4 and 8.2 frames used on average. These results demonstrate superior effectiveness and efficiency of our method over the current state-of-the-art methods, highlighting the potential of agent-based approaches in advancing long-form video understanding.

Foundation models have made significant strides in various applications, including text-to-image generation, panoptic segmentation, and natural language processing. This paper presents Instruct2Act, a framework that utilizes Large Language Models to map multi-modal instructions to sequential actions for robotic manipulation tasks. Specifically, Instruct2Act employs the LLM model to generate Python programs that constitute a comprehensive perception, planning, and action loop for robotic tasks. In the perception section, pre-defined APIs are used to access multiple foundation models where the Segment Anything Model (SAM) accurately locates candidate objects, and CLIP classifies them. In this way, the framework leverages the expertise of foundation models and robotic abilities to convert complex high-level instructions into precise policy codes. Our approach is adjustable and flexible in accommodating various instruction modalities and input types and catering to specific task demands. We validated the practicality and efficiency of our approach by assessing it on robotic tasks in different scenarios within tabletop manipulation domains. Furthermore, our zero-shot method outperformed many state-of-the-art learning-based policies in several tasks. The code for our proposed approach is available at https://github.com/OpenGVLab/Instruct2Act, serving as a robust benchmark for high-level robotic instruction tasks with assorted modality inputs.

The rise of large language models (LLMs) has transformed the role of information retrieval (IR) systems in the way to humans accessing information. Due to the isolated architecture and the limited interaction, existing IR systems are unable to fully accommodate the shift from directly providing information to humans to indirectly serving large language models. In this paper, we propose Self-Retrieval, an end-to-end, LLM-driven information retrieval architecture that can fully internalize the required abilities of IR systems into a single LLM and deeply leverage the capabilities of LLMs during IR process. Specifically, Self-retrieval internalizes the corpus to retrieve into a LLM via a natural language indexing architecture. Then the entire retrieval process is redefined as a procedure of document generation and self-assessment, which can be end-to-end executed using a single large language model. Experimental results demonstrate that Self-Retrieval not only significantly outperforms previous retrieval approaches by a large margin, but also can significantly boost the performance of LLM-driven downstream applications like retrieval augumented generation.

With the development in cognitive science and Large Language Models (LLMs), increasing connections have come to light between these two distinct fields. Building upon these connections, we propose a conjecture suggesting the existence of a duality between LLMs and Tulving's theory of memory. We identify a potential correspondence between Tulving's synergistic ecphory model (SEM) of retrieval and the emergent abilities observed in LLMs, serving as supporting evidence for our conjecture. Furthermore, we speculate that consciousness may be considered a form of emergent ability based on this duality. We also discuss how other theories of consciousness intersect with our research.

Significant advancements has recently been achieved in the field of multi-modal large language models (MLLMs), demonstrating their remarkable capabilities in understanding and reasoning across diverse tasks. However, these models are often trained for specific tasks and rely on task-specific input-output formats, limiting their applicability to a broader range of tasks. This raises a fundamental question: Can we develop a unified approach to represent and handle different multi-modal tasks to maximize the generalizability of MLLMs? In this paper, we propose UnifiedMLLM, a comprehensive model designed to represent various tasks using a unified representation. Our model exhibits strong capabilities in comprehending the implicit intent of user instructions and preforming reasoning. In addition to generating textual responses, our model also outputs task tokens and grounding tokens, serving as indicators of task types and task granularity. These outputs are subsequently routed through the task router and directed to specific expert models for task completion. To train our model, we construct a task-specific dataset and an 100k multi-task dataset encompassing complex scenarios. Employing a three-stage training strategy, we equip our model with robust reasoning and task processing capabilities while preserving its generalization capacity and knowledge reservoir. Extensive experiments showcase the impressive performance of our unified representation approach across various tasks, surpassing existing methodologies. Furthermore, our approach exhibits exceptional scalability and generality. Our code, model, and dataset will be available at https://github.com/lzw-lzw/UnifiedMLLM.

Large language Model (LLM) unlearning, i.e., selectively removing information from LLMs, is vital for responsible model deployment. Differently, LLM knowledge editing aims to modify LLM knowledge instead of removing it. Though editing and unlearning seem to be two distinct tasks, we find there is a tight connection between them. In this paper, we conceptualize unlearning as a special case of editing where information is modified to a refusal or "empty set" emptyset response, signifying its removal. This paper thus investigates if knowledge editing techniques are strong baselines for LLM unlearning. We evaluate state-of-the-art (SOTA) editing methods (e.g., ROME, MEMIT, GRACE, WISE, and AlphaEdit) against existing unlearning approaches on pretrained and finetuned knowledge. Results show certain editing methods, notably WISE and AlphaEdit, are effective unlearning baselines, especially for pretrained knowledge, and excel in generating human-aligned refusal answers. To better adapt editing methods for unlearning applications, we propose practical recipes including self-improvement and query merging. The former leverages the LLM's own in-context learning ability to craft a more human-aligned unlearning target, and the latter enables ROME and MEMIT to perform well in unlearning longer sample sequences. We advocate for the unlearning community to adopt SOTA editing methods as baselines and explore unlearning from an editing perspective for more holistic LLM memory control.

We introduce AIBrix, a cloud-native, open-source framework designed to optimize and simplify large-scale LLM deployment in cloud environments. Unlike traditional cloud-native stacks, AIBrix follows a co-design philosophy, ensuring every layer of the infrastructure is purpose-built for seamless integration with inference engines like vLLM. AIBrix introduces several key innovations to reduce inference costs and enhance performance including high-density LoRA management for dynamic adapter scheduling, LLM-specific autoscalers, and prefix-aware, load-aware routing. To further improve efficiency, AIBrix incorporates a distributed KV cache, boosting token reuse across nodes, leading to a 50% increase in throughput and a 70% reduction in inference latency. AIBrix also supports unified AI runtime which streamlines model management while maintaining vendor-agnostic engine compatibility. For large-scale multi-node inference, AIBrix employs hybrid orchestration -- leveraging Kubernetes for coarse-grained scheduling and Ray for fine-grained execution -- to balance efficiency and flexibility. Additionally, an SLO-driven GPU optimizer dynamically adjusts resource allocations, optimizing heterogeneous serving to maximize cost efficiency while maintaining service guarantees. Finally, AIBrix enhances system reliability with AI accelerator diagnostic tools, enabling automated failure detection and mock-up testing to improve fault resilience. AIBrix is available at https://github.com/vllm-project/aibrix.

Large language model (LLM) inference is often bounded by memory footprint and memory bandwidth in resource-constrained deployments, making quantization a fundamental technique for efficient serving. While post-training quantization (PTQ) maintains high fidelity at 4-bit, it deteriorates at 2-3 bits. Fundamentally, existing methods enforce a shape-invariant quantization grid (e.g., the fixed uniform intervals of UINT2) for each group, severely restricting the feasible set for error minimization. To address this, we propose Bit-Plane Decomposition Quantization (BPDQ), which constructs a variable quantization grid via bit-planes and scalar coefficients, and iteratively refines them using approximate second-order information while progressively compensating quantization errors to minimize output discrepancy. In the 2-bit regime, BPDQ enables serving Qwen2.5-72B on a single RTX 3090 with 83.85% GSM8K accuracy (vs. 90.83% at 16-bit). Moreover, we provide theoretical analysis showing that the variable grid expands the feasible set, and that the quantization process consistently aligns with the optimization objective in Hessian-induced geometry. Code: github.com/KingdalfGoodman/BPDQ.

