Privacy-Aware Split Inference with Speculative Decoding for Large Language Models over Wide-Area Networks

• URL: http://arxiv.org/abs/2602.16760v1
• Date: Wed, 18 Feb 2026 14:13:08 GMT
• Title: Privacy-Aware Split Inference with Speculative Decoding for Large Language Models over Wide-Area Networks
• Abstract summary: We present a practical system for privacy-aware large language model (LLM) inference that splits a transformer between a trusted local GPU and a cloud GPU.<n>Our system addresses the unique challenges of autoregressive LLM decoding over high-latency wide-area networks (WANs)<n>System generalizes to larger models with only 4.9 GB local VRAM and matching 7B throughput.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We present a practical system for privacy-aware large language model (LLM) inference that splits a transformer between a trusted local GPU and an untrusted cloud GPU, communicating only intermediate activations over the network. Our system addresses the unique challenges of autoregressive LLM decoding over high-latency wide-area networks (WANs), contributing: (1) an asymmetric layer split where embedding and unembedding layers remain local, ensuring raw tokens never leave the trusted device; (2) the first application of lookahead decoding to split inference over WANs, amortizing network round-trip latency across multiple tokens per iteration; (3) an empirical inversion attack evaluation showing that split depth provides a tunable privacy-performance tradeoff -- an attacker can recover ~59%% of tokens at a 2-layer split but only ~35%% at an 8-layer split, with minimal throughput impact; (4) ablation experiments showing that n-gram speculation accepts 1.2-1.3 tokens per decoding step on average (peak of 7 observed on code), with acceptance rates consistent across model scales; (5) formal verification that lookahead decoding produces token-identical output to sequential decoding under greedy argmax, with zero quality degradation; and (6) scaling validation on Mistral NeMo 12B (40 layers), demonstrating that the system generalizes to larger models with only 4.9 GB local VRAM and matching 7B throughput. Evaluated on Mistral 7B and NeMo 12B over a ~80ms WAN link, our system achieves 8.7-9.3 tok/s (7B) and 7.8-8.7 tok/s (12B) with lookahead decoding, with an RTT decomposition model (validated at <6.2%% cross-validation error) projecting 15-19 tok/s at 20ms RTT.

