coding_matrix.csv

date,short_name,title,draft_source,draft_geometry,verification_fidelity,runtime_execution,paper_url,github_url,project_urls
2026-07,DSpark,DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation,Semi-Autoregressive External Drafter [core],Block [core],Distribution-Level Exactness [core],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2607.05147,https://github.com/deepseek-ai/DeepSpec,
2026-06,WhiFlash,WhiFlash: Accelerating Speculative Decoding with Token-Level Cross-Paradigm Routing,Non-Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Block [inherited]; Tree [inherited],Token-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2606.07710,,
2026-06,VIA-SD,VIA-SD: Verification via Intra-Model Routing for Speculative Decoding,Autoregressive External Drafter [core],Chain [core],Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2606.12243,,https://zju-xyc.github.io/VIA-SD-Project-Page/
2026-06,TreeFlash,TreeFlash: Parallel AR-Approximation for Faster Speculative Decoding,Non-Autoregressive External Drafter [core],Tree [core],Token-Level Exactness [inherited],Speculation Control [inherited]; Serving Systems [inherited],https://arxiv.org/abs/2606.03819,https://github.com/ETH-DISCO/TreeFlash,
2026-06,TAPS-Tree,TAPS: Target-Aware Prefix Tree Selection for Diffusion-Drafted Speculative Decoding,Non-Autoregressive External Drafter [inherited],Tree [core],Token-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2606.00487,,https://anonymous.4open.science/r/TAPS-EMNLP2026-53DD
2026-06,SENSE,SENSE: Semantic Embedding Navigation with Soft-gated Evaluation for Retrieval-based Speculative Decoding,Context-Derived Nonparametric Source [core],Collection [core],Relaxed Acceptance [core],--,https://arxiv.org/abs/2606.00021,,
2026-06,JetSpec,JetSpec: Breaking the Scaling Ceiling of Speculative Decoding with Parallel Tree Drafting,Semi-Autoregressive External Drafter [core],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2606.18394,https://github.com/hao-ai-lab/JetSpec,
2026-06,Hybrid Verified Decoding,Hybrid Verified Decoding: Learning to Allocate Verification in Speculative Decoding,Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [inherited],Chain [inherited],Token-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2606.01019,,
2026-06,EfficientRollout,EfficientRollout: System-Aware Self-Speculative Decoding for RL Rollouts,Target-Path Execution [core],Chain [inherited],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2606.18967,https://github.com/furiosa-ai/EfficientRollout,
2026-06,Dynamic Exits,Experience-Driven Dynamic Exits for LLMs with Reinforcement Learning,Target-Path Execution [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2606.03113,,
2026-06,DFlare,DFlare: Scaling Up Draft Capacity for Block Diffusion Speculative Decoding,Non-Autoregressive External Drafter [core],Block [core],Token-Level Exactness [inherited],Speculation Control [inherited]; Serving Systems [inherited],https://arxiv.org/abs/2606.02091,https://github.com/Tencent/AngelSlim,
2026-06,D2SD,D^2SD: Accelerating Speculative Decoding with Dual Diffusion Draft Models,Non-Autoregressive External Drafter [core],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [core]; Serving Systems [inherited],https://arxiv.org/abs/2606.04446,,
2026-06,CADDTree,Cost-Aware Diffusion Draft Trees for Speculative Decoding,Non-Autoregressive External Drafter [inherited],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2606.01813,,
2026-06,BudgetDraft,BudgetDraft: Acceptance-Aware Multi-View Training for Sparse-KV Speculative Decoding,Autoregressive External Drafter [core],Chain [core],Token-Level Exactness [core],State Management [core]; Speculation Control [core],https://arxiv.org/abs/2606.00144,,
2026-06,AdaPLD,AdaPLD: Adaptive Retrieval and Reuse for Efficient Model-Free Speculative Decoding,Context-Derived Nonparametric Source [core],Tree [core],Distribution-Level Exactness [inherited],Speculation Control [inherited]; Serving Systems [inherited],https://arxiv.org/abs/2606.05742,,
2026-05,UniVer,UniVer: A Unified Perspective for Multi-step and Multi-draft Speculative Decoding,Proposal-State Modeling [inherited],Tree [core]; Collection [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2605.04543v1,,
2026-05,SpecKV,SpecKV: Adaptive Speculative Decoding with Compression-Aware Gamma Selection,Autoregressive External Drafter [core],Chain [core],Token-Level Exactness [core],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2605.02888,https://github.com/Amorfati123/SpecKV,
2026-05,SpecBlock,SpecBlock: Block-Iterative Speculative Decoding with Dynamic Tree Drafting,Semi-Autoregressive External Drafter [core],Tree [core],Token-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2605.07243,,
2026-05,SlimSpec,SlimSpec: Low-Rank Draft LM-Head for Accelerated Speculative Decoding,Proposal-State Modeling [inherited],Tree [inherited],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Performance Analysis [core],https://arxiv.org/abs/2605.10453,,
2026-05,SPECTRE,SPECTRE: Hybrid Ordinary-Parallel Speculative Serving for Resource-Efficient LLM Inference,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2605.08151,https://github.com/sgl-project/sglang/pull/22272,
2026-05,SPD,Speculative Pipeline Decoding: Higher-Accruacy and Zero-Bubble Speculation via Pipeline Parallelism,Proposal-State Modeling [core],Chain [inherited],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Serving Systems [core],https://arxiv.org/abs/2605.30852v1,,
2026-05,SNLP (SSD mode),SNLP: Layer-Parallel Inference via Structured Newton Corrections,Target-Path Execution [core],Chain [inherited],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Performance Analysis [core],https://arxiv.org/abs/2605.17842v2,https://github.com/phymhan/nanochat-snlp,
2026-05,RTP-LLM,RTP-LLM: High-Performance Alibaba LLM Inference Engine,Autoregressive External Drafter [inherited]; Target-Feature Readout [inherited]; Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [inherited],Chain [inherited]; Tree [inherited],Distribution-Level Exactness [inherited],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2605.29639,https://github.com/alibaba/rtp-llm,
2026-05,RAC,RAC: Request-adaptive Configuration for Efficient Speculative Decoding,Autoregressive External Drafter [core],Chain [core],Token-Level Exactness [core],Speculation Control [core]; Performance Analysis [core],https://doi.org/10.65109/sgxl4097,https://github.com/JhSheng/RAC-,
2026-05,PipeSD,PipeSD: An Efficient Cloud-Edge Collaborative Pipeline Inference Framework with Speculative Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],Serving Systems [core],https://arxiv.org/abs/2605.13319,https://github.com/Ghanyunhe/PipeSD,
2026-05,PPOW,Performance-Driven Policy Optimization for Speculative Decoding with Adaptive Windowing,Proposal-State Modeling [inherited],Tree [inherited],Distribution-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2605.14978,,
2026-05,PELM,PELM: Power Efficient On-Device LLM Inference with Speculative Decoding and Dynamic Voltage Frequency Scaling,Target-Path Execution [core],Chain [core],Relaxed Acceptance [core],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://doi.org/10.1145/3774906.3802783,,
2026-05,PARSE,Parallel Prefix Verification for Speculative Generation,Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Chain [core]; Collection [core],Relaxed Acceptance [core],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2605.04263,,
2026-05,PARD-2,PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding,Non-Autoregressive External Drafter [core],Block [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2605.08632,https://github.com/AMD-AGI/PARD,
2026-05,NanoSpec,NanoSpec: Accelerating Speculative Decoding using Minimalist In-Context Vocabularies,Proposal-State Modeling [inherited],Tree [inherited],Distribution-Level Exactness [inherited],Performance Analysis [core],https://arxiv.org/abs/2605.26444,,
2026-05,Mistletoe,Mistletoe: Stealthy Acceleration-Collapse Attacks on Speculative Decoding,Target-Feature Readout [inherited]; Proposal-State Modeling [inherited],Tree [core]; Collection [core],Relaxed Acceptance [core],Performance Analysis [core],https://arxiv.org/abs/2605.14005,,
2026-05,Latency Model,An Interpretable Latency Model for Speculative Decoding in LLM Serving,Proposal-State Modeling [inherited],Chain [inherited],--,Performance Analysis [core],https://arxiv.org/abs/2605.15051,,
2026-05,Graft,"Draft Less, Retrieve More: Hybrid Tree Construction for Speculative Decoding",Non-Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [core],Tree [core],Distribution-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2605.20104,,
2026-05,Future Validity,Future Validity is the Missing Statistic: From Impossibility to $Φ$-Estimation for Grammar-Faithful Speculative Decoding,--,Chain [core],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2605.07698,,
2026-05,FlexDraft,FlexDraft: Flexible Speculative Decoding via Attention Tuning and Bonus-Guided Calibration,Target-Path Execution [core],Block [core],Distribution-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2605.20022v1,,
2026-05,EvoSpec,EvoSpec: Evolving Speculative Decoding via Real-Time Vocabulary and Parameter Adaptation,Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [core],Tree [core]; Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2605.27390,,
2026-05,EVICT,Making Every Verified Token Count: Adaptive Verification for MoE Speculative Decoding,Proposal-State Modeling [inherited],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2605.00342,,
2026-05,Domino,Domino: Decoupling Causal Modeling from Autoregressive Drafting in Speculative Decoding,Semi-Autoregressive External Drafter [core],Block [core],Token-Level Exactness [inherited],Serving Systems [core],https://arxiv.org/abs/2605.29707,https://github.com/jianuo-huang/Domino,
2026-05,Cross-Language SD,Speculative Decoding Across Languages,Autoregressive External Drafter [core]; Context-Derived Nonparametric Source [core],Chain [core],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2605.30580v1,,