Large language model (LLM) applications are evolving beyond simple chatbots into dynamic, general-purpose agentic programs, which scale LLM calls and output tokens to help AI agents reason, explore, and solve complex tasks. However, existing LLM serving systems ignore dependencies between programs and calls, missing significant opportunities for optimization. Our analysis reveals that programs submitted to LLM serving engines experience long cumulative wait times, primarily due to head-of-line blocking at both the individual LLM request and the program. To address this, we introduce Autellix, an LLM serving system that treats programs as first-class citizens to minimize their end-to-end latencies. Autellix intercepts LLM calls submitted by programs, enriching schedulers with program-level context. We propose two scheduling algorithms-for single-threaded and distributed programs-that preempt and prioritize LLM calls based on their programs' previously completed calls. Our evaluation demonstrates that across diverse LLMs and agentic workloads, Autellix improves throughput of programs by 4-15x at the same latency compared to state-of-the-art systems, such as vLLM.

Large language model (LLM) routing distributes inference work across different models, advancing the cost-quality Pareto frontier of LLM serving. While coarse-grained routing at the session or query level has been widely adopted in production systems, recent algorithmic work shows that fine-grained token-level routing can yield substantial efficiency and quality gains. However, efficiently serving token-level routed inference poses significant challenges to existing systems. Built on single-LLM assumptions, current systems suffer from severe step desynchronization and frequent batch admission delays under token-level routing, and they also impose high implementation complexity on developers. To address these challenges, we design TokenRouter, an efficient and developer-friendly serving system for token-level routed LLM inference. TokenRouter follows the principle of request-centric programming, model-centric execution: developers describe routing logic from the perspective of a single request, while the runtime launches a subserver for each LLM and dispatches requests asynchronously. Each subserver employs a delayed-batching scheduler, whose optimal hyperparameters are derived from a mathematical throughput model of the system. Across diverse routing algorithms, workloads, and model pairs, TokenRouter achieves 2.01-64.15x higher decoding throughput than existing systems, substantially advancing the serving efficiency of token-level LLM routing. Our code is available at https://github.com/thu-nics/TokenRouter.

Large language model (LLM) inference workload dominates a wide variety of modern AI applications, ranging from multi-turn conversation to document analysis. Balancing fairness and efficiency is critical for managing diverse client workloads with varying prefix patterns. Unfortunately, existing fair scheduling algorithms for LLM serving, such as Virtual Token Counter (VTC), fail to take prefix locality into consideration and thus suffer from poor performance. On the other hand, locality-aware scheduling algorithms in existing LLM serving frameworks tend to maximize the prefix cache hit rate without considering fair sharing among clients. This paper introduces the first locality-aware fair scheduling algorithm, Deficit Longest Prefix Match (DLPM), which can maintain a high degree of prefix locality with a fairness guarantee. We also introduce a novel algorithm, Double Deficit LPM (D^2LPM), extending DLPM for the distributed setup that can find a balance point among fairness, locality, and load-balancing. Our extensive evaluation demonstrates the superior performance of DLPM and D^2LPM in ensuring fairness while maintaining high throughput (up to 2.87times higher than VTC) and low per-client (up to 7.18times lower than state-of-the-art distributed LLM serving system) latency.

Large language model (LLM)-based multi-agent systems (MAS) coordinate specialized agents to solve complex tasks, turning each user request into interdependent model and tool operations. Unlike independent requests, these operations become ready at different times and consume evolving context as agents coordinate and exchange results. This creates changing patterns of concurrency, computation, and memory demand that can expose different bottlenecks in LLM serving hardware. Yet benchmarking these effects requires capturing the collaboration rules that generate serving workloads, beyond the requests observed during execution. We present MAServeBench, a benchmark that treats multi-agent collaboration as a configurable dimension of serving workloads for LLM serving and hardware evaluation. Its composable stages capture how agents coordinate and exchange context in documented MAS workflows, enabling native executions to generate different workloads even when task inputs and tools remain unchanged. Hierarchical Semantic Traces preserve realized operations and dependencies for reproducible comparisons of legal release constraints at fixed work. Task-linked measurements connect these workloads to serving execution and GPU/NPU hardware behavior, revealing why serving efficiency does not always translate into timely task completion. The benchmark code is available at https://github.com/ShengxuanQiu/MAServeBench.

Large Language Model (LLM) inference in production must meet stringent service-level objectives for both time-to-first-token (TTFT) and time-between-token (TBT) while maximizing throughput under fixed compute, memory, and interconnect budgets. Modern serving systems adopt stall-free scheduling techniques such as chunked prefill, which splits long prompt processing along the token dimension and interleaves prefill with ongoing decode iterations. While effective at stabilizing TBT, chunked prefill incurs substantial overhead in Mixture-of-Experts (MoE) models: redundant expert weight loads increase memory traffic by up to 39% and inflate energy consumption. We propose layered prefill, a new scheduling paradigm that treats transformer layer groups as the primary scheduling unit. By vertically partitioning the model into contiguous layer groups and interleaving prefill and decode across the groups, layered prefill sustains stall-free decoding while eliminating chunk-induced MoE weight reloads. It reduces off-chip bandwidth demand, lowering TTFT by up to 70%, End-to-End latency by 41% and per-token energy by up to 22%. Evaluations show that layered prefill consistently improves the TTFT--TBT Pareto frontier over chunked prefill, reducing expert-load traffic and energy cost while maintaining stall-free decoding. Overall, shifting the scheduling axis from tokens to layers unlocks a new operating regime for high-efficiency, energy-aware LLM serving in co-located environments.

Large language model agents are increasingly deployed to perform complex tasks in real-world environments. However, the knowledge required for correct behavior in these environments is often implicit, undisclosed, and subject to change over time. Recent continual-learning harnesses seek to address this challenge by enabling agents to improve from serving experience. Yet the effectiveness and limitations of these methods are not yet well characterized. Existing benchmarks provide only partial coverage: some explicitly provide the target knowledge, others assume a static environment, and those that support continual adaptation remain limited in scale and knowledge diversity. To enable systematic evaluation, we formalize an evolving-environment streaming dataset (EESD), in which agents must infer, apply, and revise latent environment knowledge from interaction and outcome feedback as hidden policies evolve, and introduce ServeLearnBench, spanning retail support, banking, and sales-pitch generation with 53 environment windows and 7,718 tasks. We evaluate five learning harnesses (RAG, Mem0, SkillOpt, Continual Harness, and Prime) across six models (GPT-5.6 Terra, Opus 5, Kimi K3, GLM-5.3, DeepSeek V4.1 Flash, and GLM-5.3 Flash), covering 28 model-harness pairs and 252 learning runs. Our evaluation reveals three main findings: a substantial gap remains between task capability and learning from experience; continual adaptation is costly and can degrade already-correct behavior; and insufficient exploration emerges as a key bottleneck to effective adaptation. Overall, ServeLearnBench provides a controlled testbed for diagnosing these limitations and tracking progress toward agents that continually and reliably improve through serving experience.

