LLM Inference Acceleration

LLM: Large Language Model

Ask about this topic

Overview

This field studies how to make large language models answer faster and use less memory and computing power, especially during prompt processing and token-by-token generation. Common approaches include speculative decoding, which drafts and verifies tokens; sparse attention and KV-cache compression for long contexts; low-bit quantization; and adaptive computation that skips or reallocates work. Researchers test these methods on long-context and retrieval benchmarks such as LongBench and RULER, as well as reasoning and coding tasks, measuring both quality and serving metrics such as time to first token and throughput. A central challenge is turning theoretical reductions in computation or memory into real speedups on particular hardware without sacrificing accuracy or changing the model’s output distribution.

Latest papers

CardsList

1,101 papers202 papers in the last four weeks, up 51% on the four weeks before. 1.2% of all new papers.

May 11, 2026

GELATO: Generative Entropy- and Lyapunov-based Adaptive Token Offloading for Device-Edge Speculative LLM Inference

The recent growth of on-device Large Language Model (LLM) inference has driven significant interest in device-edge collaborative LLM inference. As a promising architecture, Speculative Decoding (SD) is increasingly adopted where a lightweight draft model rapidly generates candidate tokens to be verified by a powerful target model. However, a fundamental challenge lies in achieving per-token resource scheduling to effectively adapt SD paradigm to resource-constrained edge environment. This paper proposes a Generative Entropy- and Lyapunov-based Adaptive Token Offloading framework, named GELATO, to maximize decoding throughput under energy constraints in a device-edge collaborative SD system. Specifically, an outer drift-plus-penalty loop makes online decisions to establish a reference drafting budget, managing long-term energy-throughput trade-off. Further, a nested entropy-driven generation mechanism executes early exiting to adapt to per-token dynamic generative uncertainty. Theoretical analysis establishes a rigorous performance bound on long-term throughput for GELATO. Extensive evaluations demonstrate that GELATO achieves a globally optimal tradeoff, outperforming state-of-the-art distributed SD architectures by 64.98% in token throughput and reducing energy consumption by 47.47% under resource-constrained environments, while preserving LLM decoding quality.

Zengzipeng Tang, Yuxuan Sun, Wei Chen +2

School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing 100044, China

On-Device Language Model InferenceEdge ComputingSpeculative DecodingLLM Inference Acceleration
AskPDF
May 11, 2026

Attention Drift: What Autoregressive Speculative Decoding Models Learn

Speculative decoding accelerates LLM inference by drafting future tokens with a small model, but drafter models degrade sharply under template perturbation and long-context inputs. We identify a previously-unreported phenomenon we call attention drift: as the drafter generates successive tokens within a speculation chain, attention progressively moves from the prompt onto its own recently-generated tokens. We observe this across both EAGLE3 drafters and MTP heads, suggesting drift is a property of drafter designs. We trace this to the un-normalized residual path between chain steps: the drafter's hidden state magnitude grows monotonically with chain depth, which exhibits dynamics consistent with additional pre-norm transformer layers stacked on the target rather than as a standalone autoregressive predictor. In order to limit the growth, we propose two architectural changes: Post-norm on the drafter hidden states and per-hidden-state RMSNorm after capturing target hidden states. Our interventions improve acceptance length over the current leading model, pre-norm EAGLE3, by up to 2×2\times under template perturbation, 1.18×1.18\times on long-context tasks, and 1.10×1.10\times on seven standard benchmarks spanning multi-turn chat, math, and coding. Our changes also allow shorter train-time-test depths to generalize over longer drafting sequences.

Doğaç Eldenk, Payal Mohapatra, Yigitcan Comlek +3

Northwestern University · GE Aerospace · fal +1

Long-Context Language Model InferenceSpeculative DecodingTransformer AttentionLLM Inference AccelerationLayer NormalizationAutoregressive Transformer
AskPDF
May 11, 2026

Merlin: Deterministic Byte-Exact Deduplication for Lossless Context Optimization in Large Language Model Inference

Data-intensive applications, ranging from large-scale retrieval systems to advanced data pipelines, are increasingly bottlenecked by the processing of highly redundant text corpora. We present Merlin, a local-first, agnostic, high-throughput deduplication and context optimization engine designed to mitigate these inefficiencies. Utilizing a highly optimized, SIMD-friendly open-addressing flat hash set combined with xxHash3-64, Merlin performs rapid, byte-exact deduplication of text passages and data chunks. While broadly applicable to any text-processing workflow, its impact is particularly pronounced in Large Language Model (LLM) ecosystems, such as Retrieval-Augmented Generation (RAG). Our empirical evaluations demonstrate an input reduction ranging from 13.9% in low-redundancy datasets to over 71% in high-redundancy pipelines, maintaining absolute data fidelity. Furthermore, we detail the system's integration architecture via the Model Context Protocol (MCP), enabling secure, zero-network-interception deployment across major IDEs and autonomous agents. This paper outlines the core algorithmic design, performance benchmarks, and the architectural principles required to process data at sustained speeds of up to 8.7 GB/s.

Sietse Schelpe

Corbenic AI, Inc.

LLM Inference Acceleration
AskPDF
May 10, 2026

Entropy-informed Decoding: Adaptive Information-Driven Branching

Large language models (LLMs) achieve remarkable generative performance, yet their output quality is dependent on the decoding strategy. While sampling-based methods (e.g., top-k, nucleus) and search-and-select based methods (e.g., beam search, best-of-n, majority voting) can improve upon greedy decoding, both approaches suffer from limitations: sampling generally commits to a single path, while search often expends excessive computation regardless of task complexity. To address these, we introduce Entropy-informed decoding (EDEN), a plug-and-play, model-agnostic decoding framework that adaptively allocates computation based on the model's own uncertainty, approximating higher-width beam search with fewer expansions. At each generation step, EDEN estimates the entropy of the output token distribution and adjusts the branching factor monotonically with the entropy, expanding more candidates in high-entropy regions and following a greedier path in low-entropy regions, improving token efficiency. Experiments across complex tasks, including mathematical reasoning, code generation, and scientific questions, demonstrate that EDEN consistently improves output quality over existing decoding strategies, achieving better accuracy-expansion trade-offs than fixed-width beam search. By treating next-token selection as a noisy maximisation problem, we prove that branching factors monotone in entropy are guaranteed to find better (i.e. more probable) continuations than any fixed branching factor within the same total expansion budget, and derive explicit regret rates characterising the benefit of the adaptive allocation.

