Pith. sign in

REVIEW 20 cited by

Unlocking Efficiency in Large Language Model Inference: A Comprehensive Survey of Speculative Decoding

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2401.07851 v3 pith:RQB5FVJO submitted 2024-01-15 cs.CL

Unlocking Efficiency in Large Language Model Inference: A Comprehensive Survey of Speculative Decoding

classification cs.CL
keywords decodinginferencespeculativeanalysisautoregressivecomprehensivelanguagelarge
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

To mitigate the high inference latency stemming from autoregressive decoding in Large Language Models (LLMs), Speculative Decoding has emerged as a novel decoding paradigm for LLM inference. In each decoding step, this method first drafts several future tokens efficiently and then verifies them in parallel. Unlike autoregressive decoding, Speculative Decoding facilitates the simultaneous decoding of multiple tokens per step, thereby accelerating inference. This paper presents a comprehensive overview and analysis of this promising decoding paradigm. We begin by providing a formal definition and formulation of Speculative Decoding. Then, we organize in-depth discussions on its key facets, such as drafter selection and verification strategies. Furthermore, we present a comparative analysis of leading methods under third-party testing environments. We aim for this work to serve as a catalyst for further research on Speculative Decoding, ultimately contributing to more efficient LLM inference.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 20 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Draft Less, Retrieve More: Hybrid Tree Construction for Speculative Decoding

    cs.LG 2026-05 unverdicted novelty 7.0

    Graft combines pruning and retrieval in a sequential mechanism to build hybrid draft trees for speculative decoding, delivering up to 5.41× speedup and 21.8% better average speedup than EAGLE-3 on large models.

  2. FASER: Fine-Grained Phase Management for Speculative Decoding in Dynamic LLM Serving

    cs.DC 2026-04 unverdicted novelty 7.0

    FASER delivers up to 53% higher throughput and 1.92x lower latency in dynamic LLM serving by adjusting speculative lengths per request, early pruning of rejects, and overlapping draft/verification phases via frontiers.

  3. When RL Meets Adaptive Speculative Training: A Unified Training-Serving System

    cs.LG 2026-02 conditional novelty 7.0

    Aurora unifies speculative decoder training and serving via asynchronous RL on inference traces, delivering 1.5x day-0 speedup on frontier models and 1.25x adaptation gains on distribution shifts.

  4. Revisiting Lossy Verification in Speculative Decoding: Mechanisms, Trade-offs, and Failure Modes

    cs.CL 2026-07 conditional novelty 6.0

    Lossy speculative-decoding verification splits into truncation-based and collaborative methods; truncation-based methods underperform their matched baselines, and capping draft overshoot preserves quality.

  5. PyroDash: Cost-Efficient Token-Level Small-Large Language Model Collaborative Inference

    cs.CL 2026-07 conditional novelty 6.0

    PyroDash lets a small model decide mid-generation to hand off once to a frozen LLM, achieving 64.04% accuracy (+6.36 points vs LLM) at lambda=0.05 and 96.4% lower estimated cost at lambda=0.6.

  6. Speculation at a Distance: Where Edge-Cloud Speculative Decoding Actually Pays Off

    cs.DC 2026-06 unverdicted novelty 6.0

    Analytical bounds demonstrate edge-cloud DSD improves latency only in low-RTT regimes and primarily benefits multi-tenant throughput via draft offloading under client overlap.

  7. SURF: Separation via Unsupervised Remixing Flow

    cs.SD 2026-06 unverdicted novelty 6.0

    SURF uses teacher-student remixing within a flow-matching framework to achieve unsupervised source separation and reports new state-of-the-art results on audio and image benchmarks.

  8. Breaking the Reward Barrier: Accelerating Tree-of-Thought Reasoning via Speculative Exploration

    cs.LG 2026-05 unverdicted novelty 6.0

    SPEX delivers 1.2-3x speedup on ToT algorithms via speculative path selection, dynamic budget allocation, and adaptive early termination, reaching up to 4.1x when combined with token-level speculative decoding.

