arXiv:2610.19792v1 Announce Type: new Abstract: Speculative decoding's effectiveness is governed by two quantities: the acceptance rate of drafted tokens and the cost of producing them. We show the t...
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.
arXiv:2609.14717v1 Announce Type: cross Abstract: Speculative decoding accelerates LLM inference by verifying multiple drafted tokens in parallel, allowing a single target forward pass to accept seve...
arXiv:2606. 12243v1 Announce Type: cross Abstract: Speculative decoding (SD) addresses the high inference costs of LLMs by having lightweight drafters generate candidates for large verifiers to validate in parallel.
arXiv:2608. 03447v1 Announce Type: cross Abstract: Speculative decoding accelerates autoregressive generation by verifying a draft block with a target model in parallel.
ResiSpec is a framework that improves speculative decoding for large language models by reshaping the residual distribution during verification. It addresses the problem of residual drift, where rejected candidates cause the target distribution to diverge from the draft model’s predictions, rendering later candidates ineffective. By aligning the verification process with the draft model’s high‑confidence regions, ResiSpec prevents candidate obsolescence and achieves up to 1.92× speedup over existing multi‑candidate methods.
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.
The paper introduces Packed-SS, a self‑speculative decoding schedule that combines drafting and verification into a single model call, reducing input size to linear in speculative width. By running the diffusion drafter one step ahead, the method keeps an unresolved suffix for the next iteration, avoiding the sequential wait of standard two‑call decoding. Experiments on Nemotron 3B/8B and SetDLM 1.7B demonstrate higher throughput and lower mean wall time under greedy decoding.
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.
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.
arXiv:2606. 03819v1 Announce Type: new Abstract: One-shot block drafters for speculative decoding generate the full draft in a single forward pass, achieving strong throughput by eliminating sequential token generation.
The paper introduces DLoop, a looped speculative decoding technique that adaptively performs multiple drafting stages before verification, allowing a draft model to continue generating tokens while confident. By verifying all accumulated draft tokens together and training the draft model to handle its own hidden states for unverified tokens, DLoop reduces the number of target‑model forward passes needed. Experiments across several speculative decoding methods show wall‑clock speedups of 5–41 % without sacrificing lossless decoding.