Benjamin Patrick Evans, Sumitra Ganesh, Leo Ardon

JP Morgan AI Research, London, UK · JP Morgan AI Research, New York, USA

Language Model DecodingLLM Inference Acceleration
AskPDF
May 10, 2026

KV-RM: Regularizing KV-Cache Movement for Static-Graph LLM Serving

Static-graph LLM decoders provide predictable launches, fixed tensor shapes, and low submission overhead, but online decoding exposes highly irregular KV-cache behavior: request lengths differ, EOS events arrive asynchronously, and logical histories fragment over time. Dynamic runtimes recover flexibility through paged KV management and step-level scheduling, while static-graph executors often over-reserve memory and suffer burst-time latency outliers. This paper studies whether much of this variability can be absorbed below a fixed decode interface. We present KV-RM, a runtime design that regularizes KV-cache movement beneath a static-graph LLM decoder. KV-RM decouples logical KV histories from physical storage, tracks active KV state through a block pager, and materializes each decode step through a single committed descriptor. A merge-staged transport path coalesces non-contiguous KV mappings into a small number of large transfer groups before a fixed-shape attention kernel consumes them. Optional bounded far-history summaries can be enabled under the same interface, but the core design does not depend on them. On a 2-GPU NVIDIA A100 node, KV-RM improves mixed-length decoding throughput and tail latency relative to a static-graph baseline, reduces reserved KV memory across workload families, and removes severe burst-time latency spikes under production-trace replay. These results suggest that KV-cache movement, rather than kernel shape, can be an effective boundary for recovering runtime flexibility in static-graph LLM serving.

Zhiqing Zhong, Zhijing Ye, Jian Zhang +3

Stevens Institute of Technology Hoboken, New Jersey, USA · The Hong Kong University of Science and Technology (Guangzhou) Guangzhou, Guangdong, China · Argonne National Laboratory Lemont, Illinois, USA +1

KV CachingKV-Cache ManagementLLM Inference ServingLLM Inference Acceleration
AskPDF
May 10, 2026

Test-Time Speculation

Speculative decoding accelerates LLM inference by using a fast draft model to generate tokens and a more accurate target model to verify them. Its performance depends on the acceptance length\textit{acceptance length}, or number of draft tokens accepted by the target. Our studies show that the acceptance length of even state-of-the-art speculators, like DFlash, EAGLE-3 and PARD degrade with generation length, reaching values close to 1 (i.e. no speedup) within just a few thousand output tokens, making speculators ineffective for long-response tasks. Acceptance lengths decline because most speculators are trained offline on short sequences, but are forced to match the target model on much longer outputs at inference, well beyond their training distribution. To address this issue, we propose Test-Time Speculation (TTS)\textit{Test-Time Speculation (TTS)}, an online distillation approach that continuously adapts the speculator at test-time. TTS leverages the key insight that the token verification step already invokes the target model for each draft token, providing the training signal needed to adapt the draft at no additional cost. Treating the draft as the student and the target as a teacher, TTS adjusts the draft over several speculation rounds, with each update improving the draft's accuracy as generation proceeds. Our results across multiple models from the Qwen-3, Qwen-3.5, and Llama3.1 families show that TTS improves acceptance lengths over state-of-the-art speculators by up to 72%72\% and 41%41\% on average, with the benefits scaling with increased generation lengths.

Avinash Kumar, Sujay Sanghavi, Poulami Das

The University of Texas at Austin

Test-Time AdaptationTest-Time OptimizationSpeculative DecodingLLM Inference AccelerationLanguage Model Distillation
AskPDF
May 9, 2026

Sparse Layers are Critical to Scaling Looped Language Models

Looped language models repeat a set of transformer layers through depth, reducing memory costs and providing natural early-exit points at loop boundaries. However, looped models do not scale as favorably as standard transformers with unique layers. We compare standard and Mixture-of-Experts (MoE) transformers, with and without looping, and find two main results. First, we find Looped-MoE models scale better than the standard baseline while dense looped models do not. We trace this to routing divergence between loops: in Looped-MoE models, different experts are activated on each pass through the same shared layers, recovering expressivity without additional parameters. Our second finding is that looped models have better compute-quality trade-offs with early exits than standard models. Because each loop ends with the same layers that produce the final output, loop boundaries are superior exit points, as confirmed by earlier output convergence at these points. In sum, we provide a clear direction for scaling looped models: a Looped-MoE model with early exits can not only beat standard transformers at scale, but also enable significant memory and inference savings with minimal degradation in quality.

Ryan Lee, Jacob Biloki, Edward J. Hu +1

1USC Information Sciences Institute · 2Netflix · 3Independent Researcher

Mixture-of-Experts Language ModelsRecurrent TransformersLanguage Model ScalingLLM Inference AccelerationLanguage ModelingRecursive Language Models
AskPDF
May 9, 2026

ProxyKV: Cross-Model Proxy Pruning for Efficient Long-Context LLM Inference

Efficient long-context inference in Large Language Models (LLMs) is severely constrained by the Key-Value (KV) cache memory wall, yet existing pruning methods force a choice between low-latency heuristics that sacrifice precision and high-precision reconstruction methods that incur prohibitive prefilling overhead. To bridge this scoring-cost--accuracy gap, we propose ProxyKV, a cross-model proxy pruning framework that offloads importance scoring to a lightweight intra-family Small-Model Proxy executed asynchronously to the Large-Model Target. To bridge the architectural gap between heterogeneous models, we design the HybridAxialMapper, which disentangles temporal feature extraction from cross-head alignment, together with a Multi-Granularity Hybrid Loss that shifts the learning objective from rigid regression to relative ranking consistency. Across the Llama-3.1, Qwen-2.5, and Qwen-3 families spanning targets from 7B up to 32B parameters on LongBench, SCBench, and RULER, ProxyKV matches KVZip on aggregate (recovering \sim$$98.7\% of its mean accuracy) while delivering up to a 3.21×3.21\times prefilling speedup on Llama-3.1-8B (dual-GPU; \sim$$1.5\times shared single-GPU) and sustaining the speedup at contexts up to 170k tokens on Qwen-2.5-7B.