2026-05,Component-Aware SD,Component-Aware Self-Speculative Decoding in Hybrid Language Models,Target-Path Execution [core],Chain [core],Distribution-Level Exactness [core]; Token-Level Exactness [core],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2605.01106,https://github.com/hecboar/hybrid-speculative-decoding,
2026-05,CoSpec,Beyond the Target: From Imitation to Collaboration in Speculative Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Relaxed Acceptance [core],--,https://arxiv.org/abs/2605.24793,,
2026-05,Cassandra,Cassandra: Enabling Reasoning LLMs at Edge via Self-Speculative Decoding,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2605.26558,,
2026-05,CATS,CATS: Cascaded Adaptive Tree Speculation for Memory-Limited LLM Inference Acceleration,Target-Path Execution [core],Tree [core],Token-Level Exactness [inherited]; Relaxed Acceptance [inherited],Serving Systems [core],https://arxiv.org/abs/2605.11186,https://github.com/ElizaFuLan/CATS.git,
2026-05,Bastion,Bastion: Budget-Aware Speculative Decoding with Tree-structured Block Diffusion Drafting,Non-Autoregressive External Drafter [inherited],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2605.29727,https://github.com/kaist-ai-osi-lab/BASTION,
2026-05,Attention Drift,Attention Drift: What Autoregressive Speculative Decoding Models Learn,Proposal-State Modeling [core],Chain [inherited],Distribution-Level Exactness [inherited],Performance Analysis [core],https://arxiv.org/abs/2605.09992,https://github.com/Dogacel/Attention-Drift,
2026-05,Acceptance-Guided SD,Acceptance-Guided Adaptive Speculative Decoding for Efficient Large Language Model Inference,Proposal-State Modeling [inherited],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [core],https://doi.org/10.1109/ICASSP55912.2026.11463423,https://github.com/sunnywyang/ACCEPTANCE-GUIDED-ADAPTIVE-SPECULATIVE-DECODING,
2026-04,When Hidden States Drift,When Hidden States Drift: Can KV Caches Rescue Long-Range Speculative Decoding?,Proposal-State Modeling [core],Tree [inherited],Distribution-Level Exactness [inherited],State Management [core]; Performance Analysis [core],https://arxiv.org/abs/2604.26412,,
2026-04,SpecTr-GBV,SpecTr-GBV: Multi-Draft Block Verification Accelerating Speculative Decoding,Autoregressive External Drafter [inherited],Block [core]; Collection [core],Distribution-Level Exactness [core],Speculation Control [inherited],https://arxiv.org/abs/2604.25925,,
2026-04,SpecGuard,From Tokens to Steps: Verification-Aware Speculative Decoding for Efficient Multi-Step Reasoning,Autoregressive External Drafter [inherited],Collection [core],Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2604.15244,,
2026-04,SPEED-Bench,SPEED-Bench: A Unified and Diverse Benchmark for Speculative Decoding,Autoregressive External Drafter [inherited]; Target-Feature Readout [inherited]; Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [inherited],Chain [inherited]; Tree [inherited],--,Performance Analysis [core],https://arxiv.org/abs/2604.09557,,
2026-04,SMART,SMART: When is it Actually Worth Expanding a Speculative Tree?,Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2604.09731,,
2026-04,RL Rollout SD,Accelerating RL Post-Training Rollouts via System-Integrated Speculative Decoding,Target-Feature Readout [inherited]; Proposal-State Modeling [inherited],Chain [inherited]; Tree [inherited],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2604.26779,,
2026-04,RACER,RACER: Retrieval-Augmented Contextual Rapid Speculative Decoding,Context-Derived Nonparametric Source [core],Tree [core],Token-Level Exactness [inherited],Speculation Control [inherited],https://arxiv.org/abs/2604.14885,https://github.com/hkr04/RACER,
2026-04,On-Device MoE SD,Self-Speculative Decoding for On-device MoE Acceleration,Target-Path Execution [core],Chain [core],Relaxed Acceptance [core],State Management [core]; Speculation Control [core]; Serving Systems [core],https://doi.org/10.1145/3774904.3792218,,
2026-04,Multi-LoRA Edge SD,Unlocking the Edge deployment and ondevice acceleration of multi-LoRA enabled one-for-all foundational LLM,Target-Path Execution [core],Tree [core],Token-Level Exactness [inherited],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2604.18655v2,,
2026-04,FPGA Tree SD,Boosting the Performance of Tree-Based Speculative Decoding of LLMs on FPGAs,Proposal-State Modeling [inherited],Tree [core],Token-Level Exactness [core],State Management [core]; Serving Systems [core],https://doi.org/10.23919/date69613.2026.11539704,,
2026-04,ELMoE-3D,ELMoE-3D: Leveraging Intrinsic Elasticity of MoE for Hybrid-Bonding-Enabled Self-Speculative Decoding in On-Premises Serving,Target-Path Execution [core],Chain [inherited],Distribution-Level Exactness [inherited],Serving Systems [core],https://arxiv.org/abs/2604.14626,,
2026-04,ECHO,ECHO: Elastic Speculative Decoding with Sparse Gating for High-Concurrency Scenarios,Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [inherited],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2604.09603,,
2026-04,Dynamic Multi-Path SD,Optimizing speculative decoding via dynamic multi-path and dual-stream networks,Autoregressive External Drafter [core],Collection [core],Relaxed Acceptance [core],Speculation Control [core],https://doi.org/10.1016/j.neunet.2026.109014,,
2026-04,DIVERSED,DIVERSED: Relaxed Speculative Decoding via Dynamic Ensemble Verification,Autoregressive External Drafter [core],Chain [core],Relaxed Acceptance [core],--,https://arxiv.org/abs/2604.07622,https://github.com/comeusr/diversed,
2026-04,DDTree,Accelerating Speculative Decoding with Block Diffusion Draft Trees,Non-Autoregressive External Drafter [inherited],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [inherited]; Serving Systems [inherited],https://arxiv.org/abs/2604.12989,https://github.com/liranringel/ddtree,
2026-04,Cross-Family,Cross-Family Speculative Decoding for Polish Language Models on Apple~Silicon: An Empirical Evaluation of Bielik~11B with UAG-Extended MLX-LM,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2604.16368,https://github.com/krzysiekfonal/mlx-lm/tree/feature/slem-with-context-aware,
2026-04,ConfLayers,ConfLayers: Adaptive Confidence-based Layer Skipping for Self-Speculative Decoding,Target-Path Execution [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2604.14612,https://github.com/WA225/ConfLayers,
2026-04,Commodity-GPU SD,Memory-Efficient Speculative Decoding for Large Language Model Inference on Commodity GPUs,Autoregressive External Drafter [core],Chain [inherited],Token-Level Exactness [inherited],Serving Systems [core]; Performance Analysis [core],https://doi.org/10.1109/csnt69054.2026.11502509,,
2026-04,Calibrated,Calibrated Speculative Decoding: Frequency-Guided Candidate Selection for Efficient Inference,Autoregressive External Drafter [core]; Context-Derived Nonparametric Source [core],Chain [core],Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2604.13634,,
2026-04,Cactus,Cactus: Accelerating Auto-Regressive Decoding with Constrained Acceptance Speculative Sampling,Autoregressive External Drafter [inherited],Chain [inherited],Bounded Approximation [core],Speculation Control [inherited],https://arxiv.org/abs/2604.04987,https://github.com/MANGA-UOFA/Cactus,
2026-04,Area-,Area- and Utilization-Efficient LLM Accelerator With Fused Speculative Decoding for Edge-Side Inference,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],State Management [core]; Serving Systems [core],https://doi.org/10.1109/tvlsi.2026.3659893,,
2026-04,Adaptive Hybrid SD,Adaptive hybrid speculative decoding for accelerating large language model inference,Autoregressive External Drafter [core],Chain [core]; Block [core],Relaxed Acceptance [core],State Management [core]; Speculation Control [core],https://doi.org/10.1016/j.neucom.2026.133760,,
2026-04,AdaServe,AdaServe: Accelerating Multi-SLO LLM Serving with SLO-Customized Speculative Decoding,Autoregressive External Drafter [inherited],Tree [core],Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://dl.acm.org/doi/10.1145/3767295.3769315,https://github.com/zikun-li/AdaServe-Artifact-Evaluation,
2026-04,Acceptance,Acceptance Dynamics Across Cognitive Domains in Speculative Decoding,Autoregressive External Drafter [core],Tree [core],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2604.14682,https://github.com/saifmb0/tree-acceptance,
2026-04,AHASD,AHASD: Asynchronous Heterogeneous Architecture for LLM Adaptive Drafting Speculative Decoding on Mobile Devices,Autoregressive External Drafter [inherited],Chain [inherited],--,Serving Systems [core],https://arxiv.org/abs/2604.25326,,
2026-04,A-IO (SD/PLD study),A-IO: Adaptive Inference Orchestration for Memory-Bound NPUs,Context-Derived Nonparametric Source [inherited]; Autoregressive External Drafter [inherited],Chain [inherited],--,Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2604.09752v2,,
2026-03,VocabTrim,Balancing Coverage and Draft Latency in Vocabulary Trimming for Faster Speculative Decoding,Proposal-State Modeling [inherited],Tree [inherited],Distribution-Level Exactness [inherited],Performance Analysis [core],https://arxiv.org/abs/2603.05210,https://github.com/Ofir408/balanced-coverage-latency-spec-decoding,
2026-03,TAPS-Distribution,TAPS: Task Aware Proposal Distributions for Speculative Sampling,Proposal-State Modeling [inherited],Collection [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [inherited],https://arxiv.org/abs/2603.27027v1,https://github.com/Moe-Zbeeb/TAPS,https://huggingface.co/collections/zbeeb/taps; https://huggingface.co/datasets/zbeeb/TAPS-Datasets
2026-03,Speculative-Speculative,Speculative Speculative Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],State Management [core]; Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2603.03251,https://github.com/tanishqkumar/ssd,
2026-03,SpecForge,SpecForge: A Flexible and Efficient Open-Source Training Framework for Speculative Decoding,Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2603.18567,https://github.com/sgl-project/SpecForge,