Optimizing Large Language Model (LLM) inference in production systems is increasingly difficult due to dynamic workloads, stringent latency/throughput targets, and a rapidly expanding configuration space. This complexity spans not only distributed parallelism strategies (tensor/pipeline/expert) but also intricate framework-specific runtime parameters such as those concerning the enablement of CUDA graphs, available KV-cache memory fractions, and maximum token capacity, which drastically impact performance. The diversity of modern inference frameworks (e.g., TRT-LLM, vLLM, SGLang), each employing distinct kernels and execution policies, makes manual tuning both framework-specific and computationally prohibitive. We present AIConfigurator, a unified performance-modeling system that enables rapid, framework-agnostic inference configuration search without requiring GPU-based profiling. AIConfigurator combines (1) a methodology that decomposes inference into analytically modelable primitives - GEMM, attention, communication, and memory operations while capturing framework-specific scheduling dynamics; (2) a calibrated kernel-level performance database for these primitives across a wide range of hardware platforms and popular open-weights models (GPT-OSS, Qwen, DeepSeek, LLama, Mistral); and (3) an abstraction layer that automatically resolves optimal launch parameters for the target backend, seamlessly integrating into production-grade orchestration systems. Evaluation on production LLM serving workloads demonstrates that AIConfigurator identifies superior serving configurations that improve performance by up to 40% for dense models (e.g., Qwen3-32B) and 50% for MoE architectures (e.g., DeepSeek-V3), while completing searches within 30 seconds on average. Enabling the rapid exploration of vast design spaces - from cluster topology down to engine specific flags.

The high computational and memory requirements of generative large language models (LLMs) make it challenging to serve them quickly and cheaply. This paper introduces SpecInfer, an LLM serving system that accelerates generative LLM inference with speculative inference and token tree verification. A key insight behind SpecInfer is to combine various collectively boost-tuned small language models to jointly predict the LLM's outputs; the predictions are organized as a token tree, whose nodes each represent a candidate token sequence. The correctness of all candidate token sequences represented by a token tree is verified by the LLM in parallel using a novel tree-based parallel decoding mechanism. SpecInfer uses an LLM as a token tree verifier instead of an incremental decoder, which significantly reduces the end-to-end latency and computational requirement for serving generative LLMs while provably preserving model quality.

Large language models (LLMs) have surged in popularity and are extensively used in commercial applications, where the efficiency of model serving is crucial for the user experience. Most current research focuses on optimizing individual sub-procedures, e.g. local inference and communication, however, there is no comprehensive framework that provides a holistic system view for optimizing LLM serving in an end-to-end manner. In this work, we conduct a detailed analysis to identify major bottlenecks that impact end-to-end latency in LLM serving systems. Our analysis reveals that a comprehensive LLM serving endpoint must address a series of efficiency bottlenecks that extend beyond LLM inference. We then propose ScaleLLM, an optimized system for resource-efficient LLM serving. Our extensive experiments reveal that with 64 concurrent requests, ScaleLLM achieves a 4.3x speed up over vLLM and outperforms state-of-the-arts with 1.5x higher throughput.

The rapid evolution of large language models (LLMs) has unlocked their capabilities in advanced reasoning tasks like mathematical problem-solving, code generation, and legal analysis. Central to this progress are inference-time reasoning algorithms, which refine outputs by exploring multiple solution paths, at the cost of increasing compute demands and response latencies. Existing serving systems fail to adapt to the scaling behaviors of these algorithms or the varying difficulty of queries, leading to inefficient resource use and unmet latency targets. We present Dynasor, a system that optimizes inference-time compute for LLM reasoning queries. Unlike traditional engines, Dynasor tracks and schedules requests within reasoning queries and uses Certaindex, a proxy that measures statistical reasoning progress based on model certainty, to guide compute allocation dynamically. Dynasor co-adapts scheduling with reasoning progress: it allocates more compute to hard queries, reduces compute for simpler ones, and terminates unpromising queries early, balancing accuracy, latency, and cost. On diverse datasets and algorithms, Dynasor reduces compute by up to 50% in batch processing and sustaining 3.3x higher query rates or 4.7x tighter latency SLOs in online serving.

Large model inference is shifting from cloud to edge due to concerns about the privacy of user interaction data. However, edge devices often struggle with limited computing power, memory, and bandwidth, requiring collaboration across multiple devices to run and speed up LLM inference. Pipeline parallelism, the mainstream solution, is inefficient for single-user scenarios, while tensor parallelism struggles with frequent communications. In this paper, we argue that tensor parallelism can be more effective than pipeline on low-resource devices, and present a compute- and memory-efficient tensor parallel inference system, named TPI-LLM, to serve 70B-scale models. TPI-LLM keeps sensitive raw data local in the users' devices and introduces a sliding window memory scheduler to dynamically manage layer weights during inference, with disk I/O latency overlapped with the computation and communication. This allows larger models to run smoothly on memory-limited devices. We analyze the communication bottleneck and find that link latency, not bandwidth, emerges as the main issue, so a star-based allreduce algorithm is implemented. Through extensive experiments on both emulated and real testbeds, TPI-LLM demonstrated over 80% less time-to-first-token and token latency compared to Accelerate, and over 90% compared to Transformers and Galaxy, while cutting the peak memory footprint of Llama 2-70B by 90%, requiring only 3.1 GB of memory for 70B-scale models.

Retrieval-augmented generation (RAG) improves large language model (LLM) answer quality by grounding generation in external evidence, but processing retrieved contexts makes the prefill stage a dominant serving cost. RAG cache fusion reduces this cost by reusing precomputed key-value (KV) caches for retrieved chunks and selectively recomputing tokens under the current prompt. Existing selectors, however, face a dilemma between quality and efficiency: fast query-agnostic or final-layer query-to-context selectors can miss request-relevant evidence, whereas full-view query-aware selectors require broad context and layer visibility before recomputation and therefore stall the layer-wise cache-fusion pipeline. We present QCFuse, a compressed-view query-aware selector for RAG cache fusion. QCFuse uses chunk-anchor query probing to condition user-query states on compact per-chunk anchors and critical-layer profiling to identify recomputation tokens without all-layer inspection. We implement QCFuse in SGLang and evaluate it on four open-weight LLMs across six datasets. QCFuse reaches full-prefill-level quality. At matched quality, QCFuse achieves an average prefill-time speedup of 1.7x over full prefill and 1.5x over ProphetKV, the strongest quality-preserving baseline.

