Spiffy: Speculative Decoding Algorithm Accelerates Diffusion Language Models
Researchers have introduced Spiffy, a speculative decoding algorithm designed to accelerate inference in diffusion large language models (dLLMs) while mathematically guaranteeing the original model's output distribution is preserved. Unlike autoregressive LLMs, dLLMs generate text in a bidirectional, block-wise fashion, requiring a novel approach using directed draft graphs calibrated offline and pruned dynamically during inference. The work addresses a key bottleneck limiting the practical deployment of dLLMs, demonstrating up to 8.6× fewer model inference calls and 6.3× faster token generation rates.
Diffusion large language models have attracted attention as a potentially faster alternative to standard autoregressive models, but realizing their speed advantages in practice has remained challenging. Spiffy, presented by Sudhanshu Agrawal and colleagues, adapts speculative decoding — a technique previously used to speed up autoregressive LLMs — to the distinct generative structure of dLLMs. The core innovation is a directed draft graph that captures the bidirectional and block-wise dependencies inherent to diffusion-based generation, replacing the independent draft models typically required in autoregressive speculative decoding. Draft graphs are calibrated offline to maximize token acceptance rates and are dynamically pruned at inference time to reduce unnecessary computation. Combined with KV caching and threshold-based dynamic unmasking, Spiffy was tested on LLaDA, Dream, and SDAR models, achieving up to 8.6× reduction in model inference calls and 6.3× acceleration in token throughput. The paper provides a formal proof of losslessness — meaning the output distribution of the original model is provably unchanged — with conditions for this guarantee specified in the paper. The work has been accepted to an ICML 2026 workshop.
What's missing
The paper does not report wall-clock latency benchmarks on standardized hardware configurations that would allow direct comparison with leading autoregressive LLMs, making it difficult to assess whether dLLMs with Spiffy are competitive end-to-end. Downstream task quality evaluations (e.g., benchmark accuracy) are not discussed, leaving open whether the acceleration trades off against output quality in practice.
What different sources said
- arXiv cs.AICenter
Structuring The Future: Diffusion LLM Speculative Decoding via Calibrated Draft Graphs
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.