The paper investigates speculative decoding, a technique that speeds up language model inference by generating draft tokens and verifying them. It shows that when verification is relaxed (lossy), the resulting completions become longer—especially due to extended reasoning traces—despite accepting more tokens per round. The study evaluates various relaxed verification rules across multiple model pairs, finding that higher acceptance rates often lead to longer outputs, lower target-model probability tokens, and reduced accuracy, indicating that token acceptance alone is an unreliable metric for decoding speed.
By Chiatzen Wang, Bill Xu, Haochen Gu, Hongyang Zhang
arXiv:2602. 23881v2 Announce Type: replace Abstract: Speculative decoding accelerates autoregressive large language model (LLM) inference by using a lightweight draft model to propose candidate tokens that are then verified in parallel by the target model.
By Alexander Samarin, Sergei Krutikov, Anton Shevtsov, Sergei Skvortsov, Filipp Fisin, Alexander Golubev
The paper introduces APEX, a learned controller that improves speculative decoding for large language models by dynamically selecting the best speculation strategy and adjusting draft depth during generation. APEX uses request‑level routing among EAGLE‑3, n‑gram, and draft‑model speculation, and block‑level depth adaptation based on causal decoding signals and verifier feedback. Integrated into vLLM and tested with Qwen3‑8B, APEX achieves up to 5.24× speedup over autoregressive decoding while reducing wasted tokens by 41% compared to fixed‑depth n‑gram speculation.
By Manvi Jha, Zach Zhang, Zhichao Xu, Linbo Liu, Sai Muralidhar Jayanthi, Vinayak Arannil
The paper introduces TSS, a target-side sparsification framework that selectively skips layers in a target verifier during speculative decoding for domain-specific large language models. By exploring multi-layer skip configurations with an acceptance- and metric-aware breadth search, TSS reduces verification cost, increases draft acceptance, and can even improve downstream task performance without retraining. Experiments on Spec-Bench demonstrate consistent gains across domains and model scales, notably boosting translation throughput by 1.68× and improving BLEU scores significantly.
By Haibo Hu, Lianming Huang, Qiao Li, Nan Guan, Chun Jason Xue
arXiv:2610.07822v1 Announce Type: new
Abstract: Speculative decoding accelerates autoregressive generation by using a lightweight draft model to propose multiple tokens that are verified by a target...
By Shuhao Li, Fanghua Ye, Wanyu Lin, Tianyu Yuan, Xiaoyu Shen
arXiv:2609.24150v1 Announce Type: new
Abstract: Speculative decoding accelerates large language model (LLM) inference by using a lightweight draft model to generate multiple candidate tokens that are...
By Tianhua Xia, Mugilan Ganesan, Yifei Feng, Haiyu Wang, Maximilian Egger, Sai Qian Zhang
The paper introduces AdaptiveSpec, a training‑free speculative decoding method that simultaneously adapts the per‑step verification rule and the draft‑tree shape using signals generated during decoding. It replaces the fixed token‑match rule with a margin‑based threshold and adjusts tree depth, width, and node count based on draft confidence and recent acceptance history, allowing the total draft count to vary. Experiments on SGLang show up to 56% throughput gains over EAGLE‑3 while maintaining 93% of lossless task accuracy on GSM8K, MATH‑500, and HumanEval across three models.
By Oszk\'ar Urb\'an, Young D. Kwon, Stylianos I. Venieris, Cecilia Mascolo
arXiv:2609.38510v1 Announce Type: new
Abstract: Speculative decoding accelerates autoregressive LLMs by having a lightweight drafter propose tokens that the target model verifies in parallel. Diffusi...
By Longxuan Yu, Bingsen Chen, Peng Shi, Dongkyu Lee, Yi Xiang, Hideo Kobayashi, Sheng Zhang, Shuaichen Chang, Xing Niu, Zhuoyan Xu, Greg Ver Steeg, Jiarong Jiang
arXiv:2607. 24434v1 Announce Type: cross Abstract: Large Mixture-of-Experts (MoE) language models are attractive for end-device deployment because only a small subset of experts is active per token, but their routed expert weights often exceed accelerator memory.
By Dengke Han
arXiv:2606. 01019v1 Announce Type: cross Abstract: Large Language Model (LLM) generation remains expensive because autoregressive decoding calls the model once for each new token.
By Xin Su, Dawid Majchrowski, Fangyuan Yu, Vanshil Atul Shah, Sebastian Rogawski, Pawel Morkisz, Anahita Bhiwandiwalla, Phillip Howard
Verification-Aware Training (VAT) is a plug‑in framework that improves speculative decoding for large language models by simulating verification during training and using the resulting accept/reject patterns as supervision. VAT adds a lightweight binary verification head to predict whether each draft token will survive sequential verification, and replaces the fixed per‑position weighting with a verification‑adaptive schedule that keeps full weight up to the first rejection point. When applied to EAGLE‑3 and DFlash on Qwen3‑4B, Qwen3‑8B, and LLaMA‑3.1‑8B, VAT increases average acceptance length by up to 11.4% and wall‑clock speedup by up to 8.7%, yielding consistent gains across math, code, and chat benchmarks.
By Geonmo Gu, Byeongho Heo, HeeJae Jun, Yoohoon Kang, Sangmin Lee, Sangdoo Yun, Dongyoon Han
Large Mixture-of-Experts (MoE) language models are attractive for end-device deployment because only a small subset of experts is active per token, but their routed expert weights often exceed accelerator memory. We target latency-critical single-user settings where routed experts are staged on demand from CPU memory to a GPU or from Flash to a mobile NPU.