2026-03,Re-SpS,Re-SpS: A Reinforcement Learning Approach to Speculative Sampling,Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [core],--,https://doi.org/10.1609/aaai.v40i39.40625,https://github.com/wmd3i/ReSpS.git,
2026-03,Quasar,Quasar: Quantized Self-Speculative Acceleration for Rapid Inference via Memory-Efficient Verification,Context-Derived Nonparametric Source [inherited],Chain [core],Relaxed Acceptance [core],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2603.01399,https://github.com/Tom-HG/Quasar,
2026-03,MineDraft,MineDraft: A Framework for Batch Parallel Speculative Decoding,Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Chain [inherited],Distribution-Level Exactness [inherited],State Management [core]; Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2603.18016,,
2026-03,Lookahead Gate,Look Before You Leap: A Lookahead Reasoning Quality Gate for Speculative Decoding,Target-Path Execution [inherited],Collection [core],Relaxed Acceptance [core],Speculation Control [core],https://aclanthology.org/2026.eacl-long.367/,,
2026-03,Learning to Draft,Learning to Draft: Adaptive Speculative Decoding with Reinforcement Learning,Proposal-State Modeling [inherited],Tree [core],Token-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2603.01639,https://github.com/zhzihao/Learning-to-Draft,
2026-03,Empirical Study,An Empirical Study of Speculative Decoding for Small Language Models,Autoregressive External Drafter [inherited]; Target-Path Execution [inherited]; Target-Feature Readout [inherited]; Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [inherited],Chain [inherited]; Tree [inherited]; Collection [inherited],--,Performance Analysis [core],https://aclanthology.org/2026.eacl-long.255/,,
2026-03,EAGLE-Pangu,EAGLE-Pangu: Accelerator-Safe Tree Speculative Decoding on Ascend NPUs,Proposal-State Modeling [inherited],Tree [inherited],Token-Level Exactness [inherited],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2603.08088,,
2026-03,Distributed Adaptive SD,Distributed Adaptive Speculative Decoding: Accelerating Large Language Model Inference With Context-Aware Draft Selection,Autoregressive External Drafter [core],Chain [inherited],Token-Level Exactness [inherited],State Management [core]; Speculation Control [core]; Serving Systems [core],https://doi.org/10.1109/isdfs69419.2026.11459120,,
2026-03,DFVG,DFVG: A Heterogeneous Architecture for Speculative Decoding with Draft-on-FPGA and Verify-on-GPU,Autoregressive External Drafter [inherited],Tree [core],--,Speculation Control [core]; Serving Systems [core],https://doi.org/10.1145/3779212.3790153,https://github.com/ShaoqiangLu/DFVG,https://shaoqianglu.github.io/
2026-03,ConFu,ConFu: Contemplate the Future for Better Speculative Sampling,Proposal-State Modeling [core],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Serving Systems [inherited],https://arxiv.org/abs/2603.08899,,
2026-03,AdaSpec-Multilingual,AdaSpec: Adaptive Multilingual Speculative Decoding with Self-Synthesized Language-Aware Training and Vocabulary Simplification,Autoregressive External Drafter [core]; Context-Derived Nonparametric Source [core],Chain [core],Token-Level Exactness [core],--,https://ojs.aaai.org/index.php/AAAI/article/view/40307,,
2026-02,Watermark Trade-off,Improving the Trade-off Between Watermark Strength and Speculative Sampling Efficiency for Language Models,Autoregressive External Drafter [core],Chain [inherited],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2602.01428,https://github.com/hwq0726/watermark-tradeoff,
2026-02,Vegas,Vegas: Self-Speculative Decoding with Verification-Guided Sparse Attention,Target-Path Execution [core],Chain [core],Distribution-Level Exactness [core],State Management [core]; Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2602.07223,https://github.com/platformxlab/vegas,
2026-02,VSD,Variational Speculative Decoding: Rethinking Draft Training from Token Likelihood to Sequence Acceptance,Proposal-State Modeling [inherited],Tree [inherited],Distribution-Level Exactness [inherited],Speculation Control [inherited],https://arxiv.org/abs/2602.05774,,
2026-02,TransKV,Transactional KV Caching for Speculative Decoding under Paged KV Memory,--,Chain [inherited],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],State Management [core]; Serving Systems [core],https://doi.org/10.36227/techrxiv.177101038.80960856/v1,,
2026-02,Split Lookahead,Privacy-Aware Split Inference with Speculative Decoding for Large Language Models over Wide-Area Networks,Context-Derived Nonparametric Source [core],Chain [core],Token-Level Exactness [inherited],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2602.16760,,
2026-02,SpecVocab,Speculative Decoding with a Speculative Vocabulary,Proposal-State Modeling [inherited],Tree [inherited],Distribution-Level Exactness [inherited],Performance Analysis [core],https://arxiv.org/abs/2602.13836v1,,
2026-02,SemanticSpec,Beyond Tokens: Semantic-Aware Speculative Decoding for Efficient Inference by Probing Internal States,Autoregressive External Drafter [inherited],Chain [core],Relaxed Acceptance [core],--,https://arxiv.org/abs/2602.03708,,
2026-02,SDFP,SDFP: Speculative Decoding with FIT-Pruned Models for Training-Free and Plug-and-Play LLM Acceleration,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2602.05499,,
2026-02,PRISM,PRISM: Parametrically Refactoring Inference for Speculative Sampling Draft Models,Proposal-State Modeling [core],Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2602.01762,,
2026-02,P-EAGLE,P-EAGLE: Parallel-Drafting EAGLE with Scalable Training,Proposal-State Modeling [core],Block [core],Distribution-Level Exactness [inherited],Serving Systems [core],https://arxiv.org/abs/2602.01469v1,,
2026-02,KnapSpec,KnapSpec: Self-Speculative Decoding via Adaptive Layer Selection as a Knapsack Problem,Target-Path Execution [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2602.20217,https://github.com/kaist-flexml-lab/knapspec,
2026-02,GBV,Greedy Multi-Path Block Verification for Faster Decoding in Speculative Sampling,Autoregressive External Drafter [inherited],Block [core]; Collection [core],Distribution-Level Exactness [core],Speculation Control [inherited],https://arxiv.org/abs/2602.16961,,
2026-02,Energy Benchmarking,Benchmarking the Energy Savings with Speculative Decoding Strategies,Autoregressive External Drafter [inherited]; Target-Feature Readout [inherited]; Proposal-State Modeling [inherited],Chain [inherited]; Tree [inherited],--,Performance Analysis [core],https://arxiv.org/abs/2602.09113,,
2026-02,DFlash,DFlash: Block Diffusion for Flash Speculative Decoding,Non-Autoregressive External Drafter [core],Block [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [inherited]; Serving Systems [inherited],https://arxiv.org/abs/2602.06036,https://github.com/z-lab/dflash,
2026-02,Compiler-Assisted SD,Compiler-Assisted Speculative Sampling for Accelerated LLM Inference on Heterogeneous Edge Devices,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2602.08060,,
2026-01,HSD,Overcoming Joint Intractability with Lossless Hierarchical Speculative Decoding,Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Block [core]; Collection [inherited],Distribution-Level Exactness [core],Speculation Control [inherited],https://arxiv.org/abs/2601.05724,https://github.com/ZhouYuxuanYX/Hierarchical-Speculative-Decoding,
2026-01,FlexSpec,FlexSpec: Frozen Drafts Meet Evolving Targets in Edge-Cloud Collaborative LLM Speculative Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Token-Level Exactness [inherited],Serving Systems [core],https://doi.org/10.1109/tmc.2026.3683574,,
2026-01,DART,DART: Diffusion-Inspired Speculative Decoding for Fast LLM Inference,Non-Autoregressive External Drafter [core],Tree [core],Token-Level Exactness [inherited],Speculation Control [inherited],https://arxiv.org/abs/2601.19278,https://github.com/fvliang/DART,
2025-12,T-pro 2.0 (EAGLE),T-pro 2.0: An Efficient Russian Hybrid-Reasoning Model and Playground,Proposal-State Modeling [inherited],Tree [inherited],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [inherited]; Serving Systems [core]; Performance Analysis [core],https://aclanthology.org/2026.eacl-demo.22/,,https://huggingface.co/collections/t-tech/t-pro-20; https://huggingface.co/t-tech/T-pro-it-2.0-eagle
2025-12,Speed-of-Light,Speculative Decoding Speed-of-Light: Optimal Lower Bounds via Branching Random Walks,--,Tree [core],Token-Level Exactness [inherited],Performance Analysis [core],https://doi.org/10.18653/v1/2026.eacl-long.301,,
2025-12,SpecPV,SpecPV: Improving Self-Speculative Decoding for Long-Context Generation via Partial Verification,Proposal-State Modeling [inherited],Tree [inherited],Relaxed Acceptance [core],State Management [core]; Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2512.02337v1,https://github.com/TanZhendong/SpecPV,
2025-12,Sparse Self-SD,Accelerating Large-Scale Reasoning Model Inference with Sparse Self-Speculative Decoding,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],State Management [core]; Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2512.01278v1,https://github.com/sspec-project/SparseSpec,
2025-12,RADAR,RADAR: Accelerating Large Language Model Inference With RL-Based Dynamic Draft Trees,Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2512.14069v1,https://github.com/minaduki-sora/RADAR,
2025-12,Performance or Illusion,Speculative Decoding: Performance or Illusion?,Autoregressive External Drafter [inherited]; Target-Feature Readout [inherited]; Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [inherited],Chain [inherited]; Tree [inherited],--,Performance Analysis [core],https://proceedings.mlsys.org/paper_files/paper/2026/file/554e056fe2b6d9fd27ffcd3367ae1267-Paper-Conference.pdf,,https://zenodo.org/records/19365903; https://arxiv.org/abs/2601.11580v2