Junjie Li, Jiong Lou, Jie Li

Shanghai Jiao Tong University

LLM PruningKV CachingLong-Context Language Model InferenceKV-Cache EvictionLLM Inference AccelerationKV-Cache Pruning
AskPDF
May 9, 2026

BubbleSpec: Turning Long-Tail Bubbles into Speculative Rollout Drafts for Synchronous Reinforcement Learning

Reinforcement Learning (RL) has become a cornerstone for improving the performance of Large Language Models (LLMs). However, its rollout phase constitutes a significant efficiency bottleneck, mainly arising from the long-tail bubbles across data parallel ranks, particularly in long-context scenarios where faster GPUs remain idle while waiting for stragglers. Existing solutions, such as partial rollout or asynchronous RL, mitigate these bubbles by compromising the algorithm's strict synchronous nature. Instead, we propose BubbleSpec, a novel framework that accelerates RL rollouts while strictly keeping the mathematical exactness. Instead of attempting to eliminate bubbles, BubbleSpec exploits them. We exploit the idle time windows of faster ranks to pre-generate rollout results for subsequent steps, serving as drafts for speculative decoding. Unlike prior speculative methods that rely on historical epoch similarity and warm-ups, BubbleSpec is agnostic to dataset size and provides immediate acceleration from the onset of training. Extensive evaluations demonstrate that BubbleSpec reduces decoding steps by 50% and increases rollout throughput by up to 1.8x. Critically, BubbleSpec is seamlessly compatible with various RL frameworks and strategies as it sustains the strict synchronous property of RL algorithms.

Yuhang Xu, Kaibin Tian, Yang Tian +6

1Shanghai Jiao Tong University · 2Bytedance

Reinforcement LearningSpeculative DecodingLLM Inference AccelerationRL Post-Training
AskPDF
May 9, 2026

ReST-KV: Robust KV Cache Eviction with Layer-wise Output Reconstruction and Spatial-Temporal Smoothing

Large language models (LLMs) face growing challenges in efficient generative inference due to the increasing memory demands of Key-Value (KV) caches, especially for long sequences. Existing eviction methods typically retain KV pairs with high attention weights but overlook the impact of attention redistribution caused by token removal, as well as the spatial-temporal dynamics in KV selection. In this paper, we propose ReST-KV, a robust KV eviction method that combines layer-wise output Reconstruction and Spatial-Temporal smoothing to provide a more comprehensive perspective for the KV cache eviction task. Specifically, ReST-KV formulates KV cache eviction as an optimization problem that minimizes output discrepancies through efficient layer-wise reconstruction. By directly modeling how each token's removal affects the model output, our method naturally captures attention redistribution effects, going beyond simplistic reliance on raw attention weights. To further enhance robustness, we design exponential moving average smoothing to handle temporal variations and an adaptive window-based mechanism to capture spatial patterns. Our method, ReST-KV, significantly advances performance on long-context benchmarks. It surpasses state-of-the-art baselines by 2.58% on LongBench and 15.2% on RULER. Additionally, ReST-KV consistently outperforms existing methods on Needle-in-a-Haystack and InfiniteBench, all while achieving a remarkable 10.61×\times reduction in decoding latency at 128k context length. The code is publicly available at https://github.com/an-yongqi/rest-kv to facilitate reproducibility and further research.

Yongqi An, Chang Lu, Kuan Zhu +5

Foundation Model Research Center, Institute of Automation, Chinese Academy of Sciences, Beijing, China · School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China · University of Electronic Science and Technology of China, Chengdu, China +2

KV CachingLong-Context Language Model InferenceKV-Cache EvictionLLM Inference Acceleration
AskPDF
May 9, 2026

LAQuant: A Simple Overhead-free Large Reasoning Model Quantization by Layer-wise Lookahead Loss

Large reasoning models (LRMs) reach competition-level math and coding accuracy via long autoregressive decoding, making per-token decoding cost a primary deployment concern. Weight quantization is the standard tool for acceleration, but representative recipes -- including state-of-the-art end-to-end (E2E) QAT -- lose accuracy on long-decoding reasoning benchmarks despite preserving perplexity and short-decode accuracy. Through a systematic gradient-direction analysis, we identify two factors driving this gap: (i) KV-cache fidelity preservation under the QAT loss, which E2E supervision attenuates via the softmax Fisher metric; and (ii) Hessian-subspace alignment between calibration data and the deployment distribution. We propose LookAhead Quantization (LAQuant), a layer-wise weight-only QAT method that addresses both factors without online-transform overhead by combining reasoning-domain calibration with a one-layer lookahead loss whose implicit cross-layer co-adaptation preserves the next-layer residual stream. For Qwen3-4B under W3G128 quantization, LAQuant improves AIME25 Pass@1 over ParoQuant by 15.11pp (1.93pp over ParoQuant++ at matched calibration) while achieving a 3.42x decoding speedup over FP16 on RTX A6000, compared with ParoQuant's 3.01x.