Request scheduling is a key challenge in large-scale clusters serving agentic large language model (LLM) workloads. An effective scheduler must preserve key-value cache (KVC) reuse across long, shared prefixes, meet token-level latency service-level objectives (SLOs), and minimize GPU resource footprint. Existing schedulers struggle to reconcile these requirements: request consolidation can sacrifice cache locality and increase prefill/decode interference, compromising both SLO attainment and resource efficiency. We present PackServe, a scheduler designed to reduce resource costs while meeting latency SLOs for agentic LLM serving. PackServe uses compact white-box models to predict latency under prefill/decode interference. Guided by these predictions, it packs requests onto fewer serving instances while preserving KVC reuse and SLO constraints, trading available latency headroom for improved per-instance throughput. Evaluation on 64 NVIDIA H20 GPUs shows that PackServe uses up to 16.8% and 24.6% fewer GPU-hours than state-of-the-art schedulers under 30-ms and 50-ms TPOT targets, respectively, while meeting the target TPOT objectives. PackServe has also been deployed in our production cluster comprising over 1000 GPUs, where it reduces the resource footprint by 34.7% compared with the original production scheduler.

Large Language Models (LLMs) for Generative AI have achieved remarkable progress, evolving into sophisticated and versatile tools widely adopted across various domains and applications. However, the substantial memory overhead caused by their vast number of parameters, combined with the high computational demands of the attention mechanism, poses significant challenges in achieving low latency and high throughput for LLM inference services. Recent advancements, driven by groundbreaking research, have significantly accelerated progress in this field. This paper provides a comprehensive survey of these methods, covering fundamental instance-level approaches, in-depth cluster-level strategies, emerging scenario directions, and other miscellaneous but important areas. At the instance level, we review model placement, request scheduling, decoding length prediction, storage management, and the disaggregation paradigm. At the cluster level, we explore GPU cluster deployment, multi-instance load balancing, and cloud service solutions. For emerging scenarios, we organize the discussion around specific tasks, modules, and auxiliary methods. To ensure a holistic overview, we also highlight several niche yet critical areas. Finally, we outline potential research directions to further advance the field of LLM inference serving.

Transformers, driven by attention mechanisms, form the foundation of large language models (LLMs). As these models scale up, efficient GPU attention kernels become essential for high-throughput and low-latency inference. Diverse LLM applications demand flexible and high-performance attention solutions. We present FlashInfer: a customizable and efficient attention engine for LLM serving. FlashInfer tackles KV-cache storage heterogeneity using block-sparse format and composable formats to optimize memory access and reduce redundancy. It also offers a customizable attention template, enabling adaptation to various settings through Just-In-Time (JIT) compilation. Additionally, FlashInfer's load-balanced scheduling algorithm adjusts to dynamism of user requests while maintaining compatibility with CUDAGraph which requires static configuration. FlashInfer have been integrated into leading LLM serving frameworks like SGLang, vLLM and MLC-Engine. Comprehensive kernel-level and end-to-end evaluations demonstrate FlashInfer's ability to significantly boost kernel performance across diverse inference scenarios: compared to state-of-the-art LLM serving solutions, FlashInfer achieve 29-69% inter-token-latency reduction compared to compiler backends for LLM serving benchmark, 28-30% latency reduction for long-context inference, and 13-17% speedup for LLM serving with parallel generation.

Retrieval-Augmented Generation (RAG) is a critical paradigm for building reliable, knowledge-intensive Large Language Model (LLM) applications. However, the multi-stage pipeline (retrieve, generate) and unique workload characteristics (e.g., knowledge dependency) of RAG systems pose significant challenges for serving performance optimization. Existing generic LLM inference traces fail to capture these RAG-specific dynamics, creating a significant performance gap between academic research and real-world deployment. To bridge this gap, this paper introduces RAGPulse, an open-source RAG workload trace dataset. This dataset was collected from an university-wide Q&A system serving that has served more than 40,000 students and faculties since April 2024. We detail RAGPulse's system architecture, its privacy-preserving hash-based data format, and provide an in-depth statistical analysis. Our analysis reveals that real-world RAG workloads exhibit significant temporal locality and a highly skewed hot document access pattern. RAGPulse provides a high-fidelity foundation for researchers to develop and validate novel optimization strategies for RAG systems, such as content-aware batching and retrieval caching, ultimately enhancing the efficiency and reliability of RAG services. The code is available at https://github.com/flashserve/RAGPulse.

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.

Speculative inference is a promising paradigm employing small speculative models (SSMs) as drafters to generate draft tokens, which are subsequently verified in parallel by the target large language model (LLM). This approach enhances the efficiency of inference serving by reducing LLM inference latency and costs while preserving generation quality. However, existing speculative methods face critical challenges, including inefficient resource utilization and limited draft acceptance, which constrain their scalability and overall effectiveness. To overcome these obstacles, we present CoSine, a novel speculative inference system that decouples sequential speculative decoding from parallel verification, enabling efficient collaboration among multiple nodes. Specifically, CoSine routes inference requests to specialized drafters based on their expertise and incorporates a confidence-based token fusion mechanism to synthesize outputs from cooperating drafters, ensuring high-quality draft generation. Additionally, CoSine dynamically orchestrates the execution of speculative decoding and verification in a pipelined manner, employing batch scheduling to selectively group requests and adaptive speculation control to minimize idle periods. By optimizing parallel workflows through heterogeneous node collaboration, CoSine balances draft generation and verification throughput in real-time, thereby maximizing resource utilization. Experimental results demonstrate that CoSine achieves superior performance compared to state-of-the-art speculative approaches. Notably, with equivalent resource costs, CoSine achieves up to a 23.2% decrease in latency and a 32.5% increase in throughput compared to baseline methods.

Low-Rank Adaptation (LoRA) has become a widely adopted parameter-efficient fine-tuning (PEFT) technique for adapting large language models (LLMs) to downstream tasks. While prior work has explored strategies for integrating LLM training and serving, there still remains a gap in unifying fine-tuning and inference for LoRA-based models. We present Loquetier, a virtualized multi-LoRA framework that seamlessly integrates LoRA fine-tuning and serving within a single runtime. Loquetier introduces two key components: (1) a Virtualized Module that isolates PEFT-based modifications and supports multiple adapters on a shared base model, and (2) an optimized computation flow with a kernel design that merges fine-tuning and inference paths in forward propagation, enabling efficient batching and minimizing kernel invocation overhead. Extensive experiments across three task settings show that Loquetier consistently outperforms existing baselines in both performance and flexibility, achieving up to 3.0times the throughput of the state-of-the-art co-serving system on inference-only tasks and 46.4times higher SLO attainment than PEFT on unified fine-tuning and inference tasks. The implementation of Loquetier is publicly available at https://github.com/NJUDeepEngine/Loquetier.

Large language model (LLM) routing reduces serving cost by assigning each query to an appropriate model while preserving response quality. Learning such a router, however, often requires executing multiple candidate models on historical queries to collect query--model quality feedback, creating a nontrivial supervision cost before deployment. Existing work largely focuses on serving-time efficiency, overlooking whether the resulting savings are sufficient to recover this upfront expenditure. We further observe that routing quality often saturates well before all query--model feedback is collected, suggesting that dense supervision can be economically over-provisioned. We propose SaveRouter, a sparse-supervision routing framework that selectively acquires informative model feedback and shares capability information across related queries, while retaining query-level refinement for fine-grained routing. We evaluate routing by jointly accounting for supervision expenditure and subsequent serving-time savings. Across four routing benchmarks, the main setting uses only about 33--41% of available training feedback while maintaining competitive or better routing quality, and reduces the break-even deployment volume by approximately 1.9--9.5 times compared with the fastest conventional router. Further analysis shows that acquiring more supervision is not always economically preferable: the supervision level that minimizes serving cost can differ from the one that achieves the earliest payback. Our code is publicly available at https://github.com/LAMDA-Model-Reuse/SaveRouter.

