FMplex: New System for Efficiently Serving Multiple Foundation Models Through Shared Backbones
Researchers have developed FMplex, a serving system that allows multiple customized tasks to share the same foundation model backbone while maintaining task-specific extensions and isolation. The system uses virtual foundation models (vFMs) backed by shared physical models, combined with a fair-queueing scheduler for efficient batching. In testing across 92 downstream tasks and 7 foundation models, FMplex reduced latency by up to 80% compared to spatial partitioning and enabled hosting 6x more tasks at cluster scale.
FMplex addresses a key inefficiency in current foundation model deployment: each customized task typically runs as an independent model instance, duplicating expensive backbone models and wasting accelerator memory. The system introduces a virtualization layer where multiple tasks share a single physical foundation model while each receives a logically private virtual foundation model (vFM), preserving task-specific customizations, independent lifecycles, and isolation guarantees. The approach includes a batch-aware fair-queueing scheduler that optimizes both inter-task and intra-task batching to maximize resource utilization. Evaluation across 7 foundation model backbones (16 variants) and 92 downstream tasks demonstrates significant improvements: latency reductions of up to 80% over spatial partitioning approaches and 33.3% over best-effort co-location, while supporting up to 6x more concurrent tasks at cluster scale.
What's missing
The paper does not discuss potential limitations such as memory overhead of the virtualization layer, performance variability under extreme load imbalance, or applicability to models with dynamic architectures. The evaluation scope and specific foundation models tested are not detailed in the abstract.
What different sources said
- arXiv cs.AICenter
FMplex: Model Virtualization for Serving Extensible Foundation Models
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.