Euntae Choi, Sumin Song, Sungjoo Yoo

Seoul National University

LLM QuantizationQuantization-Aware TrainingLLM Inference AccelerationWeight-Only QuantizationNeural Network Quantization
AskPDF
May 9, 2026

Structured Recurrent Mixers for Massively Parallelized Sequence Generation

Over the last two decades, language modeling has experienced a shift from the use of predominantly recurrent architectures that process tokens sequentially during training and inference to non-recurrent models that process sequence elements in parallel during training, which results in greater training efficiency and stability at the expense of lower inference throughput. Here we introduce the Structured Recurrent Mixer, an architecture that allows for algebraic conversion between a sequence parallel representation at train time and a recurrent representation at inference, notably without the need for specialized kernels or device-specific memory management. We show experimentally that this dual representation allows for greater training efficiency, higher input information capacity, and larger inference throughput and concurrency when compared to other linear complexity models. We postulate that recurrent models are poorly suited to extended sequence length scaling for information-rich inputs typical of language, but are well suited to scaling in the sample (batch) dimension due to their constant memory per sample. We provide Mojo/MAX inference implementations of SRMs exhibiting 12x the throughput and 170x the concurrency of similarly powerful Transformers inferenced on vLLM, increases characteristic of Pytorch implementations resulting in a 30% increase in compute-constant GSM8k Pass@k. We conclude by demonstrating that SRMs are effective reinforcement learning training candidates.

Benjamin L. Badger

IBM

Recurrent Neural NetworksAutoregressive Language ModelingEfficient Language Model TrainingLLM Inference AccelerationRecurrent Language Models
AskPDF
May 9, 2026

PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding

Speculative decoding accelerates Large Language Models (LLMs) inference by using a lightweight draft model to propose candidate tokens that are verified in parallel by the target model. However, existing draft model training objectives are not directly aligned with the inference-time goal of maximizing consecutive token acceptance. To address this issue, we reformulate the draft model optimization objective, shifting the focus from token prediction accuracy to the overall acceptance length. In this paper, we build upon PARD to propose PARD-2, a dual-mode speculative decoding framework with Confidence-Adaptive Token (CAT) optimization. This approach adaptively reweights each token to better align with the verification process. Notably, PARD-2 enables a single draft model to support both target-dependent and target-independent modes. Experiments across diverse models and tasks demonstrate that PARD-2 achieves up to 6.94×\times lossless acceleration, surpassing EAGLE-3 by 1.9×\times and PARD by 1.3×\times on Llama3.1-8B. Our code is available at https://github.com/AMD-AGI/PARD.

Zihao An, Taichi Liu, Ziqiong Liu +3

Advanced Micro Devices, Inc. · Rutgers University · Michigan State University

Speculative DecodingLLM Inference Acceleration
AskPDF
May 9, 2026

HAServe: Horizon-Aware Scheduling and Memory Co-Design for Efficient LLM Serving

Online large language model (LLM) services process requests that share context but differ in prefix structure and arrival timing. Finite capacity couples request admission and key-value (KV) retention. This paper addresses one online serving problem: using a unified view of visible demand to choose the feasible joint admission and KV-retention candidate with the lowest predicted latency cost as prefix structures and arrival rates vary. To model this problem, we introduce HorizonSchema, a workload abstraction that connects exact prefix paths, permitted serializations, and released consumers as shared inputs to both decisions. Our analysis shows that useful reuse depends on demand and prefix survival, while service order changes latency even at equal computation. Furthermore, we present HAServe, which turns this abstraction into a scheduling--memory co-design method of systematic optimization for online LLM serving. We implement HAServe on SGLang and evaluate Qwen3-4B-Instruct-2507 and Llama2-13B on five workloads using eight H100 GPUs. Across the evaluated workloads, HAServe achieves up to 1.141.14--5.90×5.90\times P99 TTFT speedups and up to 1.081.08--1.44×1.44\times the throughput, with KV-cache hit rate up to 20.59 percentage points higher than the strongest baseline for each metric and setting.

Xingyu Qu, Tianhao Lin, Yiqi Li +2

School of Computer Science, Wuhan University · Amazon

LLM ServingKV CachingKV-Cache ManagementLLM Inference AccelerationPrefix CachingLLM Request Scheduling
AskPDF
May 9, 2026

Uncovering Intra-expert Activation Sparsity for Efficient Mixture-of-Expert Model Execution

Mixture of Experts (MoE) architecture has become the standard for state-of-the-art large language models, owing to its computational efficiency through sparse expert activation. However, sparsity through finer expert granularity is becoming increasingly difficult to achieve due to fundamental training challenges such as expert collapse and load imbalance. In this work, we explore and leverage intra-expert activation sparsity as a complementary and underexplored dimension of sparsity in MoE models. Surprisingly, substantial intra-expert sparsity is readily available in existing pre-trained MoE models, without any modification to the activation function or model parameters, providing up to 90% sparsity within each expert without significant accuracy loss. We explore intra-expert activation sparsity across eight off-the-shelf MoE models ranging from 1B to 400B parameters, and extend the MoE execution pipeline of vLLM to leverage intra-expert activation sparsity by skipping the computations of inactive neurons, on top of its existing optimizations, achieving up to 2.5 times speedup in MoE layer execution and 1.2 times end-to-end speedup compared to the original dense vLLM baseline.

Jongseok Park, Sunga Kim, Zhenyu Gu +2

University of California, Berkeley · Work done during an internship at Advanced Micro Devices, Inc. · Advanced Micro Devices, Inc.

Activation SparsityMixture-of-Experts InferenceLLM Inference AccelerationMixture-of-Experts Models
AskPDF
May 9, 2026

Different Prompts, Different Ranks: Prompt-aware Dynamic Rank Selection for SVD-based LLM Compression

Large language models (LLMs) have rapidly grown in scale, creating substantial memory and computational costs that hinder efficient deployment. Singular value decomposition (SVD) has emerged as an effective post-training compression technique, but existing SVD-based methods rely on static rank truncation, applying a fixed prefix of singular components to all inputs regardless of their diversity. We identify two limitations of this static design: the optimal rank varies across individual prompts, and the selected rank is sensitive to the choice of calibration set, leading to suboptimal performance across diverse inputs. To address these challenges, we propose PARSE\textbf{PARSE}, a post-training framework for P\textbf{P}rompt-A\textbf{A}ware R\textbf{R}ank S\textbf{S}election as E\textbf{E}xperts in SVD-compressed LLMs. PARSE trains a linear router offline to perform prompt-aware rank selection, decoupling it from calibration information by supervising the router against dense-model outputs on a large-scale corpus. We further observe that rank-selection patterns are shared across semantically similar prompts and remain stable across decoding steps, allowing appropriate rank subsets to be served directly from a pattern cache at inference. Complemented by expert memory aggregation and kernel fusion for system-level efficiency, PARSE is orthogonal to existing SVD-based pipelines and consistently improves both model quality and inference efficiency. Integrated with four representative SVD-based methods, PARSE improves average task accuracy by up to 10% at a compression ratio of 0.6 on LLaMA-7B, and achieves up to 2.5 ×\times prefill and 2.4 ×\times decode speedup over native SVD execution.

