The paper introduces SwitchSD, an adaptive framework that controls speculative decoding by distinguishing genuine copy intent from accidental repetitions using lightweight probes on a model’s internal representations. SwitchSD dynamically switches between neural drafting and context-based copying, achieving up to 15% throughput gains over existing baselines such as EAGLE3. The approach turns copying from a noisy heuristic into a principled, model-aware decoding regime.
By Roy Eisenstadt, Ido Cohen, Edo Cohen-Karlik, Lior Wolf, Itamar Zimerman
arXiv:2601.03199v2 Announce Type: replace-cross
Abstract: Diffusion language models (DLMs) have shown strong potential for general natural language tasks with in-context examples. Existing In-Context...
By Yang Li, Han Meng, Chenan Wang, Zhenyu Bi, Xuan Wang, Haipeng Chen
arXiv:2509.11252v3 Announce Type: replace-cross
Abstract: LLMs have become the mainstream approaches to code generation. Existing LLMs mainly employ autoregressive generation, i.e. generating code to...
By Chengze Li, Yitong Zhang, Jia Li, Liyi Cai, Ge Li
arXiv:2609.36173v1 Announce Type: cross
Abstract: Parallel speculative drafting generates multiple candidates in one backbone pass, but independent token selection can produce inconsistent continuati...
By Haohui Zhang, Keyu Chen, Haocheng Sun, Weibo Gu, Ruizhi Qiao, Xing Sun, Bo Jiang
The paper introduces SwitchSD, an adaptive framework that treats copying as a latent control signal in large language model decoding. By training lightweight probes on internal representations, SwitchSD accurately detects genuine copy intent (AUC > 0.99) and dynamically switches between neural drafting and context-based copying. Experiments on Llama and Qwen models show up to 15 % throughput gains over state‑of‑the‑art baselines such as EAGLE3.
The paper introduces DCOCO (DFlash Contextual Corrector), a lightweight mechanism that restores intra-block dependencies in speculative decoding by evaluating multiple candidate tokens for contextual compatibility across adjacent draft positions. DCOCO maintains parallel drafting while enabling contextually informed token selection at low additional cost. Across multiple verifier backbones and benchmarks, DCOCO achieves the highest average throughput, delivering speedups of up to 6.6× over vanilla decoding and outperforming competing methods by 6–23% under concurrent serving loads.
By Roy Eisenstadt, Harel Lidar, Tal Gonen, Omer Luxembourg, Itamar Zimerman, Lior Wolf, Eli David, Ilan Naiman
The paper introduces PILL, a new infilling technique for diffusion language models that eliminates the need for a preset initial length and reduces inference overhead. PILL uses probing-based length-free decoding, cutting down on extra forward passes and speeding up generation. Experiments across five diffusion models and eight benchmarks show PILL outperforms the strongest baseline with higher pass rates and BLEU-2 scores while running 1.82× faster.
By Haobo Xu, Sirui Chen, Yuanchen Bei, Lingjie Chen, Yuchen Yan, Dongqi Fu, Jingrui He, Hanghang Tong
arXiv:2606.21803v2 Announce Type: replace
Abstract: Next-token prediction is the self-supervised signal that trains language models, and every observed prompt token provides the same signal at test t...
By Xuan Ouyang, Zefan Cai, Junjie Hu
arXiv:2607. 27600v1 Announce Type: new Abstract: Key-value (KV) cache management through compression and eviction strategies has emerged as an important research direction in recent years.
By Stephen Gould, Anton van den Hengel
TeCoP is a training‑free decoder that accelerates diffusion large language model (dLLM) decoding by decoupling prediction readiness from context utility. It uses tentative context pipelining and draft‑and‑verify to reuse stable but uncommitted predictions as provisional cross‑block context, with commitment based on confidence and stability checks. Across four dLLMs on GSM8K, Minerva MATH, and MBPP, TeCoP cuts forward evaluations by up to 87% and boosts throughput up to 7.5‑fold without KV caching and 11.77‑fold with caching, outperforming Fast‑dLLM and DSB while maintaining accuracy within one percentage point.
By Jui-Chien Tsou, Hsuan-Chi Liu, Wei-Chin Wang, Yu-Syuan Xu, Chun-Yi Lee
The paper presents a method for compiling autoregressive large language model (LLM) control into a policy‑exact parallel inference scheme for Hierarchical Retrieval‑Augmented Generation (RAG). By constructing a shallow action‑prefix tree and using a single root prefill plus tree‑causal probes, the authors eliminate the need for serial decoding while preserving the exact control policy in exact arithmetic. Experiments on MetaX devices show that the compiled approach matches the serial policy on 84–89 % of questions and reduces control latency by 35–47 % compared to token‑wise serial decoding.
By Zehao Lin, Qianfeng Shen, Xixuan Hao, Shaobo Cui, Chunyu Li, Kai Chen, Bo Tang, Zhenjun Wang, Jin Yan, Zhiyu Li, Feiyu Xiong
arXiv:2410.21647v5 Announce Type: replace-cross
Abstract: Recently, a number of repository-level code generation benchmarks-such as CoderEval, DevEval, RepoEval, RepoBench, and LongCodeArena-have eme...
By Shanchao Liang, Yiran Hu, Nan Jiang, Lin Tan