2025-12,Parallel Token Prediction,Parallel Token Prediction for Language Models,Non-Autoregressive External Drafter [core],Block [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2512.21323v2,https://github.com/mandt-lab/ptp,
2025-12,Nightjar,Nightjar: Dynamic adaptive speculative decoding for large language models serving,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://doi.org/10.1016/j.sysarc.2026.103889,,
2025-12,FailFast,"Fail Fast, Win Big: Rethinking the Drafting Strategy in Speculative Decoding via Diffusion LLMs",Non-Autoregressive External Drafter [inherited],Chain [inherited],Token-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2512.20573,https://github.com/ruipeterpan/failfast,
2025-12,Device-Server SD,Device-Server Collaborative Speculative Decoding for Real-Time LLM Streaming,Autoregressive External Drafter [inherited],Chain [inherited],Relaxed Acceptance [core],Serving Systems [core],https://doi.org/10.1109/globecom59602.2025.11432593,,
2025-12,Aggressive ASD,Aggressive Speculative Decoding for Efficient Chain-of-Thought Reasoning,Autoregressive External Drafter [inherited],Chain [inherited],Bounded Approximation [core]; Relaxed Acceptance [core],Speculation Control [core],https://www.sciopen.com/article/10.26599/TST.2026.9010003,,
2025-12,AdaSD,AdaSD: Adaptive Speculative Decoding for Efficient Language Model Inference,Autoregressive External Drafter [core],Chain [core],Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2512.11280,,
2025-12,ASDT,ASDT: Adaptive Speculative Decoding Tree Based on Token Acceptance Feedback,Autoregressive External Drafter [inherited],Tree [core],Distribution-Level Exactness [core],Speculation Control [core],https://doi.org/10.1109/aibdf67964.2025.11440682,,
2025-11,WiP,WiP: Efficient Speculative Decoding for AI PCs via Hierarchical N-Gram Retrieval,Context-Derived Nonparametric Source [core],Tree [core],Token-Level Exactness [core],Serving Systems [core],https://doi.org/10.1145/3737902.3768352,,
2025-11,SpecFormer,Scaling LLM Speculative Decoding: Non-Autoregressive Forecasting in Large-Batch Scenarios,Proposal-State Modeling [core],Block [core],Token-Level Exactness [inherited],Performance Analysis [core],https://doi.org/10.1609/aaai.v40i39.40576,https://github.com/ShiLuohe/SpecFormer,
2025-11,SpecDiff-2,SpecDiff-2: Scaling Diffusion Drafter Alignment For Faster Speculative Decoding,Non-Autoregressive External Drafter [core],Block [core]; Collection [core],Token-Level Exactness [inherited],Speculation Control [core],https://arxiv.org/abs/2511.00606,,
2025-11,SDSquare,Steering Pretrained Drafters During Speculative Decoding,Autoregressive External Drafter [core],Chain [inherited],Distribution-Level Exactness [inherited],Speculation Control [inherited],https://doi.org/10.1609/aaai.v40i36.40255,https://github.com/ETH-DISCO/SD-square,
2025-11,Next-Latent,Next-Latent Prediction Transformers Learn Compact World Models,Proposal-State Modeling [core],Chain [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2511.05963v4,https://github.com/JaydenTeoh/NextLat,
2025-11,N-Gram Trie SD,Faster In-Context Learning for LLMs via N-Gram Trie Speculative Decoding,Context-Derived Nonparametric Source [core],Tree [core],Token-Level Exactness [core],--,https://aclanthology.org/2025.emnlp-main.911/,https://github.com/mrlife219/Ngram-Trie,
2025-11,Global Resolution,Global Resolution: Optimal Multi-Draft Speculative Sampling via Convex Minimization,Autoregressive External Drafter [inherited],Tree [inherited]; Collection [core],Bounded Approximation [core],--,https://arxiv.org/abs/2511.15898v1,,
2025-11,FractalLLM,FractalLLM: Lossless Self-Speculative Decoding with Layer Embedded Self-Compression,Proposal-State Modeling [core],Chain [core],Token-Level Exactness [core],--,https://aclanthology.org/2025.findings-emnlp.1286/,https://github.com/YEonleo/FractalLLM,
2025-11,FLy,Training-Free Loosely Speculative Decoding: Accepting Semantically Correct Drafts Beyond Exact Match,Autoregressive External Drafter [inherited]; Context-Derived Nonparametric Source [inherited],Chain [inherited],Relaxed Acceptance [core],--,https://doi.org/10.48550/arxiv.2511.22972,,
2025-11,Collective Adaptive SD,Towards Efficient LLM Inference via Collective and Adaptive Speculative Decoding,Autoregressive External Drafter [inherited],Collection [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.1145/3712285.3759834,,
2025-11,Cacheback,Cacheback: Speculative Decoding With Nothing But Cache,Context-Derived Nonparametric Source [core],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Serving Systems [core],https://aclanthology.org/2025.emnlp-main.1581/,https://github.com/zyma98/Spec-Bench/tree/cacheback,
2025-10,TokenTiming,TokenTiming: A Dynamic Alignment Method for Universal Speculative Decoding Model Pairs,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2510.15545v4,,
2025-10,ReSpec,ReSpec: Towards Optimizing Speculative Decoding in Reinforcement Learning Systems,Autoregressive External Drafter [core],Tree [core],Distribution-Level Exactness [core],State Management [core]; Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2510.26475v1,,
2025-10,Polybasic,Polybasic Speculative Decoding Through a Theoretical Perspective,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2510.26527v1,,
2025-10,Mirror SD,Mirror Speculative Decoding: Breaking the Serial Barrier in LLM Inference,Autoregressive External Drafter [core]; Target-Path Execution [core],Tree [core],Token-Level Exactness [core],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2510.13161v2,,
2025-10,HiSpec,HiSpec: Hierarchical Speculative Decoding for LLMs,Target-Path Execution [core]; Target-Feature Readout [inherited],Chain [core],Token-Level Exactness [core]; Distribution-Level Exactness [core],State Management [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2510.01336v2,,
2025-10,DynaSpec,DynaSpec: Context-aware Dynamic Speculative Sampling for Large-Vocabulary Language Models,Proposal-State Modeling [inherited],Tree [inherited],Distribution-Level Exactness [inherited],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2510.13847v3,,
2025-10,Draft-Verify-Improve,"Draft, Verify, and Improve: Toward Training-Aware Speculative Decoding",Target-Path Execution [core]; Target-Feature Readout [core],Chain [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2510.05421,,
2025-10,Disparate Impacts,The Disparate Impacts of Speculative Decoding,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2510.02128v1,,
2025-10,DiffuSpec,DiffuSpec: Unlocking Diffusion Language Models for Speculative Decoding,Non-Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2510.02358,,
2025-10,Comm-Efficient SD,Communication-Efficient Speculative Decoding for Large Language Models Inference,Autoregressive External Drafter [core],Chain [inherited],Relaxed Acceptance [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.1109/icct67417.2025.11374072,,
2025-10,CAS-Spec,CAS-Spec: Cascade Adaptive Self-Speculative Decoding for On-the-Fly Lossless Inference Acceleration of LLMs,Target-Path Execution [core]; Context-Derived Nonparametric Source [inherited],Tree [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2510.26843v1,,
2025-10,BatchSpec,Batch Speculative Decoding Done Right,Autoregressive External Drafter [inherited],Chain [inherited],Token-Level Exactness [core],State Management [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2510.22876v3,https://github.com/eBay/spec_dec,
2025-10,AdaSPEC,AdaSPEC: Selective Knowledge Distillation for Efficient Speculative Decoders,Autoregressive External Drafter [core],Chain [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2510.19779,https://github.com/yuezhouhu/adaspec,
2025-10,3-Model,3-Model Speculative Decoding,Autoregressive External Drafter [core],Chain [core],Relaxed Acceptance [core],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2510.12966v1,,
2025-09,SpecMamba,SpecMamba: Accelerating Mamba Inference on FPGA with Speculative Decoding,Autoregressive External Drafter [core],Tree [core],Distribution-Level Exactness [core],Serving Systems [core],https://doi.org/10.1109/iccad66269.2025.11240945,,
2025-09,SSBD,Self-Speculative Biased Decoding for Faster Re-Translation,Context-Derived Nonparametric Source [core],Chain [core],Token-Level Exactness [inherited]; Relaxed Acceptance [core],State Management [core],https://arxiv.org/abs/2509.21740,,
2025-09,SPEC-RL,SPEC-RL: Accelerating On-Policy Reinforcement Learning with Speculative Rollouts,Context-Derived Nonparametric Source [core],Chain [inherited],Relaxed Acceptance [core],State Management [core]; Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2509.23232v3,,
2025-09,SN40L SD,Speculative Decoding on the SN40L Reconfigurable Dataflow Unit,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Serving Systems [core],https://doi.org/10.1109/mm.2025.3592570,,
2025-09,Pipeline Early-Exit,Pipeline Parallelism is All You Need for Optimized Early-Exit Based Self-Speculative Decoding,Target-Path Execution [core]; Target-Feature Readout [core],Chain [inherited],Token-Level Exactness [core],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2509.19368v1,,
2025-09,FastMTP,FastMTP: Accelerating LLM Inference with Enhanced Multi-Token Prediction,Proposal-State Modeling [core],Chain [inherited],Token-Level Exactness [inherited],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2509.18362v1,https://github.com/Tencent-BAC/FastMTP,
2025-09,DSDE,DSDE: Dynamic Speculative Decoding with KLD Stability for Real-World Serving,Autoregressive External Drafter [inherited],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2509.01083,,
2025-09,Collaborative SD,Communication-Efficient Collaborative LLM Inference via Distributed Speculative Decoding,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2509.04576,,
2025-08,SuperSpec,SuperSpec: Enhanced Verification and Sampling for End-to-End LLM Speculative Decoding,Autoregressive External Drafter [inherited],Collection [core]; Tree [core],Distribution-Level Exactness [core],State Management [core]; Speculation Control [core]; Serving Systems [core],https://doi.org/10.1109/hpcc67675.2025.00064,,