Hengyi Zhu, Zhendong Mi, Grace Li Zhang +1

1Stevens Institute of Technology, Hoboken, USA · 2Technical University of Darmstadt, Darmstadt, Germany

LLM CompressionLow-Rank Matrix DecompositionLLM Inference AccelerationLow-Rank CompressionLow-Rank Approximation
AskPDF
May 8, 2026

Fast Byte Latent Transformer

Recent byte-level language models (LMs) match the performance of token-level models without relying on subword vocabularies, yet their utility is limited by slow, byte-by-byte autoregressive generation. We address this bottleneck in the Byte Latent Transformer (BLT) through new training and generation techniques. First, we introduce BLT Diffusion (BLT-D), a new model and our fastest BLT variant, trained with an auxiliary block-wise diffusion objective alongside the standard next-byte prediction loss. This enables an inference procedure that generates multiple bytes in parallel per decoding step, substantially reducing the number of forward passes required to generate a sequence. Second, we propose two extensions inspired by speculative decoding that trade some of this speed for higher generation quality: BLT Self-speculation (BLT-S), in which BLT's local decoder continues generating past its normal patch boundaries to draft bytes, which are then verified with a single full-model forward pass; and BLT Diffusion+Verification (BLT-DV), which augments BLT-D with an autoregressive verification step after diffusion-based generation. All methods may achieve an estimated memory-bandwidth cost over 50% lower than BLT on generation tasks. Each approach offers its own unique advantages, together removing key barriers to the practical use of byte-level LMs.

Julie Kallini, Artidoro Pagnoni, Tomasz Limisiewicz +5

FAIR at Meta · Stanford University · University of Washington

Speculative DecodingLLM Inference AccelerationByte-Level Language ModelLanguage ModelingDiffusion Language Model DecodingDiffusion Language Models
AskPDF
May 8, 2026

FlashSVD v1.5: Making Low-Rank Transformers Inference Actually Fast

SVD-based Low-rank compression reduces transformer parameters and nominal FLOPs, but these savings often translate poorly into real LLM serving speedups. We show that this gap is largely a runtime problem: factorized checkpoints fragment execution paths, and the resulting overhead differs substantially between prefill and autoregressive decode. We present FlashSVD v1.5, a unified inference runtime for serving SVD-compressed transformers. FlashSVD v1.5 maps diverse public SVD compression families to a common factorized representation and combines phase-specific kernels with dense-KV decode, packed MLP execution, and per-layer CUDA-graph replay to reorganize the low-rank serving path into a thin runtime. Across representative decoder-serving settings, FlashSVD v1.5 achieves up to 2.55x decode and 2.39x end-to-end speedup, and it attains 1.48x average decode and 1.44x average end-to-end speedup across multiple popular SVD compression families. These results suggest that practical low-rank acceleration requires runtime co-design, not compression algorithms alone. Our code is available at: https://github.com/Zishan-Shao/FlashSVD.

Wenhao Wu, Zishan Shao, Kangning Cui +5

Department of Electrical & Computer Engineering, Duke University · Department of Statistical Science, Duke University · Department of Computer Science, Wake Forest University +1

Efficient Transformer InferenceLLM CompressionLLM Inference AccelerationLow-Rank CompressionSingular Value Decomposition
AskPDF
May 8, 2026

An Efficient Hybrid Sparse Attention with CPU-GPU Parallelism for Long-Context Inference

Long-context inference increasingly operates over CPU-resident KV caches, either because decoding-time KV states exceed GPU memory capacity or because disaggregated prefill-decode systems place KV data in host memory. Although block-sparse attention reduces attention cost in this setting, sparsity alone is insufficient for end-to-end efficiency. GPU-only designs remain constrained by PCIe bandwidth and metadata memory overhead, while CPU-GPU hybrid designs still suffer from substantial GPU idle time and bottlenecks in CPU-side top-k selection and sparse attention computation. Fluxion is built on three key insights: output-aware KV budgeting, head-specific and granularity-aware sparse configuration, and cross-device coordinated execution for sparse attention over CPU-resident KV caches. Guided by these insights, Fluxion combines a lightweight head-property predictor, a granularity-budget selector, and a priority-based scheduler to jointly optimize budget allocation, sparse configuration, and CPU-GPU execution overlap. This co-design enables hybrid sparse attention to achieve both accuracy and system efficiency in long-context inference. Across 2 models, 3 benchmarks, and 40 tasks, Fluxion preserves quality well -- the worst average degradation is only -0.26 relative to FULL, while delivering 1.5×\times-3.7×\times speedup over the strongest fixed sparse hybrid baseline, whose KV budget is only 0.05.