Related papers

• Architecture-Dependent Causal Transfer of Activation States Across Large Language Models [51.56484100374058]
  Internal language activation states can be transferred causally between different large language models.<n>Injecting projected activations into a target model during generation produces a statistically significant, preregistered causal effect.<n>We conclude that end-to-end activation-state transfer between LLMs, as currently implemented, is architecture-dependent rather than universal.
  arXiv  Detail & Related papers  (2026-08-17T09:53:27Z)
• Differentiable Logic Gate Networks for Low-Latency EEG Classification on Edge Devices [0.0]
  Real-time EEG classification on edge devices is bottlenecked by the floating-point arithmetic of conventional neural networks.<n>We investigated Differentiable Logic Gate Networks (Diff-Logic) as a hardware-native alternative.<n>Our results establish logic-based neural architectures as a practical paradigm for resource-constrained brain-computer interfaces.
  arXiv  Detail & Related papers  (2026-07-20T16:49:30Z)
• Constrained Decoding for Diffusion Language Models via Efficient Inference over Finite Automata [57.27430779838529]
  We present an exact and tractable algorithm for sampling from the constrained mean-field posterior under any constraint expressible as a finite automaton.<n>The approach guarantees constraint satisfaction by construction, supports both greedy and sampling-based decoding, and is compatible with parallel and block-wise decoding.<n> Empirical evaluations on Dream-7B and LLaDA-8 show substantial accuracy gains across various tasks.
  arXiv  Detail & Related papers  (2026-07-08T05:48:57Z)
• MIRAGE: A Polarity-Flipping Encoding Subspace in LLM Agents [2.896184275229374]
  LLM agents are coerced into covertly encoding sensitive data.<n>The resulting outputs evade output-side detection but the underlying computation does not.<n>We build MIRAGE, a two-channel real-time monitor exploiting both signals.
  arXiv  Detail & Related papers  (2026-06-09T01:45:19Z)
• Operator Fusion for LLM Inference on the Tensix Architecture [13.215957873912181]
  This study addresses on-device inference of Transformer models on Tenstorrent's Tensix architecture.<n> RMSNorm is fused with matrix multiplication in self-attention and in the FFN, enabling back-to-back execution of memory-bound and compute-bound operators.
  arXiv  Detail & Related papers  (2026-06-03T02:48:34Z)
• Scaling Parallel Sequence Models to Foundation-Scale Vision Encoders [123.58723804218151]
  Vision foundation models are bottlenecked by the quadratic cost of self-attention.<n>We present C-GSPN based on 2D spatial propagation propagation.
  arXiv  Detail & Related papers  (2026-05-30T14:29:43Z)
• Human-Inspired Memory Architecture for LLM Agents [0.9507070656654629]
  We present a biologically-grounded memory architecture comprising six cognitive mechanisms.<n>Each mechanism addresses a specific failure mode of naive memory accumulation.<n>At S-tier scale (50 sessions), dedup-based consolidation yields a +13.3 pp improvement in preference recall.
  arXiv  Detail & Related papers  (2026-05-08T22:52:37Z)
• WISV: Wireless-Informed Semantic Verification for Distributed Speculative Decoding in Device-Edge LLM Inference [56.297697169678095]
  WISV (Wireless-Informed Semantic Verification) is a novel distributed speculative decoding framework.<n>We show that WISV achieves up to a 60.8% increase in accepted length, a 37.3% reduction in interaction rounds, and a 31.4% improvement in end-to-end latency.<n>We validate WISV on a hardware testbed comprising an NVIDIA Jetson AGX Orin and an A40-equipped server.
  arXiv  Detail & Related papers  (2026-04-20T01:29:56Z)
• STEP3-VL-10B Technical Report [115.89015065130127]
  STEP3-VL-10B is a lightweight foundation model designed to redefine the trade-off between compact efficiency and frontier-level multimodal intelligence.<n>We implement Parallel Coordinated Reasoning (PaCoRe) to scale test-time compute, allocating resources to scalable perceptual reasoning.<n>It records 92.2% on MMBench and 80.11% on MMMU, while excelling in complex reasoning with 94.43% on AIME2025 and 75.95% on MathVision.
  arXiv  Detail & Related papers  (2026-01-14T17:58:24Z)
• Towards a Science of Scaling Agent Systems [79.64446272302287]
  We formalize a definition for agent evaluation and characterize scaling laws as the interplay between agent quantity, coordination structure, modelic, and task properties.<n>We derive a predictive model using coordination metrics, that cross-validated R2=0, enabling prediction on unseen task domains.<n>We identify three effects: (1) a tool-coordination trade-off: under fixed computational budgets, tool-heavy tasks suffer disproportionately from multi-agent overhead, and (2) a capability saturation: coordination yields diminishing or negative returns once single-agent baselines exceed 45%.
  arXiv  Detail & Related papers  (2025-12-09T06:52:21Z)
• Incentivised Orchestrated Training Architecture (IOTA): A Technical Primer for Release [0.0]
  IOTA transforms SN9's previously isolated competitors into a single cooperating unit.<n> Validators measure each miner's contribution and allocate token emissions proportionally.<n>We use model-bottlenecks to cut communication bandwidths of activations by up to 128x.
  arXiv  Detail & Related papers  (2025-07-16T15:16:21Z)
• FusionLLM: A Decentralized LLM Training System on Geo-distributed GPUs with Adaptive Compression [55.992528247880685]
  Decentralized training faces significant challenges regarding system design and efficiency. We present FusionLLM, a decentralized training system designed and implemented for training large deep neural networks (DNNs) We show that our system and method can achieve 1.45 - 9.39x speedup compared to baseline methods while ensuring convergence.
  arXiv  Detail & Related papers  (2024-10-16T16:13:19Z)
• KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization [67.74400574357472]
  LLMs are seeing growing use for applications which require large context windows, and with these large context windows KV cache activations surface as the dominant contributor to memory consumption during inference.<n> Quantization is a promising approach for compressing KV cache activations; however, existing solutions fail to represent activations accurately in sub-4-bit precision.<n>Our work, KVQuant, facilitates low precision KV cache quantization by incorporating several novel methods.
  arXiv  Detail & Related papers  (2024-01-31T18:58:14Z)
• Benchmarking Local Robustness of High-Accuracy Binary Neural Networks for Enhanced Traffic Sign Recognition [0.0]
  This paper introduces a set of benchmark problems featuring layers that challenge state-of-the-art verification tools. The difficulty of the verification problem is given by the high number of network parameters.
  arXiv  Detail & Related papers  (2023-09-25T01:17:14Z)
• Semi-supervised multiscale dual-encoding method for faulty traffic data detection [0.0]
  We introduce a variational autoencoder (VAE) based semi-supervised method for detection of faulty traffic data. Our experiments show that the proposed architecture with the dual encoding scheme, coupled with attention module, outperforms other encoding schemes.
  arXiv  Detail & Related papers  (2022-12-27T20:07:52Z)
• Sub 8-Bit Quantization of Streaming Keyword Spotting Models for Embedded Chipsets [7.5195830365852085]
  We propose a novel sub 8-bit quantization aware training algorithm for all components of a 250K parameter feedforward, streaming, state-free keyword spotting model. We conduct large scale experiments, training on 26,000 hours of de-identified production, far-field and near-field audio data.
  arXiv  Detail & Related papers  (2022-07-13T17:46:08Z)
• A Unified Model for Multi-class Anomaly Detection [33.534990722449066]
  UniAD accomplishes anomaly detection for multiple classes with a unified framework. We evaluate our algorithm on MVTec-AD and CIFAR-10 datasets.
  arXiv  Detail & Related papers  (2022-06-08T06:05:09Z)
• Enabling certification of verification-agnostic networks via memory-efficient semidefinite programming [97.40955121478716]
  We propose a first-order dual SDP algorithm that requires memory only linear in the total number of network activations. We significantly improve L-inf verified robust accuracy from 1% to 88% and 6% to 40% respectively. We also demonstrate tight verification of a quadratic stability specification for the decoder of a variational autoencoder.
  arXiv  Detail & Related papers  (2020-10-22T12:32:29Z)
• ReActNet: Towards Precise Binary Neural Network with Generalized Activation Functions [76.05981545084738]
  We propose several ideas for enhancing a binary network to close its accuracy gap from real-valued networks without incurring any additional computational cost. We first construct a baseline network by modifying and binarizing a compact real-valued network with parameter-free shortcuts. We show that the proposed ReActNet outperforms all the state-of-the-arts by a large margin.
  arXiv  Detail & Related papers  (2020-03-07T02:12:02Z)

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

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