2025-08,SSS,Reward-Shifted Speculative Sampling Is An Efficient Test-Time Weak-to-Strong Aligner,Autoregressive External Drafter [inherited],Chain [inherited],Relaxed Acceptance [core],--,https://aclanthology.org/2025.emnlp-main.578/,https://github.com/lblaoke/CARDS,
2025-08,SSD (Safety-Aware),Speculative Safety-Aware Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Relaxed Acceptance [core],Speculation Control [core],https://aclanthology.org/2025.emnlp-main.648/,https://github.com/k-k1w-w1x-x/Speculative-Safety-Aware-Decoding,https://arxiv.org/abs/2508.17739
2025-08,ResDecode,ResDecode: Accelerating Large Language Models Inference via Residual Decoding Heads,Proposal-State Modeling [core],Tree [core],Distribution-Level Exactness [core],--,https://doi.org/10.26599/bdma.2024.9020074,https://github.com/ZeroNLP/ResDecode,
2025-08,KVPSN (text-to-text mode),Novel Parasitic Dual-Scale Modeling for Efficient and Accurate Multilingual Speech Translation,Autoregressive External Drafter [core],Chain [inherited],Token-Level Exactness [inherited]; Relaxed Acceptance [core],State Management [core],https://arxiv.org/abs/2508.11189v1,,
2025-08,Confidence-Modulated,Confidence-Modulated Speculative Decoding for Large Language Models,Non-Autoregressive External Drafter [inherited],Block [inherited],Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2508.15371v1,,
2025-08,CARD,CARD: A Cache-Assisted Parallel Speculative Decoding Framework via Query-and-Correct Paradigm for Accelerating LLM Inference,Autoregressive External Drafter [core]; Context-Derived Nonparametric Source [core],Chain [inherited],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2508.04462v2,https://github.com/hunzhizi/CARD,
2025-07,HVT,Hierarchical Verification of Speculative Beams for Accelerating LLM Inference,Autoregressive External Drafter [inherited],Tree [core],--,Speculation Control [core],https://arxiv.org/abs/2508.03726v1,,
2025-07,EDD/PCT,Faster Speculative Decoding via Effective Draft Decoder with Pruned Candidate Tree,Autoregressive External Drafter [core],Tree [core],Token-Level Exactness [core],Speculation Control [core],https://aclanthology.org/2025.acl-long.486/,,
2025-07,Disaggregated,Disaggregated Speculative Decoding for Carbon-Efficient LLM Serving,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],Serving Systems [core]; Performance Analysis [core],https://doi.org/10.1109/lca.2025.3630094,,
2025-07,DSSD,DSSD: Efficient Edge-Device LLM Deployment and Collaborative Inference via Distributed Split Speculative Decoding,Autoregressive External Drafter [core],Chain [inherited],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2507.12000v2,https://github.com/JasonNing96/DSSD-Efficient-Edge-Computing,
2025-06,VOCABTRIM,VOCABTRIM: Vocabulary Pruning for Efficient Speculative Decoding in LLMs,Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Chain [core]; Tree [core],Distribution-Level Exactness [core],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2506.22694v2,,
2025-06,Utility-Driven,Utility-Driven Speculative Decoding for Mixture-of-Experts,Context-Derived Nonparametric Source [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Performance Analysis [core],https://doi.org/10.48550/arxiv.2506.20675,,
2025-06,SwiftSpec,SwiftSpec: Ultra-Low Latency LLM Decoding by Scaling Asynchronous Speculative Decoding,Autoregressive External Drafter [inherited],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2506.11309,,
2025-06,SpecMemo,SpecMemo: Speculative Decoding is in Your Pocket,Target-Feature Readout [inherited],Tree [core],Token-Level Exactness [core],State Management [core]; Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2506.01986v1,,
2025-06,SpecEE,SpecEE: Accelerating Large Language Model Inference with Speculative Early Exiting,Proposal-State Modeling [inherited],Tree [inherited],Relaxed Acceptance [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.1145/3695053.3730996,https://github.com/infinigence/SpecEE,
2025-06,SpecBranch,SpecBranch: Speculative Decoding via Hybrid Drafting and Rollback-Aware Branch Parallelism,Autoregressive External Drafter [inherited],Tree [core],Token-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.48550/arxiv.2506.01979,,
2025-06,S4C,S4C: Speculative Sampling with Syntactic and Semantic Coherence for Efficient Inference of Large Language Models,Proposal-State Modeling [core],Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2506.14158v1,,
2025-06,POSS,POSS: Position Specialist Generates Better Draft for Speculative Decoding,Proposal-State Modeling [core],Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2506.03566v1,https://github.com/shrango/PosS,
2025-06,Model-Free Test-Time Scaling,Accelerated Test-Time Scaling with Model-Free Speculative Sampling,Context-Derived Nonparametric Source [core],Tree [core],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2506.04708v3,,
2025-06,Mamba,Mamba Drafters for Speculative Decoding,Autoregressive External Drafter [core],Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2506.01206,,
2025-06,Lookahead Reasoning,Scaling Speculative Decoding with Lookahead Reasoning,Autoregressive External Drafter [inherited],Chain [core],Relaxed Acceptance [core],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2506.19830,https://github.com/hao-ai-lab/LookaheadReasoning,
2025-06,List-Level,List-Level Distribution Coupling with Applications to Speculative Decoding and Lossy Compression,Autoregressive External Drafter [inherited],Collection [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2506.05632,,
2025-06,GSI,Guided Speculative Inference for Efficient Test-Time Alignment of LLMs,Autoregressive External Drafter [inherited],Collection [core],Bounded Approximation [core]; Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2506.04118,https://github.com/j-geuter/GSI,
2025-06,Dynamic-K SD,Dynamic K Mechanism for Accelerating Speculative Decoding in vLLM,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.1109/icecai66283.2025.11170671,,
2025-06,3D-TokSIM,3D-TokSIM: Stacking 3D Memory with Token-Stationary Compute-in-Memory for Speculative LLM Inference,--,Chain [inherited],Token-Level Exactness [inherited],Serving Systems [core]; Performance Analysis [core],https://doi.org/10.1109/dac63849.2025.11132883,,
2025-05,Traversal,Traversal Verification for Speculative Tree Decoding,Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [core],--,https://doi.org/10.48550/arxiv.2505.12398,,
2025-05,SpecRouter,SpecRouter: Adaptive Routing for Multi-Level Speculative Decoding in Large Language Models,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],State Management [core]; Speculation Control [core],https://arxiv.org/abs/2505.07680v1,,
2025-05,Quantization Compat.,Speculative Decoding Meets Quantization: Compatibility Evaluation and Hierarchical Framework Design,Autoregressive External Drafter [core]; Proposal-State Modeling [inherited],Tree [core]; Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2505.22179v2,,
2025-05,PipeSpec,PipeSpec: Breaking Stage Dependencies in Hierarchical LLM Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Token-Level Exactness [inherited],Serving Systems [core],https://aclanthology.org/2025.findings-acl.669/,https://github.com/BradMcDanel/PipeSpec,
2025-05,MTP Curriculum,Pre-Training Curriculum for Multi-Token Prediction in Language Models,Target-Feature Readout [core],Block [inherited],Token-Level Exactness [inherited],--,https://arxiv.org/abs/2505.22757v1,https://github.com/aynetdia/mtp_curriculum,
2025-05,LAPS-SD,Semi-Clairvoyant Scheduling of Speculative Decoding Requests to Minimize LLM Inference Latency,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://doi.org/10.24963/ijcai.2025/951,,
2025-05,KNN-SSD,KNN-SSD: Enabling Dynamic Self-Speculative Decoding via Nearest Neighbor Layer Set Optimization,Target-Path Execution [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://doi.org/10.18653/v1/2026.findings-eacl.31,https://github.com/mbsong/KNN-SSD,
2025-05,Ghidorah,Ghidorah: Fast LLM Inference on Edge with Speculative Decoding and Hetero-Core Parallelism,Target-Feature Readout [inherited],Tree [core],Token-Level Exactness [inherited],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2505.23219,,
2025-05,CLaSp,CLaSp: In-Context Layer Skip for Self-Speculative Decoding,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2505.24196v1,,
2025-05,BanditSpec,BanditSpec: Adaptive Speculative Decoding via Bandit Algorithms,Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [inherited],Chain [core]; Tree [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2505.15141v2,https://github.com/sail-sg/BanditSpec,
2025-05,Automatic,Automatic Task Detection and Heterogeneous LLM Speculative Decoding,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2505.08600v1,,
2025-05,Alignment-Augmented,Alignment-Augmented Speculative Decoding with Alignment Sampling and Conditional Verification,Context-Derived Nonparametric Source [core],Chain [core],Relaxed Acceptance [core],--,https://arxiv.org/abs/2505.13204v2,,
2025-04,SpecReason,SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning,Autoregressive External Drafter [inherited],Chain [core],Relaxed Acceptance [core],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2504.07891,https://github.com/ruipeterpan/specreason,
2025-04,SPIRe,SPIRe: Boosting LLM Inference Throughput with Speculative Decoding,Autoregressive External Drafter [core],Chain [inherited],Distribution-Level Exactness [core],State Management [core]; Performance Analysis [core],https://arxiv.org/abs/2504.06419v1,,
2025-04,SD$^2$,SD$^2$: Self-Distilled Sparse Drafters,Autoregressive External Drafter [core],Chain [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2504.08838v2,,
2025-04,PARD,PARD: Accelerating LLM Inference with Low-Cost PARallel Draft Model Adaptation,Non-Autoregressive External Drafter [core],Block [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2504.18583v4,https://github.com/AMD-AIG-AIMA/PARD,