Large Language Model (LLM)-based applications are graduating from research prototypes to products serving millions of users, influencing how people write and consume information. A prominent example is the appearance of Answer Engines: LLM-based generative search engines supplanting traditional search engines. Answer engines not only retrieve relevant sources to a user query but synthesize answer summaries that cite the sources. To understand these systems' limitations, we first conducted a study with 21 participants, evaluating interactions with answer vs. traditional search engines and identifying 16 answer engine limitations. From these insights, we propose 16 answer engine design recommendations, linked to 8 metrics. An automated evaluation implementing our metrics on three popular engines (You.com, Perplexity.ai, BingChat) quantifies common limitations (e.g., frequent hallucination, inaccurate citation) and unique features (e.g., variation in answer confidence), with results mirroring user study insights. We release our Answer Engine Evaluation benchmark (AEE) to facilitate transparent evaluation of LLM-based applications.

Large Language Model (LLM) libraries have emerged as the foundational infrastructure powering today's AI revolution, serving as the backbone for LLM deployment, inference optimization, fine-tuning, and production serving across diverse applications. Despite their critical role in the LLM ecosystem, these libraries face frequent quality issues and bugs that threaten the reliability of AI systems built upon them. To address this knowledge gap, we present the first comprehensive empirical investigation into bug characteristics and testing practices in modern LLM libraries. We examine 313 bug-fixing commits extracted across two widely-adopted LLM libraries: HuggingFace Transformers and vLLM.Through rigorous manual analysis, we establish comprehensive taxonomies categorizing bug symptoms into 5 types and root causes into 14 distinct categories.Our primary discovery shows that API misuse has emerged as the predominant root cause (32.17%-48.19%), representing a notable transition from algorithm-focused defects in conventional deep learning frameworks toward interface-oriented problems. Additionally, we examine 7,748 test functions to identify 7 distinct test oracle categories employed in current testing approaches, with predefined expected outputs (such as specific tensors and text strings) being the most common strategy. Our assessment of existing testing effectiveness demonstrates that the majority of bugs escape detection due to inadequate test cases (41.73%), lack of test drivers (32.37%), and weak test oracles (25.90%). Drawing from these findings, we offer some recommendations for enhancing LLM library quality assurance.

Efficiently serving large language model (LLM) inference tasks is crucial both for user-perceived latency such as time-to-first-token (TTFT) and for GPU utilization. However, LLM request routing, that is, assigning each inference request to a GPU instance, is particularly challenging: execution is highly input-dependent; batching and KV-cache reuse create strong cross-request coupling; and latency responds nonlinearly to context length, model/engine settings, and heterogeneous accelerators. As a result, simple traditional load balancing algorithms, and even heuristics tailored for LLM inference, fail to achieve good performance. We present Lodestar, a novel learning-based request routing system for distributed GPU clusters. Lodestar continuously collects a snapshot of the cluster at per-request level, including real-time instance state, request characteristics, and observed performance, and trains an online reward predictor that it uses to route inference requests to the instance that will maximize given reward (e.g., minimizing TTFT). Lodestar is cloud-native and works seamlessly with existing serving stacks (vLLM). With continuous online adaptation to changing workloads and infrastructure conditions, Lodestar achieves 1.41x lower average TTFT and 1.47x lower P99 TTFT on average (up to 2.15x/1.86x on homogeneous and 4.38x/4.42x on heterogeneous clusters) compared to a state-of-the-art prefix cache and load-aware heuristic, and learns these efficient routing strategies within about 5 minutes, based on experiments in a public cloud GPU cluster.

Deploying a large language model (LLM) inference service remains costly because centralized serving depends on specialized GPU clusters and high-bandwidth interconnects in datacenters. An appealing alternative is to leverage collaborative decentralized GPU pools. However, heterogeneity in GPU and limited interconnected network bandwidth, along with potentially dynamic availability, make efficient scheduling the central challenge in this scenario. In this paper, we present Parallax, a decentralized LLM serving system that turns a pool of heterogeneous GPUs into an efficient inference platform via a two-phase scheduler. Parallax decomposes planning into (i) model allocation, which places layers of each replica across diverse GPUs to jointly optimize latency and throughput under memory and link-bandwidth constraints, and (ii) request-time GPU pipeline selection, which stitches layers from different replicas into end-to-end execution chains that balance load and adapt to current conditions. We implement Parallax and evaluate it on open-source LLMs deployed over real volunteer nodes. Parallax consistently reduces latency and increases throughput relative to decentralized baselines, demonstrating that principled scheduling can make volunteer compute a practical, affordable substrate for LLM inference. Github Repo at: https://github.com/GradientHQ/parallax.

Multimodal Large Language Model (MLLM)-based Graphical User Interface (GUI) agents develop rapidly, with visual grounding that maps natural language instructions to target UI elements serving as the core capability. Existing GUI agents typically fine-tune MLLM on massive datasets to handle challenges in understanding instructions and UI interfaces, which not only incurs high data annotation costs but also makes performance dependent on data quality and distribution. To avoid such cumbersome yet ineffective training, we notice that complex UI interfaces can be decomposed into basic visual elements directly understandable by common MLLMs. Consequently, we propose ZoomUI that leverages inference scaling to guide common MLLMs in progressively anchor instruction elements to increasingly detailed interface elements. Specifically, ZoomUI first optimizes the latent thinking to transform original instruction into element visual features description, and subsequently leverages internal attention to iteratively zoom in target element interface region. Evaluations on extensive benchmarks demonstrate that ZoomUI reaches or even surpasses SOTA baselines.

Serverless computing has grown rapidly for serving Large Language Model (LLM) inference due to its pay-as-you-go pricing, fine-grained GPU usage, and rapid scaling. However, our analysis reveals that current serverless can effectively serve general LLM but fail with Low-Rank Adaptation (LoRA) inference due to three key limitations: 1) massive parameter redundancy among functions where 99% of weights are unnecessarily duplicated, 2) costly artifact loading latency beyond LLM loading, and 3) magnified resource contention when serving multiple LoRA LLMs. These inefficiencies lead to massive GPU wastage, increased Time-To-First-Token (TTFT), and high monetary costs. We propose ServerlessLoRA, a novel serverless inference system designed for faster and cheaper LoRA LLM serving. ServerlessLoRA enables secure backbone LLM sharing across isolated LoRA functions to reduce redundancy. We design a pre-loading method that pre-loads comprehensive LoRA artifacts to minimize cold-start latency. Furthermore, ServerlessLoRA employs contention aware batching and offloading to mitigate GPU resource conflicts during bursty workloads. Experiment on industrial workloads demonstrates that ServerlessLoRA reduces TTFT by up to 86% and cuts monetary costs by up to 89% compared to state-of-the-art LLM inference solutions.

