Graft: Hybrid Tree Construction for Speculative Decoding via Prune-Then-Retrieve
Graft is a training-free framework that improves speculative decoding by coupling dynamic-depth pruning with retrieval-based token compensation. Pruning reduces VRAM and compute overhead while freeing budget for retrieval, which fills topological gaps in the draft tree with near-zero additional cost. On short-context benchmarks, Graft achieves up to 5.41× speedup and improves average speedup over EAGLE-3 by up to 21.8% on Qwen3-235B. The method is evaluated across short- and long-context settings and extended to block-drafting paradigms.
Related guides (3)
Related events (8)
VIA-SD: Multi-tier speculative decoding via intra-model routing cuts rejection rates and boosts inference speed
VIA-SD introduces a three-tier verification framework for speculative decoding that routes draft tokens to a lightweight 'slim verifier' submodel for medium-confidence cases, reserving full-model verification only for uncertain tokens. Across four tasks and multiple model families, the method reduces rejection rates by 0.10–0.22 and achieves 10–20% speedups over strong speculative decoding baselines, with 2.5–3x acceleration over standard decoding. The approach is compatible with existing speculative decoding frameworks without retraining. The work proposes multi-tier speculative decoding as a general paradigm for scalable LLM inference.
Reroute: Training-free recoverable visual token routing for vision-language models
A new arXiv preprint proposes Reroute, a training-free plug-in that replaces the standard rank-and-remove visual token pruning paradigm in VLMs with a recoverable routing mechanism. Instead of permanently discarding low-ranked tokens, Reroute defers them to re-enter the candidate pool at later decoder stages, addressing the problem that token importance shifts across decoder depth. Evaluated on LLaVA-1.5 and Qwen backbones augmented with FastV, PDrop, and Nüwa pruning methods, Reroute improves grounding performance under aggressive token reduction without sacrificing general VQA accuracy. The approach preserves the theoretical compute and KV-cache budget of the underlying pruning method.
Faster Text Generation with Self-Speculative Decoding via LayerSkip
This Hugging Face blog post covers LayerSkip, a self-speculative decoding technique that accelerates text generation by using early exit from transformer layers to draft tokens, then verifying them with the full model. Unlike standard speculative decoding, LayerSkip requires no separate draft model, reducing memory overhead while still achieving inference speedups. The post likely covers integration with the Hugging Face ecosystem and practical performance benchmarks.
Accelerating Qwen3-8B Agent on Intel Core Ultra with Depth-Pruned Draft Models
Hugging Face and Intel demonstrate speculative decoding acceleration for the Qwen3-8B model on Intel Core Ultra client hardware using depth-pruned draft models. The approach applies structured pruning to create a smaller draft model that enables speculative decoding, targeting on-device agent workloads. This work addresses inference efficiency for mid-size open-weight models on consumer-grade x86 silicon.
ASRD: Training-free anchor-guided revocable decoding for diffusion LLMs improves accuracy and throughput
A new arXiv preprint introduces ASRD (Anchor Supervised Revocable Decoding), a training-free framework for improving decoding quality in diffusion large language models. The method addresses error propagation and local error reinforcement in revocable decoding by separating trusted 'anchor tokens' (identified via temporal consistency) from uncertain candidates, then applying anchor-guided generation and anchor-perturbed verification. Experiments on math and coding benchmarks show up to 6.4% accuracy improvement and 7.2× inference throughput gains over remasking baselines.
GGRO: Gradient-Guided Reward Optimization for inference-time LLM alignment
Researchers introduce Gradient-Guided Reward Optimization (GGRO), an inference-time alignment method that uses gradient signals from a reward model to inject 'nudging tokens' at high-uncertainty decoding steps, rather than relying on sampling-intensive re-ranking approaches like Best-of-N. The method monitors token-level entropy to detect distribution drift and steers generation trajectories directly, claiming improved robustness to reward hacking with minimal computational overhead. Experiments show gains across safety, helpfulness, and reasoning benchmarks compared to standard inference-time alignment baselines.
Universal Assisted Generation: Faster Decoding with Any Assistant Model
Hugging Face introduces Universal Assisted Generation (UAG), a technique that extends speculative decoding to work with any assistant model regardless of tokenizer or vocabulary differences. The approach enables using smaller, mismatched draft models to accelerate inference of larger target models, removing the previous constraint that both models share the same tokenizer. This broadens the practical applicability of speculative decoding across the open-weights ecosystem.
SimSD: Speculative Decoding Adapted for Diffusion Language Models
SimSD introduces a training-free speculative decoding algorithm for diffusion large language models (dLLMs), which previously could not use standard token-level speculative decoding due to their bidirectional attention and masked language modeling formulation. The method uses a plug-and-play masking strategy that introduces reference tokens from a draft model and a custom attention mask, enabling valid logit computation for drafted tokens in a single forward pass. Evaluated on SDAR-family dLLMs across four benchmarks, SimSD achieves up to 7.46x decoding throughput improvement while maintaining or improving generation quality. The approach is compatible with other acceleration techniques such as KV cache and blockwise decoding.