Feiyu Yao, Zhixiong Niu, Xiaqing Li +3

Beijing University of Technology Beijing, China · Microsoft Research Beijing, China · Beijing Jiaotong University Beijing, China

GPU AccelerationSelf-AttentionLong-Context Language Model InferenceKV-Cache ManagementLLM Inference AccelerationHeterogeneous Computing+2
AskPDF
May 8, 2026

Priming: Hybrid State Space Models From Pre-trained Transformers

Hybrid State-Space models combine Attention with recurrent State-Space Model (SSM) layers, balancing eidetic memory from Attention with compressed fading memory from SSMs. This yields smaller Key-Value caches and faster decoding than Transformers, along with a richer architectural design space. Exploring that design space at scale has so far required training from scratch, a barrier that has kept most large-model Hybrid research within a narrow range of architectures. We introduce Priming, a method that turns Hybrid architecture design from a pre-training problem into a knowledge transfer one. Priming initializes a Hybrid model from a pre-trained Transformer and, through short alignment and post-training phases, recovers downstream quality using less than 0.5% of the source model's pre-training token budget. Priming is agnostic to the source Transformer family (e.g., Qwen, Llama, Mistral), model class (dense or Mixture-of-Experts), and model scale. Priming enables us to run the first controlled comparison of SSM layer types at scale under identical conditions. We evaluate, Gated KalmaNet (GKA), Gated DeltaNet (GDN), and Mamba-2, and show that their expressiveness hierarchy, GKA>GDN>Mamba-2, directly predicts downstream performance on long-context reasoning tasks. We scale Priming to 8B/32B reasoning models with native 128K contexts. Our Hybrid GKA 32B improves over its source Qwen3-32B by +3.8 average reasoning points, while staying within 1% of a Transformer post-trained on the same data and enabling up to 2.3x higher decode throughput. To foster research on Hybrid architectures, we release a model zoo of primed Hybrid models for long-context reasoning and instruction following, together with the Priming training and inference code (Sequence Parallelism algorithms for long-context training, optimized GKA kernels, and vLLM serving plugin), all under Apache~2.0 License.

Aditya Chattopadhyay, Elvis Nunez, Prannay Kaul +6

AWS Agentic AI

State Space ModelsLong-Context Language Model InferenceLLM Inference AccelerationTransfer LearningLong-Context InferenceHybrid Transformer-State Space Model Architectures+2
AskPDF
May 8, 2026

MISA: Mixture of Indexer Sparse Attention for Long-Context LLM Inference

DeepSeek Sparse Attention (DSA) sets the state of the art for fine-grained inference-time sparse attention by introducing a learned token-wise indexer that scores every prefix token and selects the most relevant ones for the main attention. To remain expressive, the indexer uses many query heads (for example, 64 on DeepSeek-V3.2) that share the same selected token set; this multi-head design is precisely what makes the indexer the dominant cost on long contexts. We propose MISA (Mixture of Indexer Sparse Attention), a drop-in replacement for the DSA indexer that treats its indexer heads as a pool of mixture-of-experts. A lightweight router uses cheap block-level statistics to pick a query-dependent subset of only a few active heads, and only those heads run the heavy token-level scoring. This preserves the diversity of the original indexer pool while reducing the per-query cost from scoring every prefix token with every head to scoring it with only a handful of routed heads, plus a negligible router term computed on a small set of pooled keys. We further introduce a hierarchical variant of MISA that uses the routed pass to keep an enlarged candidate set and then re-ranks it with the original DSA indexer to recover the final selected tokens almost exactly. With only eight active heads and no additional training, MISA matches the dense DSA indexer on LongBench across DeepSeek-V3.2 and GLM-5 while running with eight and four times fewer indexer heads respectively, and outperforms HISA on average. It also preserves fully green Needle-in-a-Haystack heatmaps up to a 128K-token context and recovers more than 92% of the tokens selected by the DSA indexer per layer. Our TileLang kernel delivers roughly a 3.82 times speedup over DSA's original indexer kernel on a single NVIDIA H200 GPU.

Ruijie Zhou, Fanxu Meng, Yufei Xu +4

Self-AttentionLong-Context Language Model InferenceLLM Inference AccelerationSparse Attention
AskPDF
May 8, 2026

CSR: Infinite-Horizon Real-Time Policies with Massive Cached State Representations

Deploying massive large language models (LLMs) as continuous cognitive engines for robotics is bottlenecked by the time-to-first-token (TTFT) latency required to process extensive state histories. Existing solutions like RAG or sliding windows compromise global context or incur prohibitive re-computation costs. We formalize the optimal task structure for minimizing latency and theoretically prove that prefix stability, incremental extensibility, and asynchronous state reconciliation are necessary conditions for real-time performance. Building on these proofs, we introduce the Cached State Representation (CSR) framework as the practical instantiation of these properties, ensuring optimal KV-cache reuse. To sustain these properties over infinite horizons, we further propose an Asynchronous State Reconciliation (ASR) algorithm that offloads state memory eviction to a parallel computational resource to eliminate latency spikes. On a physical robot wirelessly connected to an on-premise GPU server, CSR achieves a 26-fold latency reduction (14.67s to 0.56s) for 120K token contexts with a 235B parameter model compared to a standard baseline. On an embodied AI benchmark, we achieve SOTA recall (0.836 vs. 0.459) while maintaining RAG-level latency. ASR is validated to sustain bounded, spike-free TTFT over 10 eviction cycles in continuous real-world operation. Together, CSR and ASR enable massive LLMs to function as continuously operating, high-frequency (> 2 Hz) embodied policies.

Robin Karlsson, Go Suzui

1GODOT Inc, Hyogo, Japan · 2Graduate School of Informatics, Nagoya University, Aichi, Japan.

Robotic ControlLanguage Model-Based ControlKV-Cache ManagementLLM Inference AccelerationMemory-Augmented Language Model Agents
AskPDF
May 8, 2026

LaTER: Efficient Test-Time Reasoning via Latent Exploration and Explicit Verification

Chain-of-thought (CoT) reasoning improves large language models (LLMs) on difficult tasks, but it also makes inference expensive because every intermediate step must be generated as a discrete token. Latent reasoning reduces visible token generation by propagating continuous states, yet replacing explicit derivations with latent computation can hurt tasks that require symbolic checking. We propose Latent-Then-Explicit Reasoning (LaTER), a two-stage paradigm that first performs bounded exploration in a continuous latent space and then switches to explicit CoT for verification and answer generation. In a training-free instantiation, LaTER projects final-layer hidden states back to the input embedding space, preserves the latent KV cache, and uses entropy and model-native stop-token probes to decide when to switch. We find that strong reasoning models already exhibit structured latent trajectories under this interface. On Qwen3-14B, training-free LaTER reduces total token usage by 16%-32% on several benchmarks while matching or improving accuracy on most of them; for example, it improves AIME 2025 from 70.0% to 73.3% while reducing tokens from 15,730 to 10,661. We further construct Latent-Switch-69K, a supervised corpus that pairs condensed solution intuitions with shortened explicit derivations. Fine-tuning with latent rollout and halting supervision yields additional gains: trained LaTER reaches 80.0% accuracy on AIME 2025, 10.0 points above the standard CoT baseline, while using 33% fewer tokens. Our code, data, and model are available at https://github.com/TioeAre/LaTER.