Serving an answer from a large language model requires deciding when to abstain, yet a verifier's ranking accuracy alone does not determine the error rate among served answers. We introduce PriceCheck, which builds a compact family of decision rules from label-free checks such as re-solving a problem. Each check has a price: its agreement rates on correct and incorrect answers and its cost per run. Prices fitted on a small, class-enriched labelled set compose into predictions of a schedule's coverage and cost, guiding which checks to run and when to stop. A calibration test then selects a schedule at a stated selective-risk target. In mathematics, the selected schedules serve 76.1% of answers on average and keep held-out selective risk below 1.5% on all 15 splits. Under the shared testing protocol, PriceCheck serves more answers at that target than reward models, a prompted judge, the generator's confidence and a trained correctness classifier. At matched coverage, it keeps the fewest wrong answers among these scorers. Across 118 diagnostic schedules, price-based coverage predictions have a rank correlation of 0.97 with observed coverage. These results show that choosing how checks are combined and stopped matters alongside how well a verifier ranks answers. Code is available at https://github.com/js-lee-AI/PriceCheck.

GPU collective communication is typically optimized for bandwidth, yet many emerging workloads are increasingly limited by latency. Long-context decode-heavy large language model (LLM) inference is a prime example, where serving large models requires multiple GPUs, and many small collectives lie directly on the critical path of token generation. Therefore, even microsecond of overhead can impact performance and cost. In this work, we study how to approach the hardware Speed-of-Light (SoL) lower bound for GPU collectives within a scale-up network. We identify key principles for near-optimal designs, including barrier-free synchronization and efficient use of symmetric memory and multicast. Building on NCCL's device-side API, we develop low-latency interfaces for constructing custom collective kernels and use them to implement new symmetric collectives in NCCL. Microbenchmarks show substantial latency reductions for small and medium messages, reducing overhead to within 7% of the absolute SoL lower bound. When integrated into real applications, these kernels improve inter-token latency and throughput in LLM inference and accelerate cuSOLVERMp, demonstrating benefits for both AI inference and traditional HPC workloads.

GPU kernel optimization has long been a central challenge at the intersection of high-performance computing and machine learning. Efficient kernels are crucial for accelerating large language model (LLM) training and serving, yet attaining high performance typically requires extensive manual tuning. Compiler-based systems reduce some of this burden, but still demand substantial manual design and engineering effort. Recently, researchers have explored using LLMs for GPU kernel generation, though prior work has largely focused on translating high-level PyTorch modules into CUDA code. In this work, we introduce Astra, the first LLM-based multi-agent system for GPU kernel optimization. Unlike previous approaches, Astra starts from existing CUDA implementations extracted from SGLang, a widely deployed framework for serving LLMs, rather than treating PyTorch modules as the specification. Within Astra, specialized LLM agents collaborate through iterative code generation, testing, profiling, and planning to produce kernels that are both correct and high-performance. On kernels from SGLang, Astra achieves an average speedup of 1.32x using zero-shot prompting with OpenAI o4-mini. A detailed case study further demonstrates that LLMs can autonomously apply loop transformations, optimize memory access patterns, exploit CUDA intrinsics, and leverage fast math operations to yield substantial performance gains. Our work highlights multi-agent LLM systems as a promising new paradigm for GPU kernel optimization. Our code is publicly available at https://github.com/Anjiang-Wei/Astra.

In this paper, we propose a novel factored agent architecture designed to overcome the limitations of traditional single-agent systems in agentic AI. Our approach decomposes the agent into two specialized components: (1) a large language model (LLM) that serves as a high level planner and in-context learner, which may use dynamically available information in user prompts, (2) a smaller language model which acts as a memorizer of tool format and output. This decoupling addresses prevalent issues in monolithic designs, including malformed, missing, and hallucinated API fields, as well as suboptimal planning in dynamic environments. Empirical evaluations demonstrate that our factored architecture significantly improves planning accuracy and error resilience, while elucidating the inherent trade-off between in-context learning and static memorization. These findings suggest that a factored approach is a promising pathway for developing more robust and adaptable agentic AI systems.

We introduce xLLM, an intelligent and efficient Large Language Model (LLM) inference framework designed for high-performance, large-scale enterprise-grade serving, with deep optimizations for diverse AI accelerators. To address these challenges, xLLM builds a novel decoupled service-engine architecture. At the service layer, xLLM-Service features an intelligent scheduling module that efficiently processes multimodal requests and co-locates online and offline tasks through unified elastic scheduling to maximize cluster utilization. This module also relies on a workload-adaptive dynamic Prefill-Decode (PD) disaggregation policy and a novel Encode-Prefill-Decode (EPD) disaggregation policy designed for multimodal inputs. Furthermore, it incorporates a distributed architecture to provide global KV Cache management and robust fault-tolerant capabilities for high availability. At the engine layer, xLLM-Engine co-optimizes system and algorithm designs to fully saturate computing resources. This is achieved through comprehensive multi-layer execution pipeline optimizations, an adaptive graph mode and an xTensor memory management. xLLM-Engine also further integrates algorithmic enhancements such as optimized speculative decoding and dynamic EPLB, collectively serving to substantially boost throughput and inference efficiency. Extensive evaluations demonstrate that xLLM delivers significantly superior performance and resource efficiency. Under identical TPOT constraints, xLLM achieves throughput up to 1.7x that of MindIE and 2.2x that of vLLM-Ascend with Qwen-series models, while maintaining an average throughput of 1.7x that of MindIE with Deepseek-series models. xLLM framework is publicly available at https://github.com/jd-opensource/xllm and https://github.com/jd-opensource/xllm-service.

Due to the resource-intensive nature of training vision-language models on expansive video data, a majority of studies have centered on adapting pre-trained image-language models to the video domain. Dominant pipelines propose to tackle the visual discrepancies with additional temporal learners while overlooking the substantial discrepancy for web-scaled descriptive narratives and concise action category names, leading to less distinct semantic space and potential performance limitations. In this work, we prioritize the refinement of text knowledge to facilitate generalizable video recognition. To address the limitations of the less distinct semantic space of category names, we prompt a large language model (LLM) to augment action class names into Spatio-Temporal Descriptors thus bridging the textual discrepancy and serving as a knowledge base for general recognition. Moreover, to assign the best descriptors with different video instances, we propose Optimal Descriptor Solver, forming the video recognition problem as solving the optimal matching flow across frame-level representations and descriptors. Comprehensive evaluations in zero-shot, few-shot, and fully supervised video recognition highlight the effectiveness of our approach. Our best model achieves a state-of-the-art zero-shot accuracy of 75.1% on Kinetics-600.

