The paper introduces sparse collaborative decoding, where a draft model generates most tokens and a target model is called only for short corrective segments. It finds that local uncertainty metrics perform poorly as triggers, whereas aggregate trajectory-shape features achieve high predictive accuracy (AUC 0.86) for deciding when to invoke the target. Using this learned trigger improves accuracy by up to 20 points on GSM8K, 13.5 on MATH-500, and 8.3 on StrategyQA compared to open-loop schedules.
arXiv:2610.27379v1 Announce Type: cross Abstract: Block-diffusion speculative decoding accelerates large language model inference by drafting multiple tokens in parallel. However, one-step drafting i...
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...
arXiv:2610.28619v1 Announce Type: new Abstract: Multi-token prediction (MTP) can accelerate language-model decoding by drafting several tokens and verifying them in a single target-model pass. Howeve...
The paper introduces the Expected Decoding Rounds (EDR) objective, a theoretical framework that models speculative decoding as a Markov reward process and directly optimizes the number of decoding rounds. EDR weights local rejection costs by state occupancies, eliminating auxiliary hyperparameters and enabling unbiased stochastic optimization via temporal‑difference gradients. Applying EDR to fine‑tune state‑of‑the‑art draft models (DSpark and DFly) improves mean accepted length and outperforms prior training objectives across nine benchmarks in math reasoning, code generation, and chat.
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:2610.00623v1 Announce Type: new Abstract: Speculative decoding has achieved substantial lossless speedups for LLMs, but remains less effective for large vision-language models (LVLMs), where li...
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 explores a new lossy strategy for speculative decoding called verifier skipping, which allows certain draft prefixes to be committed without invoking the verifier. It investigates which confidence signals—raw token confidence, learned marginal survival, or learned conditional survival—best schedule these skips, finding that raw confidence often performs best despite higher positionwise AUROC for learned signals. Experiments on HumanEval with DiffuCoder-7B-Instruct and Qwen3-32B show that all three signals can reduce verifier calls by 9.6% to 13.5% while maintaining the same pass@1 as Strict SDD.
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.
LEAP: Learning Efficient Action Proposals For LLM Agents proposes a method to speed up large language model agents by training a small 0.6B drafter to predict target actions accurately. The approach uses a latency framework that balances drafting, verification, and execution costs, achieving up to 60% faster end‑to‑end wall clock time without reducing task success. LEAP can be online trained, eliminating the need for prior trace collection and making it practical for real‑world deployment.
arXiv:2609.37169v1 Announce Type: cross Abstract: Mid-training equips pretrained large language models with specialized and reasoning capabilities, but the returns of this stage are bounded since add...
Osprey is a target‑agnostic pre‑training method that bootstraps draft models for speculative decoding from existing small language models. By pruning to a shallow backbone, restoring language‑modeling capability with next‑token pretraining, and adapting via vocabulary alignment and distillation, Osprey reduces per‑target work to a lightweight adaptation step. Experiments show that a single Osprey backbone improves mean acceptance length by up to 22.7% and increases tokens per second by 17.5% across several large target models, especially on out‑of‑domain and multilingual data.