2025-04,Hierarchical,Hierarchical Speculative Decoding with Dynamic Window,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://aclanthology.org/2025.findings-naacl.462/,,
2025-04,Exponential Races,Speculative Sampling via Exponential Races,Autoregressive External Drafter [core],Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2504.15475v1,,
2025-04,EdgeLLM,EdgeLLM: Fast On-Device LLM Inference With Speculative Decoding,Autoregressive External Drafter [core],Tree [core],Token-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.1109/tmc.2024.3513457,,
2025-04,DEL,DEL: Context-Aware Dynamic Exit Layer for Efficient Self-Speculative Decoding,Target-Path Execution [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2504.05598v2,https://github.com/hoenza/DEL,
2025-04,ASSD,Reviving Any-Subset Autoregressive Models with Principled Parallel Sampling and Speculative Decoding,Target-Path Execution [core],Block [core],Distribution-Level Exactness [core],Speculation Control [inherited],https://arxiv.org/abs/2504.20456,https://github.com/gabeguo/any-order-speculative-decoding,
2025-03,vLLM SpecInfer,Speculative Inference with vLLM: Optimizing Heterogeneous Computing in Training-Inference Integrated Environments,--,Chain [inherited],--,Serving Systems [core],https://doi.org/10.1109/iscait64916.2025.11010601,,
2025-03,RASD,RASD: Retrieval-Augmented Speculative Decoding,Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [core],Tree [core],Distribution-Level Exactness [core],--,https://aclanthology.org/2025.findings-acl.320/,,
2025-03,Multi-Sample SD,Speculative Decoding for Multi-Sample Inference,Context-Derived Nonparametric Source [core],Collection [core],Distribution-Level Exactness [core],--,https://doi.org/10.18653/v1/2025.findings-emnlp.668,,
2025-03,ML-SpecQD,ML-SpecQD: Multi-Level Speculative Decoding with Quantized Drafts,Autoregressive External Drafter [core],Chain [inherited],Token-Level Exactness [inherited],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2503.13565v1,,
2025-03,EAGLE-3,EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test,Proposal-State Modeling [core],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [inherited]; Serving Systems [inherited],https://openreview.net/forum?id=4exx1hUffq,https://github.com/SafeAILab/EAGLE,
2025-03,Dynamic Adjustment,Improving Speculative Decoding with Dynamic Adjustment and Probability Smoothing,Autoregressive External Drafter [core],Collection [core],Distribution-Level Exactness [core],--,https://doi.org/10.1109/icaace65325.2025.11019425,,
2025-03,DuoDecoding,DuoDecoding: Hardware-aware Heterogeneous Speculative Decoding with Dynamic Multi-Sequence Drafting,Autoregressive External Drafter [core],Collection [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2503.00784v1,https://github.com/KaiLv69/DuoDecoding,
2025-03,Domain Draft Models,Training Domain Draft Models for Speculative Decoding: Best Practices and Insights,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2503.07807v2,,
2025-03,AdaSpec-SLO,"AdaSpec: Adaptive Speculative Decoding for Fast, SLO-Aware Large Language Model Serving",Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],Speculation Control [core],https://doi.org/10.1145/3772052.3772239,https://github.com/cerebellumking/AdaSpec,
2025-02,TETRIS,TETRIS: Optimal Draft Token Selection for Batch Speculative Decoding,Autoregressive External Drafter [inherited]; Target-Feature Readout [inherited],Collection [core],Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2502.15197v2,,
2025-02,QuantSpec,QuantSpec: Self-Speculative Decoding with Hierarchical Quantized KV Cache,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2502.10424v1,,
2025-02,Optimal Multi-Draft,Towards Optimal Multi-draft Speculative Decoding,Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Collection [core],Distribution-Level Exactness [core],--,https://www.semanticscholar.org/paper/50af31e0aa76103dfb0ed901acea0bddbdbaef8d,,
2025-02,Hierarchical Drafting,Lossless Acceleration of Large Language Models with Hierarchical Drafting based on Temporal Locality in Speculative Decoding,Context-Derived Nonparametric Source [core],Chain [core],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2502.05609v1,https://github.com/zomss/Hierarchy_Drafting,
2025-02,Heterogeneous Vocab SD,Accelerating LLM Inference with Lossless Speculative Decoding Algorithms for Heterogeneous Vocabularies,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Serving Systems [core]; Performance Analysis [core],https://www.semanticscholar.org/paper/b2fecb8aa7218b00334e8fff8162509aa31761fe,https://github.com/keyboardAnt/hf-bench,
2025-02,GRIFFIN,GRIFFIN: Effective Token Alignment for Faster Speculative Decoding,Proposal-State Modeling [core],Tree [core],Token-Level Exactness [core],--,https://proceedings.neurips.cc/paper_files/paper/2025/hash/b6e67ae290635d0874c4cb43ba2a2cfb-Abstract-Conference.html,https://github.com/hsj576/GRIFFIN,
2025-02,Fuzzy,Fuzzy Speculative Decoding for a Tunable Accuracy-Runtime Tradeoff,Autoregressive External Drafter [core]; Proposal-State Modeling [inherited],Chain [core],Relaxed Acceptance [core],Speculation Control [core]; Performance Analysis [core],https://aclanthology.org/2025.findings-acl.1346/,https://github.com/maxholsman/fsd,
2025-02,FR-Spec,FR-Spec: Accelerating Large-Vocabulary Language Models via Frequency-Ranked Speculative Sampling,Target-Feature Readout [inherited]; Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [core],Serving Systems [core]; Performance Analysis [core],https://aclanthology.org/2025.acl-long.198/,https://github.com/thunlp/FR-Spec,
2025-02,Dynamic Tree Attention,Acceleration Multiple Heads Decoding for LLM via Dynamic Tree Attention,Target-Feature Readout [inherited],Tree [core],Token-Level Exactness [core],Serving Systems [core],https://arxiv.org/abs/2502.05947,https://github.com/zzd1992/MEDUSA-Plus,
2025-02,DReSD,DReSD: Dense Retrieval for Speculative Decoding,Context-Derived Nonparametric Source [core],Chain [core]; Collection [core],Token-Level Exactness [core],--,https://aclanthology.org/2025.findings-acl.1017/,https://github.com/huawei-noah/HEBO/tree/DReSD,
2025-02,CopySpec,CopySpec: Accelerating LLMs with Speculative Copy-and-Paste,Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [core],Chain [core],Token-Level Exactness [core],--,https://aclanthology.org/2025.emnlp-main.1337/,https://github.com/RazvanDu/CopySpec,
2025-01,SpecStreaming-MSA,Speculative Streaming: Efficient and Scalable Speculative Decoding with Multi-Stream Attention,Proposal-State Modeling [core],Collection [core],Distribution-Level Exactness [core]; Relaxed Acceptance [core],--,https://doi.org/10.18653/v1/2025.emnlp-main.986,,
2025-01,RSD,Reward-Guided Speculative Decoding for Efficient LLM Reasoning,Autoregressive External Drafter [inherited],Chain [core],Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2501.19324v1,https://github.com/BaohaoLiao/RSD,
2025-01,Judge Decoding,Judge Decoding: Faster Speculative Sampling Requires Going Beyond Model Alignment,Autoregressive External Drafter [inherited],Chain [inherited],Relaxed Acceptance [core],Performance Analysis [core],https://arxiv.org/abs/2501.19309v1,,
2025-01,Drop-In Adaptation,A Drop-In Solution for On-the-Fly Adaptation of Speculative Decoding in Large Language Models,Autoregressive External Drafter [core]; Proposal-State Modeling [inherited],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://doi.org/10.18653/v1/2025.acl-long.482,,
2024-12,PLD+,PLD+: Accelerating LLM inference by leveraging Language Model Artifacts,Context-Derived Nonparametric Source [core],Chain [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2412.01447v1,,
2024-12,Falcon,Falcon: Faster and Parallel Inference of Large Language Models through Enhanced Semi-Autoregressive Drafting and Custom-Designed Decoding Tree,Proposal-State Modeling [core],Block [core]; Tree [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2412.12639v3,,
2024-12,Dovetail,Dovetail: A CPU/GPU Heterogeneous Speculative Decoding for LLM inference,Proposal-State Modeling [core],Chain [core],Distribution-Level Exactness [core],Serving Systems [core]; Performance Analysis [core],https://aclanthology.org/2025.emnlp-main.879/,https://github.com/ddInference/Dovetail,
2024-12,CTC Drafter,Speculative Decoding with CTC-based Draft Model for LLM Inference Acceleration,Target-Feature Readout [core],Block [core]; Tree [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2412.00061,,
2024-11,Theoretical,A Theoretical Perspective for Speculative Decoding Algorithm,Autoregressive External Drafter [inherited],Chain [inherited],Bounded Approximation [core],Performance Analysis [core],https://arxiv.org/abs/2411.00841,,
2024-11,The N-Grammys,The N-Grammys: Accelerating Autoregressive Inference with Learning-Free Batched Speculation,Context-Derived Nonparametric Source [core],Collection [core],Token-Level Exactness [inherited],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2411.03786,,
2024-11,SuffixDecoding,SuffixDecoding: Extreme Speculative Decoding for Emerging AI Applications,Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [core],Tree [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.48550/arxiv.2411.04975,https://github.com/snowflakedb/ArcticInference,
2024-11,Speculative Side Channels,When Speculation Spills Secrets: Side Channels via Speculative Decoding In LLMs,Proposal-State Modeling [inherited],Chain [inherited]; Tree [inherited],--,Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2411.01076,,
2024-11,SpecHub,SpecHub: Provable Acceleration to Multi-Draft Speculative Decoding,Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Collection [core],Distribution-Level Exactness [core],--,https://aclanthology.org/2024.emnlp-main.1148/,https://github.com/MasterGodzilla/Speculative_decoding_OT,
2024-11,Self-Verification SD,Draft Model Knows When to Stop: Self-Verification Speculative Decoding for Long-Form Generation,Autoregressive External Drafter [core]; Proposal-State Modeling [inherited],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://aclanthology.org/2025.emnlp-main.844/,,