  9. Breaking the Reward Barrier: Accelerating Tree-of-Thought Reasoning via Speculative Exploration

    cs.LG 2026-05 unverdicted novelty 6.0

    SPEX accelerates Tree-of-Thought LLM reasoning 1.2-3x via speculative path selection, dynamic budget allocation across queries, and adaptive early termination, with up to 4.1x when combined with token speculative decoding.

  10. When Hidden States Drift: Can KV Caches Rescue Long-Range Speculative Decoding?

    cs.CL 2026-04 unverdicted novelty 6.0

    KV cache reuse improves long-range draft acceptance in speculative decoding but delivers only marginal end-to-end speedups due to drafter limitations.

  11. When Hidden States Drift: Can KV Caches Rescue Long-Range Speculative Decoding?

    cs.CL 2026-04 unverdicted novelty 6.0

    KV cache reuse improves long-range draft acceptance rates in speculative decoding but delivers only marginal end-to-end speedups because shallow drafters cannot accurately estimate target queries and receive sparse gr...

  12. DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models

    cs.CL 2024-02 unverdicted novelty 6.0

    DeepSeekMath 7B reaches 51.7% on MATH via continued pretraining on curated web math data and Group Relative Policy Optimization.

  13. AngelSpec: Towards Real-World High Performance Inference with Speculative Decoding

    cs.CL 2026-07 conditional novelty 5.0

    AngelSpec + DFly pair a chat MTP drafter with a code/math block-diffusion drafter and load-aware verification pruning, reaching up to 2.4x AR throughput on Hy3-A21B.

  14. D-cut: Adaptive Verification Depth Pruning for Batched Speculative Decoding

    cs.CL 2026-07 conditional novelty 5.0

    D-Cut prunes low-confidence draft suffixes across a batch of requests, using a confidence-based ranking and a profiled hardware cost model, to raise batched speculative decoding speedup from 1.26x to 1.65x.

  15. RTP-LLM: High-Performance Alibaba LLM Inference Engine

    cs.OS 2026-05 unverdicted novelty 5.0

    RTP-LLM is a new LLM inference engine achieving 4.7x-6.3x model loading speedup and 1.12x-2.52x throughput gains over vLLM and SGLang via disaggregated phases, multi-tier KV cache, and modular optimizations in product...

  16. D-PACE: Dynamic Position-Aware Cross-Entropy for Parallel Speculative Drafting

    cs.LG 2026-05 unverdicted novelty 5.0

    D-PACE derives per-position weights from a surrogate of expected accepted draft length to shift training focus toward currently limiting positions, yielding measured gains in wall-clock speedup and emitted length acro...

  17. ECHO: Elastic Speculative Decoding with Sparse Gating for High-Concurrency Scenarios

    cs.DC 2026-03 unverdicted novelty 5.0

    ECHO uses sparse gating and elastic budget pivoting in a super-tree structure to achieve up to 5.35x speedup for LLM inference under high concurrency.

  18. Network Edge Inference for Large Language Models: Principles, Techniques, and Opportunities

    cs.DC 2026-04 unverdicted novelty 3.0

    A survey synthesizing challenges, system architectures, model optimizations, deployment methods, and resource management techniques for large language model inference at the network edge.

  19. A Survey on Efficient Inference for Large Language Models

    cs.CL 2024-04 accept novelty 3.0

    The paper surveys techniques to speed up and reduce the resource needs of LLM inference, organized by data-level, model-level, and system-level changes, with comparative experiments on representative methods.

  20. Small Language Models (SLMs) Can Still Pack a Punch: A survey (updated 2026)

    cs.CL 2025-01 unverdicted novelty 2.0

    A literature survey of Small Language Models (1-8B parameters) that can perform comparably or better than larger models, covering general-purpose and task-specific approaches plus creation techniques.