Xuan Li, Yining Wang, Yuchen Liu +7

University of Science and Technology of China · Unisound AI Technology Co., Ltd · MAIS, Institute of Automation, Chinese Academy of Sciences

CoT ReasoningEfficient Language Model ReasoningLLM Inference AccelerationContinuous Latent ReasoningLLM ReasoningLatent Reasoning+2
AskPDF
May 8, 2026

SpecBlock: Block-Iterative Speculative Decoding with Dynamic Tree Drafting

Speculative decoding accelerates LLM inference by drafting a tree of candidate continuations and verifying it in one target forward. Existing drafters fall into two camps with opposite weaknesses. Autoregressive drafters such as EAGLE-3 preserve dependence along each draft path but call the drafter once per tree depth, making drafting a non-trivial share of per-iteration latency. Parallel drafters cut drafter calls by predicting multiple future positions in one forward, but each position is predicted without seeing the others, producing paths the verifier rejects. In this paper, we propose SpecBlock, a block-iterative drafter that combines path dependence with cheap drafting. Each drafter forward produces K dependent positions and we call this a block. The draft tree grows through repeated block expansions. Two mechanisms explicitly carry path dependence to keep later draft positions accurate. Within each block, a layer-wise shift carries the previous position's hidden state into every decoder layer. Across blocks, each new block can start from any position of the previous block, inheriting its hidden state to extend the path. To spend verifier budget where acceptance is likely, a co-trained rank head replaces the fixed top-k tree by allocating per-position branching during drafting. To avoid training the drafter on prefixes it never produces at inference, a valid-prefix mask drops the loss at later positions once an earlier one is wrong. Beyond static drafting, a cost-aware bandit at deployment uses free verifier feedback to update the drafter selectively, only when the expected throughput gain exceeds the update cost. Experiments show that SpecBlock improves mean speedup by 8-13% over EAGLE-3 at 44-52% of its drafting cost, and cost-aware adaptation extends this lead to 11-19%.

Weijie Shi, Qiang Xu, Fan Deng +9

Hong Kong University of Science and Technology · MetaX · Zhejiang Normal University +1

Speculative DecodingLLM Inference Acceleration
AskPDF
May 7, 2026

Sparse Attention as a Range Searching Problem: Towards an Inference-Efficient Index for KV Cache

Sparse attention improves LLM inference efficiency by selecting a subset of key-value entries, but at the cost of potential accuracy degradation. In particular, omitting critical KV entries can induce substantial errors in model outputs. Existing methods typically operate under fixed or adaptive token budgets and provide empirical robustness or partial theoretical guarantees, yet they do not ensure zero false negatives in decoding steps, particularly since the set of relevant tokens is both query- and step-dependent. Our empirical observations confirm that missing even one critical key can lead to sharp error spikes, especially in long reasoning tasks where the set of important tokens varies throughout decoding. This observation motivates the need for indexing methods that dynamically adapt to these variations across decoding steps while guaranteeing a full recall of the relevant keys above a certain threshold. We address this challenge by reformulating sparse attention as the halfspace range searching problem. However, existing range searching indices are not suitable for modern LLM inference due to their computational and implementation overheads. To overcome this, we introduce Louver, a novel index structure tailored for efficient KV cache retrieval. Louver (i) guarantees zero false negatives with respect to a specified threshold in both theory and practice, (ii) is lightweight to integrate into existing LLM pipelines, and (iii) incorporates hardware-aware optimizations for both CPU and GPU executions. Our experiments demonstrate that Louver outperforms prior sparse attention methods in both accuracy and runtime, and is faster than highly optimized dense attentions such as FlashAttention. These results highlight that recall guarantees are a critical and overlooked dimension of sparse attention, and open a new direction for building theoretically grounded, efficient KV cache indices.

Mohsen Dehghankar, Abolfazl Asudeh

Department of Computer Science University of Illinois Chicago

Self-AttentionKV CachingKV-Cache ManagementLLM Inference AccelerationSparse Attention
AskPDF
May 7, 2026

Grokers: Bottom-Up Inductive Comprehension and Write-Time Intelligence over Typed Knowledge Graphs

We present Grokers, an architecture for building persistent, structured comprehension of typed knowledge graphs through bottom-up inductive traversal of dependency subgraphs. Unlike retrieval-augmented generation (RAG), which pays full comprehension cost at every query, Grokers pushes intelligence to write time: autonomous Groker agents analyze nodes in a typed stream graph, extract structured attributes via governed language model (LM) calls, and inductively compose that understanding upward through dependency relations, writing enriched typed attributes that serve all future queries at zero additional LM cost. We prove three formal properties: (1) the Byte-Identity Theorem, establishing that context blocks assembled from a transactionally-maintained denormalization index are byte-identical across LM turns between semantic changes, enabling KV-cache hit rates approaching 100%; (2) the Accumulation Monotonicity Theorem, establishing that the fraction of interactions resolved without LM calls is non-decreasing in the number of completed interactions under a governed wisdom library growth protocol; and (3) the Dual-Traversal Ordering Theorem, establishing that top-down generation and bottom-up comprehension are the unique correct traversal orderings for their respective tasks over a dependency DAG, and that their composition closes into a complete generation-comprehension cycle. We further present a deterministic alternative to embedding-based semantic search, with a synonym caching protocol whose LM fallback rate converges to zero for finite-vocabulary domains. A reference implementation is provided in the open-source Qbix / Safebox / Safebots stack.