2024-11,SSSD,SSSD: Simply-Scalable Speculative Decoding,Context-Derived Nonparametric Source [core],Tree [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core]; Performance Analysis [core],https://doi.org/10.48550/arxiv.2411.05894,https://github.com/huawei-csl/sssd_speculator,
2024-11,SAM Decoding,SAM Decoding: Speculative Decoding via Suffix Automaton,Proposal-State Modeling [inherited]; Context-Derived Nonparametric Source [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.48550/arxiv.2411.10666,,
2024-11,FastDraft,FastDraft: How to Train Your Draft,Autoregressive External Drafter [core],Chain [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2411.11055,,
2024-10,Tensor-MTP,Faster Language Models with Better Multi-Token Prediction Using Tensor Decomposition,Target-Feature Readout [core],Block [inherited]; Tree [inherited],Token-Level Exactness [inherited],Speculation Control [inherited],https://arxiv.org/abs/2410.17765v2,,
2024-10,Temperature-Centric,Temperature-Centric Investigation of Speculative Decoding with Knowledge Distillation,Autoregressive External Drafter [core],Chain [inherited],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2410.10141,https://github.com/ozyyshr/TempSpec,
2024-10,SWIFT,SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration,Target-Path Execution [core],Chain [core]; Tree [core],Token-Level Exactness [core]; Distribution-Level Exactness [core],State Management [core]; Speculation Control [core],https://arxiv.org/abs/2410.06916v2,https://github.com/hemingkx/SWIFT,
2024-10,QSpec,QSpec: Speculative Decoding with Complementary Quantization Schemes,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],Serving Systems [core]; Performance Analysis [core],https://doi.org/10.48550/arxiv.2410.11305,,
2024-10,ParallelSpec,ParallelSpec: Parallel Drafter for Efficient Speculative Decoding,Non-Autoregressive External Drafter [core],Block [core]; Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2410.05589v1,,
2024-10,Multi-Draft,Multi-Draft Speculative Sampling: Canonical Decomposition and Theoretical Limits,Autoregressive External Drafter [inherited],Collection [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2410.18234,,
2024-10,Inevitable,Inevitable Trade-off between Watermark Strength and Speculative Sampling Efficiency for Language Models,Autoregressive External Drafter [core],Chain [inherited],Distribution-Level Exactness [core],Performance Analysis [core],https://arxiv.org/abs/2410.20418v1,,
2024-10,DySpec,DySpec: Faster speculative decoding with dynamic token tree structure,Autoregressive External Drafter [core],Tree [core],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://doi.org/10.1007/s11280-025-01344-0,,
2024-10,Draft on the Fly,Draft on the Fly: Adaptive Self-Speculative Decoding using Cosine Similarity,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2410.01028v1,,
2024-10,Cerberus,Cerberus: Efficient Inference with Adaptive Parallel Decoding and Sequential Knowledge Enhancement,Proposal-State Modeling [core],Tree [core],Token-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2410.13344v1,,
2024-10,AMUSD,AMUSD: Asynchronous Multi-Device Speculative Decoding for LLM Acceleration,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2410.17375v1,,
2024-09,Multi-Candidate SD,Improving Multi-candidate Speculative Decoding,Autoregressive External Drafter [core]; Target-Path Execution [core],Tree [core]; Collection [core],Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2409.10644v3,,
2024-09,Dynamic Depth Decoding,Dynamic Depth Decoding: Faster Speculative Decoding for LLMs,Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2409.00142v1,,
2024-08,SCD,Speculative Contrastive Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Relaxed Acceptance [core],Performance Analysis [core],https://doi.org/10.18653/v1/2024.acl-short.5,,
2024-08,PEARL,PEARL: Parallel Speculative Decoding with Adaptive Draft Length,Autoregressive External Drafter [core],Chain [inherited],Distribution-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2408.11850v3,,
2024-08,LASER,Efficiency Unleashed: Inference Acceleration for LLM-based Recommender Systems with Speculative Decoding,Context-Derived Nonparametric Source [core],Tree [core],Relaxed Acceptance [core],--,https://arxiv.org/abs/2408.05676,https://github.com/YunjiaXi/LASER,
2024-08,KOALA,KOALA: Enhancing Speculative Decoding for LLM via Multi-Layer Draft Heads with Adversarial Learning,Target-Feature Readout [core]; Proposal-State Modeling [core],Chain [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2408.08146,,
2024-08,Harmonized Repr.,Learning Harmonized Representations for Speculative Sampling,Proposal-State Modeling [core],Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2408.15766v3,,
2024-08,Feature Alignment,Boosting Lossless Speculative Decoding via Feature Sampling and Partial Alignment Distillation,Proposal-State Modeling [core],Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2408.15562v1,,
2024-08,Draft & Verify,Draft & Verify: Lossless Large Language Model Acceleration via Self-Speculative Decoding,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],Speculation Control [core],https://aclanthology.org/2024.acl-long.607/,https://github.com/dilab-zju/self-speculative-decoding,
2024-08,DISD,Coupling Without Communication and Drafter-Invariant Speculative Decoding,Autoregressive External Drafter [inherited],Collection [inherited],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2408.07978,https://github.com/majid-daliri/DISD,
2024-08,Context-Aware,Context-Aware Assistant Selection for Improved Inference Acceleration with Large Language Models,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2408.08470v4,,
2024-08,CREST,CREST: Effectively Compacting a Datastore For Retrieval-Based Speculative Decoding,Context-Derived Nonparametric Source [core],Tree [core],Token-Level Exactness [core],Speculation Control [core]; Performance Analysis [core],https://doi.org/10.48550/arxiv.2408.04678,,
2024-07,S2D,S2D: Sorted Speculative Decoding For More Efficient Deployment of Nested Large Language Models,Autoregressive External Drafter [core],Chain [core],Token-Level Exactness [core],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2407.01955v1,,
2024-07,Graph-Structured,Graph-Structured Speculative Decoding,Autoregressive External Drafter [core],Tree [core],Token-Level Exactness [core]; Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2407.16207v1,,
2024-07,Dynamic Self-SD,Accelerating Large Language Model Inference with Self-Supervised Early Exits,Target-Path Execution [core]; Target-Feature Readout [core],Chain [core],Token-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2407.21082v2,https://github.com/FlorianVal/BranchyLLM,
2024-06,SpecExec,SpecExec: Massively Parallel Speculative Decoding for Interactive LLM Inference on Consumer Devices,Autoregressive External Drafter [core],Tree [core],Distribution-Level Exactness [core],State Management [core]; Speculation Control [core]; Serving Systems [core],https://doi.org/10.48550/arxiv.2406.02532,https://github.com/yandex-research/specexec,
2024-06,Optimized,Optimized Speculative Sampling for GPU Hardware Accelerators,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited]; Relaxed Acceptance [core],Serving Systems [core],https://arxiv.org/abs/2406.11016v2,,
2024-06,OPT-Tree,OPT-Tree: Speculative Decoding with Adaptive Draft Tree Structure,Autoregressive External Drafter [inherited]; Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [core],Speculation Control [core],https://doi.org/10.1162/tacl_a_00735,https://github.com/Jikai0Wang/OPT-Tree,
2024-06,Multilingual Drafters,Towards Fast Multilingual LLM Inference: Speculative Decoding and Specialized Drafters,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Performance Analysis [core],https://doi.org/10.18653/v1/2024.emnlp-main.602,https://github.com/Kthyeon/Multilingual-SpecBench,
2024-06,GenZ,Demystifying AI Platform Design for Distributed Inference of Next-Generation LLM models,Autoregressive External Drafter [inherited],Chain [inherited],--,Performance Analysis [core],https://arxiv.org/abs/2406.01698v3,https://github.com/abhibambhaniya/GenZ-LLM-Analyzer,
2024-06,Early-Exit Thompson,Speculative Decoding via Early-exiting for Faster LLM Inference with Thompson Sampling Control Mechanism,Target-Path Execution [core]; Target-Feature Readout [core],Chain [core],Token-Level Exactness [core]; Distribution-Level Exactness [core],State Management [inherited]; Speculation Control [core],https://doi.org/10.18653/v1/2024.findings-acl.179,,
2024-06,EAGLE-2,EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees,Proposal-State Modeling [inherited],Tree [core],Distribution-Level Exactness [inherited],Speculation Control [core]; Serving Systems [inherited],https://arxiv.org/abs/2406.16858v2,,
2024-06,Amphista,Amphista: Bi-directional Multi-head Decoding for Accelerating LLM Inference,Proposal-State Modeling [core],Tree [core]; Collection [core],Token-Level Exactness [core]; Distribution-Level Exactness [core],--,https://arxiv.org/abs/2406.13170v2,,
2024-05,Speculative Cascades,Faster Cascades via Speculative Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [core]; Relaxed Acceptance [core],Performance Analysis [core],https://arxiv.org/abs/2405.19261v2,,
2024-05,SpecDec++,SpecDec++: Boosting Speculative Decoding via Adaptive Candidate Lengths,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://doi.org/10.48550/arxiv.2405.19715,,
2024-05,S3D,S3D: A Simple and Cost-Effective Self-Speculative Decoding Scheme for Low-Memory GPUs,Target-Path Execution [core],Block [core],Token-Level Exactness [core]; Relaxed Acceptance [core],Serving Systems [core]; Performance Analysis [core],https://arxiv.org/abs/2405.20314v2,,
2024-05,NEST,Nearest Neighbor Speculative Decoding for LLM Generation and Attribution,Context-Derived Nonparametric Source [core],Chain [inherited],Relaxed Acceptance [core],Speculation Control [inherited],https://doi.org/10.52202/079017-2574,,