In recent years, Contrastive Language-Image Pre-training (CLIP) has become a cornerstone in multimodal intelligence. However, recent studies have identified that the information loss in the CLIP encoding process is substantial, and CLIP tends to capture only coarse-grained features from the input. This deficiency significantly limits the ability of a single CLIP model to handle images rich in visual detail. In this work, we propose a simple yet effective model-agnostic strategy, Diversified Multiplet Upcycling (DMU), for CLIP. DMU efficiently fine-tunes a series of CLIP models that capture different feature spaces, from a dense pre-trained CLIP checkpoint, sharing parameters except for the Feed-Forward Network (FFN). These models can then be transformed into a CLIP-MoE with a larger model capacity, leading to significantly enhanced performance with minimal computational overhead. To the best of our knowledge, Diversified Multiplet Upcycling is the first approach to introduce sparsely activated MoE into CLIP foundation models. Extensive experiments demonstrate the significant performance of CLIP-MoE across various zero-shot retrieval, zero-shot image classification tasks, and downstream Multimodal Large Language Model (MLLM) benchmarks by serving as a vision encoder. Furthermore, Diversified Multiplet Upcycling enables the conversion of any dense CLIP model into CLIP-MoEs, which can seamlessly replace CLIP in a plug-and-play manner without requiring further adaptation in downstream frameworks. Through Diversified Multiplet Upcycling, we aim to provide valuable insights for future research on developing more efficient and effective multimodal learning systems.

Large language model (LLM) based agentic workflows have become a popular paradigm for coordinating multiple specialized agents to solve complex tasks. To improve serving efficiency, existing LLM systems employ prefix caching to reuse key-value (KV) tensors corresponding to agents' fixed prompts, thereby avoiding redundant computation across repeated invocations. However, current systems typically evict KV caches using a Least Recently Used (LRU) policy, which fails to anticipate future agent usage and often discards KV caches shortly before their reuse. This leads to frequent cache misses and substantial recomputation or swapping overhead. We present KVFlow, a workflow-aware KV cache management framework tailored for agentic workloads. KVFlow abstracts the agent execution schedule as an Agent Step Graph and assigns each agent a steps-to-execution value that estimates its temporal proximity to future activation. These values guide a fine-grained eviction policy at the KV node level, allowing KVFlow to preserve entries likely to be reused and efficiently manage shared prefixes in tree-structured caches. Moreover, KVFlow introduces a fully overlapped KV prefetching mechanism, which proactively loads required tensors from CPU to GPU in background threads for agents scheduled in the next step, thereby avoiding cache miss stalls during generation. Compared to SGLang with hierarchical radix cache, KVFlow achieves up to 1.83times speedup for single workflows with large prompts, and up to 2.19times speedup for scenarios with many concurrent workflows.

EduChat (https://www.educhat.top/) is a large-scale language model (LLM)-based chatbot system in the education domain. Its goal is to support personalized, fair, and compassionate intelligent education, serving teachers, students, and parents. Guided by theories from psychology and education, it further strengthens educational functions such as open question answering, essay assessment, Socratic teaching, and emotional support based on the existing basic LLMs. Particularly, we learn domain-specific knowledge by pre-training on the educational corpus and stimulate various skills with tool use by fine-tuning on designed system prompts and instructions. Currently, EduChat is available online as an open-source project, with its code, data, and model parameters available on platforms (e.g., GitHub https://github.com/icalk-nlp/EduChat, Hugging Face https://huggingface.co/ecnu-icalk ). We also prepare a demonstration of its capabilities online (https://vimeo.com/851004454). This initiative aims to promote research and applications of LLMs for intelligent education.

Conversational AI tools that can generate and discuss clinically correct radiology reports for a given medical image have the potential to transform radiology. Such a human-in-the-loop radiology assistant could facilitate a collaborative diagnostic process, thus saving time and improving the quality of reports. Towards this goal, we introduce RaDialog, the first thoroughly evaluated and publicly available large vision-language model for radiology report generation and interactive dialog. RaDialog effectively integrates visual image features and structured pathology findings with a large language model (LLM) while simultaneously adapting it to a specialized domain using parameter-efficient fine-tuning. To keep the conversational abilities of the underlying LLM, we propose a comprehensive, semi-automatically labeled, image-grounded instruct dataset for chest X-ray radiology tasks. By training with this dataset, our method achieves state-of-the-art clinical correctness in report generation and shows impressive abilities in interactive tasks such as correcting reports and answering questions, serving as a foundational step toward clinical dialog systems. Our code is available on github: https://github.com/ChantalMP/RaDialog.

To interact with daily-life articulated objects of diverse structures and functionalities, understanding the object parts plays a central role in both user instruction comprehension and task execution. However, the possible discordance between the semantic meaning and physics functionalities of the parts poses a challenge for designing a general system. To address this problem, we propose SAGE, a novel framework that bridges semantic and actionable parts of articulated objects to achieve generalizable manipulation under natural language instructions. More concretely, given an articulated object, we first observe all the semantic parts on it, conditioned on which an instruction interpreter proposes possible action programs that concretize the natural language instruction. Then, a part-grounding module maps the semantic parts into so-called Generalizable Actionable Parts (GAParts), which inherently carry information about part motion. End-effector trajectories are predicted on the GAParts, which, together with the action program, form an executable policy. Additionally, an interactive feedback module is incorporated to respond to failures, which closes the loop and increases the robustness of the overall framework. Key to the success of our framework is the joint proposal and knowledge fusion between a large vision-language model (VLM) and a small domain-specific model for both context comprehension and part perception, with the former providing general intuitions and the latter serving as expert facts. Both simulation and real-robot experiments show our effectiveness in handling a large variety of articulated objects with diverse language-instructed goals.

Neuroimaging is a ubiquitous tool for evaluating patients with neurological diseases. The global demand for magnetic resonance imaging (MRI) studies has risen steadily, placing significant strain on health systems, prolonging turnaround times, and intensifying physician burnout. These challenges disproportionately impact patients in low-resource and rural settings. Here, we utilized a large academic health system as a data engine to develop Prima, the first vision language model (VLM) serving as an AI foundation for neuroimaging that supports real-world, clinical MRI studies as input. Trained on over 220,000 MRI studies, Prima uses a hierarchical vision architecture that provides general and transferable MRI features. Prima was tested in a 1-year health system-wide study that included 30K MRI studies. Across 52 radiologic diagnoses from the major neurologic disorders, including neoplastic, inflammatory, infectious, and developmental lesions, Prima achieved a mean diagnostic area under the ROC curve of 92.0, outperforming other state-of-the-art general and medical AI models. Prima offers explainable differential diagnoses, worklist priority for radiologists, and clinical referral recommendations across diverse patient demographics and MRI systems. Prima demonstrates algorithmic fairness across sensitive groups and can help mitigate health system biases, such as prolonged turnaround times for low-resource populations. These findings highlight the transformative potential of health system-scale VLMs and Prima's role in advancing AI-driven healthcare.