Gregory Magarshak

Qbix, Inc. & Intercoin, Inc. · IE University NYC · New York, NY, USA

KG ConstructionMemory-Augmented Language ModelsKnowledge RepresentationLLM Inference AccelerationGraph ReasoningKnowledge Graphs
AskPDF
May 7, 2026

Litespark Inference For CPUs: Ultra-Fast SIMD Framework for Ternary (1.58-bit) Language Models

Large language models (LLMs) have transformed artificial intelligence, but their computational requirements remain prohibitive for most users. Standard inference demands expensive datacenter GPUs or cloud API access, leaving over one billion personal computers underutilized for AI workloads. Ternary models offer a path forward: their weights are constrained to {-1, 0, +1}, theoretically eliminating the need for floating-point multiplication. However, existing frameworks fail to exploit this structure, treating ternary models as dense floating-point networks. We address this gap with custom SIMD kernels that replace matrix multiplication with simple addition and subtraction operations, targeting the integer dot product instructions available on modern CPUs. Our implementation, Litespark-Inference, is pip-installable and integrates directly with Hugging-Face, achieving 18.15x higher throughput, 7.15x faster time-to-first-token and 6.03x memory reduction compared to standard PyTorch inference on Apple Silicon, with comparable or higher throughput speedups up to 95.81x on Intel and AMD processors.

Nii Osae Osae Dade, Tony Morri, Moinul Hossain Rahat +2

Mindbeam AI

LLM QuantizationEfficient Neural Network InferenceLLM InferenceLLM Inference AccelerationLanguage Modeling
AskPDF
May 7, 2026

LatentRAG: Latent Reasoning and Retrieval for Efficient Agentic RAG

Single-step retrieval-augmented generation (RAG) provides an efficient way to incorporate external information for simple question answering tasks but struggles with complex questions. Agentic RAG extends this paradigm by replacing single-step retrieval with a multi-step process, in which the large language model (LLM) acts as a search agent that generates intermediate thoughts and subqueries to iteratively interact with the retrieval system. This iterative process incurs substantial latency due to the autoregressive generation of lengthy thoughts and subqueries. To address this limitation, we propose LatentRAG, a novel framework that shifts both reasoning and retrieval from discrete language space to continuous latent space. Unlike existing explicit methods that generate natural language thoughts or subqueries token-by-token, LatentRAG produces latent tokens for thoughts and subqueries directly from the hidden states in a single forward pass. We align LLMs with dense retrieval models in the latent space, enabling retrieval over latent subquery tokens and supporting end-to-end joint optimization. To improve transparency and encourage semantically meaningful latent representations, we incorporate a parallel latent decoding mechanism that translates latent tokens back into natural language. Extensive experiments on seven benchmark datasets show that LatentRAG achieves performance comparable to explicit agentic RAG methods while reducing inference latency by approximately 90%, substantially narrowing the latency gap with traditional single-step RAG.

Yijia Zheng, Marcel Worring

University of Amsterdam, Amsterdam, the Netherlands

LLM Inference AccelerationAgentic RAGLatent Reasoning
AskPDF
May 7, 2026

UniPrefill: Universal Long-Context Prefill Acceleration via Block-wise Dynamic Sparsification

As large language models (LLMs) continue to advance rapidly, they are becoming increasingly capable while simultaneously demanding ever-longer context lengths. To improve the inference efficiency of long-context processing, several novel low-complexity hybrid architectures have recently been proposed, effectively alleviating the computational burden of long-context inference. However, existing research on long-context prefill acceleration remains predominantly focused on sparse attention mechanisms, which achieve their maximum speedup only on full-attention models. When transferred to emerging architectures--such as linear/full attention hybrids or sliding window/full attention hybrids--these prefill acceleration approaches suffer significant performance degradation. Furthermore, such methods are generally incompatible with continuous batching, making them difficult to integrate into modern inference engines such as vLLM. To this end, we propose UniPrefill, a prefill acceleration framework applicable to virtually any model architecture, which directly accelerates the model's computation at the token level. We further implement UniPrefill as a continuous batching operator and extend vLLM's scheduling strategy to natively support prefill-decode co-processing and tensor parallel for UniPrefill, enabling its seamless integration into vLLM. UniPrefill achieves up to 2.1x speedup in Time-To-First-Token (TTFT), with the acceleration becoming increasingly pronounced as the number of concurrent requests grows.

Qihang Fan, Huaibo Huang, Zhiying Wu +2

1MAIS&NLPR, CASIA · 2UCAS · 3WeChat, Tencent

Long-Context Language Model InferenceLLM Inference Acceleration
AskPDF
May 7, 2026

Federation of Experts: Communication Efficient Distributed Inference for Large Language Models

Mixture of experts has emerged as the primary mechanism for making Large Language Models (LLMs) computationally efficient. However, in distributed settings, communicating token embeddings between experts is a significant bottleneck. We present the novel Federation of Experts (FoE) architecture. FoE restructures the MoE block of a transformer layer into multiple MoE clusters. Each cluster is responsible for only one of the KV heads and expert parallelism is applied between those experts. Between clusters, a sum synchronizes the post-attention residuals, which then drives routing and dispatch for the next MoE block. In a single-node setting, FoE completely eliminates all-to-all communication as all experts within a group are contained on the same GPU. In multi-node settings, FoE confines all-to-all communication to the intra-node fabric, thus significantly reducing communication overhead. An implementation of FoE finds that on LongBench, FoE significantly improves inference throughput and latency in both single-node and multi-node settings, reducing the end-to-end forward-pass latency by up to 5.2x, TTFT by 3.62x, and TBT by 1.95x. It does so while achieving comparable generation quality to a mixture of experts model of the same size and training configuration.

Muhammad Shahir Abdurrahman, Chun Deng, Azalia Mirhoseini +1

Stanford University

Expert ParallelismMixture-of-Experts InferenceLLM Inference AccelerationDistributed Inference
AskPDF