2024-05,How Speculative,How Speculative Can Speculative Decoding Be?,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],Performance Analysis [core],https://aclanthology.org/2024.lrec-main.725/,https://github.com/ZhuoruiLiu12/SpecGame,
2024-05,Dynamic Lookahead,Dynamic Speculation Lookahead Accelerates Speculative Decoding of Large Language Models,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://doi.org/10.48550/arxiv.2405.04304,,
2024-05,Clover,Clover: Regressive Lightweight Speculative Decoding with Sequential Knowledge,Proposal-State Modeling [core],Tree [core]; Collection [core],Token-Level Exactness [core]; Distribution-Level Exactness [core],--,https://arxiv.org/abs/2405.00263v1,,
2024-04,LayerSkip,LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding,Target-Path Execution [core],Chain [core],Token-Level Exactness [core],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2404.16710,https://github.com/facebookresearch/LayerSkip,
2024-04,Hidden Transfer,Parallel Decoding via Hidden Transfer for Lossless Large Language Model Acceleration,Target-Path Execution [core]; Proposal-State Modeling [core],Tree [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2404.12022v1,,
2024-04,Exploring,Exploring and Improving Drafts in Blockwise Parallel Decoding,Target-Feature Readout [core]; Context-Derived Nonparametric Source [core],Block [core]; Collection [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2404.09221v2,,
2024-04,BASS,BASS: Batched Attention-optimized Speculative Sampling,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2404.15778v2,,
2024-03,SDSAT,SDSAT: Accelerating LLM Inference through Speculative Decoding with Semantic Adaptive Tokens,Target-Path Execution [core],Block [core]; Tree [core],Token-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2403.18647v2,https://github.com/ainergy-ml/SDSAT,
2024-03,Recurrent,Recurrent Drafter for Fast Speculative Decoding in Large Language Models,Proposal-State Modeling [core],Chain [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2403.09919,https://github.com/apple/ml-recurrent-drafter,
2024-03,Direct,Direct Alignment of Draft Model for Speculative Decoding with Chat-Fine-Tuned LLMs,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2403.00858v4,,
2024-03,Block Verification,Block Verification Accelerates Speculative Decoding,Autoregressive External Drafter [inherited],Block [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2403.10444,,
2024-02,SpecStreaming,Speculative Streaming: Fast LLM Inference without Auxiliary Models,Proposal-State Modeling [core],Tree [core],Token-Level Exactness [core],State Management [core]; Serving Systems [core],https://arxiv.org/abs/2402.11131v1,,
2024-02,Smart Auto-Correct,Generation Meets Verification: Accelerating Large Language Model Inference with Smart Parallel Auto-Correct Decoding,Target-Path Execution [core],Block [core],Distribution-Level Exactness [core],--,https://doi.org/10.18653/v1/2024.findings-acl.313,,
2024-02,Sequoia,Sequoia: Scalable and Robust Speculative Decoding,Autoregressive External Drafter [inherited],Tree [core],Distribution-Level Exactness [core],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2402.12374,https://github.com/Infini-AI-Lab/Sequoia,
2024-02,ProPD,ProPD: Dynamic Token Tree Pruning and Generation for LLM Parallel Decoding,Target-Feature Readout [inherited],Tree [core],Token-Level Exactness [core],State Management [core]; Speculation Control [core],https://arxiv.org/abs/2402.13485v1,,
2024-02,Ouroboros,Ouroboros: Generating Longer Drafts Phrase by Phrase for Faster Speculative Decoding,Autoregressive External Drafter [core]; Context-Derived Nonparametric Source [core],Block [core]; Tree [core],Distribution-Level Exactness [core],Speculation Control [core],https://doi.org/10.18653/v1/2024.emnlp-main.742,,
2024-02,Lookahead,Break the Sequential Dependency of LLM Inference Using Lookahead Decoding,Context-Derived Nonparametric Source [core],Collection [core],Distribution-Level Exactness [core],State Management [core]; Speculation Control [core],https://proceedings.mlr.press/v235/fu24a.html,https://github.com/hao-ai-lab/LookaheadDecoding,
2024-02,Hydra,Hydra: Sequentially-Dependent Draft Heads for Medusa Decoding,Target-Feature Readout [core],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited]; Relaxed Acceptance [inherited],Speculation Control [inherited]; Serving Systems [inherited],https://arxiv.org/abs/2402.05109v2,https://github.com/zankner/Hydra,
2024-02,GliDe with a CaPE,GliDe with a CaPE: A Low-Hassle Method to Accelerate Speculative Decoding,Autoregressive External Drafter [core],Collection [core],Distribution-Level Exactness [core],State Management [core]; Speculation Control [core],https://arxiv.org/abs/2402.02082v1,,
2024-02,Decoding,Decoding Speculative Decoding,Autoregressive External Drafter [inherited],Chain [inherited],Distribution-Level Exactness [inherited],Performance Analysis [core],https://arxiv.org/abs/2402.01528,,
2024-02,Chimera,Chimera: A Lossless Decoding Method for Accelerating Large Language Models Inference by Fusing all Tokens,Target-Feature Readout [core],Tree [core],Token-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2402.15758,,
2024-01,Multi-Candidate,Multi-Candidate Speculative Decoding,Autoregressive External Drafter [core],Collection [core]; Tree [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2401.06706v1,,
2024-01,Medusa,Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads,Target-Feature Readout [core],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited]; Relaxed Acceptance [core],Serving Systems [core],https://arxiv.org/abs/2401.10774v3,https://github.com/FasterDecoding/Medusa,
2024-01,Kangaroo,Kangaroo: Lossless Self-Speculative Decoding for Accelerating LLMs via Double Early Exiting,Target-Path Execution [core]; Target-Feature Readout [core],Chain [core],Token-Level Exactness [core],State Management [inherited]; Speculation Control [core],https://doi.org/10.52202/079017-0381,,
2024-01,EAGLE,EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty,Proposal-State Modeling [core],Chain [core]; Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Serving Systems [inherited],https://arxiv.org/abs/2401.15077v3,,
2023-11,REST,REST: Retrieval-Based Speculative Decoding,Context-Derived Nonparametric Source [core],Tree [core],Token-Level Exactness [inherited]; Distribution-Level Exactness [inherited],Speculation Control [inherited]; Serving Systems [inherited],https://aclanthology.org/2024.naacl-long.88/,https://github.com/FasterDecoding/REST,
2023-11,PaSS,PaSS: Parallel Speculative Sampling,Target-Path Execution [core],Block [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2311.13581v1,,
2023-10,SpecTr,SpecTr: Fast Speculative Decoding via Optimal Transport,Autoregressive External Drafter [inherited],Tree [core]; Collection [core],Distribution-Level Exactness [core],--,https://arxiv.org/abs/2310.15141,,
2023-10,Online SD,Online Speculative Decoding,Autoregressive External Drafter [core]; Target-Feature Readout [inherited],Chain [inherited],Distribution-Level Exactness [inherited],Speculation Control [core]; Serving Systems [core],https://proceedings.mlr.press/v235/liu24y.html,https://github.com/LiuXiaoxuanPKU/OSD,
2023-10,DistillSpec,DistillSpec: Improving Speculative Decoding via Knowledge Distillation,Autoregressive External Drafter [core],Chain [inherited],Distribution-Level Exactness [inherited]; Relaxed Acceptance [core],Speculation Control [inherited],https://arxiv.org/abs/2310.08461,,
2023-10,Batching Synergy,The Synergy of Speculative Decoding and Batching in Serving Large Language Models,Autoregressive External Drafter [core],Chain [core],Token-Level Exactness [core],State Management [core]; Speculation Control [core]; Performance Analysis [core],https://doi.org/10.48550/arxiv.2310.18813,,
2023-08,Staged,Accelerating LLM Inference with Staged Speculative Decoding,Autoregressive External Drafter [core]; Context-Derived Nonparametric Source [core],Tree [core],Distribution-Level Exactness [inherited],Speculation Control [core]; Performance Analysis [core],https://arxiv.org/abs/2308.04623v1,,
2023-05,SpecInfer,SpecInfer: Accelerating Large Language Model Serving with Tree-based Speculative Inference and Verification,Autoregressive External Drafter [inherited],Tree [core],Token-Level Exactness [core]; Distribution-Level Exactness [core],State Management [core]; Serving Systems [core],https://doi.org/10.1145/3620666.3651335,https://github.com/flexflow/FlexFlow/,
2023-04,Inference with Reference,Inference with Reference: Lossless Acceleration of Large Language Models,Context-Derived Nonparametric Source [core],Chain [core],Token-Level Exactness [core],--,https://arxiv.org/abs/2304.04487,https://github.com/microsoft/unilm,
2023-02,Speculative Sampling,Accelerating Large Language Model Decoding with Speculative Sampling,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://arxiv.org/abs/2302.01318v1,,
2023-02,Big-Little Decoder,Speculative Decoding with Big Little Decoder,Autoregressive External Drafter [core],Chain [core],Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/2302.07863,https://github.com/kssteven418/BigLittleDecoder,
2022-11,Speculative Decoding,Fast Inference from Transformers via Speculative Decoding,Autoregressive External Drafter [core],Chain [core],Distribution-Level Exactness [core],Speculation Control [core],https://doi.org/10.48550/arxiv.2211.17192,,
2022-03,SpecDec,Speculative Decoding: Exploiting Speculative Execution for Accelerating Seq2seq Generation,Non-Autoregressive External Drafter [core],Block [core],Relaxed Acceptance [core],Speculation Control [core]; Serving Systems [core],https://arxiv.org/abs/2203.16487v6,,
2018-11,Blockwise,Blockwise Parallel Decoding for Deep Autoregressive Models,Target-Feature Readout [core],Block [core],Token-Level Exactness [core]; Relaxed Acceptance [core],Speculation Control [core],https://arxiv.org/abs/1811.03115v1,https://github.com/tensorflow/tensor2tensor,