dLLM-Cache: New Caching Framework Accelerates Diffusion-Based Language Models
Researchers have introduced bicache, a novel key-value (KV) caching technique designed to enable shared prefix caching in diffusion language models (DLMs), improving serving throughput by 36.3%–98.3%. Unlike traditional autoregressive large language models, DLMs use bidirectional attention, which causes existing caching methods to corrupt shared prefix data and collapse model accuracy to near zero. The technique addresses a critical bottleneck for deploying DLMs at scale in high-throughput production environments.
A preprint posted to arXiv introduces bicache, described as the first KV caching method for shared prefixes specifically designed for diffusion language models (DLMs). The core challenge is that DLMs rely on bidirectional attention, meaning any token update dynamically alters the entire context and its associated key-value pairs—an assumption that breaks existing caching techniques built for autoregressive LLMs. The authors found that applying standard prefix caching to DLMs causes model accuracy to collapse to near zero. Their analysis revealed that shared prefix KVs remain stable and reusable in shallower transformer layers, and that the safe depth of those layers depends on the proportion of shared prefix tokens in a given request. Bicache exploits this property by dynamically identifying a safe layer depth for reusing cached KVs while eliminating redundant computation in deeper layers. Evaluations show throughput improvements of 36.3%–98.3% over existing techniques, with accuracy degradation of only 0–1.8%.
What's missing
The interaction between bicache and other inference optimization techniques (e.g., quantization, speculative decoding) is not addressed.
What different sources said
- arXiv cs.CLCenter
Diffusion Language Model Parallel Decoding via Product-of-Experts Bridge
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.