The rapid evolution of large language models (LLMs), driven by growing parameter scales, adoption of mixture-of-experts (MoE) architectures, and expanding context lengths, imposes unprecedented demands on AI infrastructure. Traditional AI clusters face limitations in compute intensity, memory bandwidth, inter-chip communication, and latency, compounded by variable workloads and strict service-level objectives. Addressing these issues requires fundamentally redesigned hardware-software integration. This paper introduces Huawei CloudMatrix, a next-generation AI datacenter architecture, realized in the production-grade CloudMatrix384 supernode. It integrates 384 Ascend 910C NPUs and 192 Kunpeng CPUs interconnected via an ultra-high-bandwidth Unified Bus (UB) network, enabling direct all-to-all communication and dynamic pooling of resources. These features optimize performance for communication-intensive operations, such as large-scale MoE expert parallelism and distributed key-value cache access. To fully leverage CloudMatrix384, we propose CloudMatrix-Infer, an advanced LLM serving solution incorporating three core innovations: a peer-to-peer serving architecture that independently scales prefill, decode, and caching; a large-scale expert parallelism strategy supporting EP320 via efficient UB-based token dispatch; and hardware-aware optimizations including specialized operators, microbatch-based pipelining, and INT8 quantization. Evaluation with the DeepSeek-R1 model shows CloudMatrix-Infer achieves state-of-the-art efficiency: prefill throughput of 6,688 tokens/s per NPU and decode throughput of 1,943 tokens/s per NPU (<50 ms TPOT). It effectively balances throughput and latency, sustaining 538 tokens/s even under stringent 15 ms latency constraints, while INT8 quantization maintains model accuracy across benchmarks.

Large language models (LLMs) power a new generation of interactive AI applications exemplified by ChatGPT. The interactive nature of these applications demand low job completion time (JCT) for model inference. Existing LLM serving systems use run-to-completion processing for inference jobs, which suffers from head-of-line blocking and long JCT. We present FastServe, a distributed inference serving system for LLMs. FastServe exploits the autoregressive pattern of LLM inference to enable preemption at the granularity of each output token. FastServe uses preemptive scheduling to minimize JCT with a novel skip-join Multi-Level Feedback Queue scheduler. Based on the new semi information-agnostic setting of LLM inference, the scheduler leverages the input length information to assign an appropriate initial queue for each arrival job to join. The higher priority queues than the joined queue are skipped to reduce demotions. We design an efficient GPU memory management mechanism that proactively offloads and uploads intermediate states between GPU memory and host memory for LLM inference. We build a system prototype of FastServe based on NVIDIA FasterTransformer. Experimental results show that compared to the state-of-the-art solution Orca, FastServe improves the average and tail JCT by up to 5.1times and 6.4times, respectively.

Six-bit quantization (FP6) can effectively reduce the size of large language models (LLMs) and preserve the model quality consistently across varied applications. However, existing systems do not provide Tensor Core support for FP6 quantization and struggle to achieve practical performance improvements during LLM inference. It is challenging to support FP6 quantization on GPUs due to (1) unfriendly memory access of model weights with irregular bit-width and (2) high runtime overhead of weight de-quantization. To address these problems, we propose TC-FPx, the first full-stack GPU kernel design scheme with unified Tensor Core support of float-point weights for various quantization bit-width. We integrate TC-FPx kernel into an existing inference system, providing new end-to-end support (called FP6-LLM) for quantized LLM inference, where better trade-offs between inference cost and model quality are achieved. Experiments show that FP6-LLM enables the inference of LLaMA-70b using only a single GPU, achieving 1.69x-2.65x higher normalized inference throughput than the FP16 baseline. The source code will be publicly available soon.

The growing demand for Large Language Models (LLMs) across diverse applications has prompted a paradigm shift in the design of deep learning serving systems. Deploying LLMs, especially in multi-tenant environments, presents considerable challenges due to their high computational and memory demands. We present BlockLLM, a serving system that exploits the potential of sharing components among fine-tuned LLM models to offer an efficient and flexible solution for LLM workloads. BlockLLM partitions the models into finer-grained blocks to enable the reuse of model components and independent provisioning to improve the computation efficiency. BlockLLM consists of an offline block zoo, for storing the blocks, and an online system to serve the requests through chains of blocks. It offers multi-fold flexibility: (1) Adaptive assembly of block chains on-the-fly is achieved with the help of equivalence evaluation among blocks in the zoo. (2) We enable per-block batch size and configure best-effort KV cache coordination at individual block level. (3) We adopt speculative execution and locality-aware block placement to mitigate the communication costs from dynamic block resource allocation. Our evaluation demonstrates that BlockLLM reduces memory and storage footprints and improves computation efficiency, outperforming existing serving approach in 95\%ile latency and GPU utilization by 33.5\% and 20.1\%, respectively.

Reducing the inference latency of large language models (LLMs) is crucial, and speculative decoding (SD) stands out as one of the most effective techniques. Rather than letting the LLM generate all tokens directly, speculative decoding employs effective proxies to predict potential outputs, which are then verified by the LLM without compromising the generation quality. Yet, deploying SD in real online LLM serving systems (with continuous batching) does not always yield improvement -- under higher request rates or low speculation accuracy, it paradoxically increases latency. Furthermore, there is no best speculation length work for all workloads under different system loads. Based on the observations, we develop a dynamic framework SmartSpec. SmartSpec dynamically determines the best speculation length for each request (from 0, i.e., no speculation, to many tokens) -- hence the associated speculative execution costs -- based on a new metric called goodput, which characterizes the current observed load of the entire system and the speculation accuracy. We show that SmartSpec consistently reduces average request latency by up to 3.2x compared to non-speculative decoding baselines across different sizes of target models, draft models, request rates, and datasets. Moreover, SmartSpec can be applied to different styles of speculative decoding, including traditional, model-based approaches as well as model-free methods like prompt lookup and tree-style decoding.

Mixture of experts (MoE) is a popular technique to improve capacity of Large Language Models (LLMs) with conditionally-activated parallel experts. However, serving MoE models on memory-constrained devices is challenging due to the large parameter size. Typical solutions such as memory swapping or expert pruning may lead to significantly higher latency or severe accuracy loss. In this paper, we introduce SwapMoE, a framework for efficient serving of MoE-based large language models with tunable memory budgets. The main idea of SwapMoE is to keep a small dynamic set of important experts, namely Virtual Experts, in the main memory for inference, while seamlessly maintaining how the Virtual Experts map to the actual experts. Experiments have shown that SwapMoE can reduce the memory footprint while maintaining reasonable accuracy. For example, on text summarization tasks with Switch Transformer, SwapMoE can reduce the memory consumption from 14.2 GiB to 4.7 GiB, together with 50\% latency reduction and a slight Rouge-2 score drop of 0.041.

The use of propagandistic techniques in online content has increased in recent years aiming to manipulate online audiences. Fine-grained propaganda detection and extraction of textual spans where propaganda techniques are used, are essential for more informed content consumption. Automatic systems targeting the task over lower resourced languages are limited, usually obstructed by lack of large scale training datasets. Our study investigates whether Large Language Models (LLMs), such as GPT-4, can effectively extract propagandistic spans. We further study the potential of employing the model to collect more cost-effective annotations. Finally, we examine the effectiveness of labels provided by GPT-4 in training smaller language models for the task. The experiments are performed over a large-scale in-house manually annotated dataset. The results suggest that providing more annotation context to GPT-4 within prompts improves its performance compared to human annotators. Moreover, when serving as an expert annotator (consolidator), the model provides labels that have higher agreement with expert annotators, and lead to specialized models that achieve state-of-the-art over an unseen Arabic testing set. Finally, our work is the first to show the potential of utilizing LLMs to develop annotated datasets for propagandistic spans detection task prompting it with annotations from human annotators with limited expertise. All scripts and annotations will be shared with the community.